Professional Documents
Culture Documents
Periodic Orbits - Moulton
Periodic Orbits - Moulton
1920
^o
PERIODIC ORBITS
>
BY
F. R.
MOULTON
IN COLLABORATION
WITH
DANIEL BUCHANAN, THOMAS BUCK, FRANK L. GRIFFIN, WILLIAM R. LONGLEY and WILLIAM D. MacMILLAN
>?
f,
Washington, 1920
OB
WASHINGTON,
D. C.
Hi
INTRODUCTION.
The problem
of three bodies received a great
impetus
in 1878,
when
His inves-
tigations were carried out with practical objects in mind, and comparatively little attention was given to the underlying logic of the processes which he invented. For example, the legitimacy of the use of infinite
determinants was assumed, the validity of the solution of infinite systems of non-linear equations was not questioned, and the conditions for the convergence of the infinite series which he used were stated to be quite unknown. These deficiencies in the logic of his work do not detract from the brilliancy and value of his ideas, and his skill in carrying them out excites only the highest admiration. The work of Hill was followed in the early nineties by the epoch-making researches of Poincare, which were published in detail in his Les Methodes Nouvelles de la Mecanique Celeste. Poincare brought to bear on the problem all the resources of modern analysis. The new methods of treating the difficult problem of three bodies which he invented were so numerous and powerful as to be positively bewildering. They opened so many new fields that a generation will be required for their complete exploration. On the one hand, the results were in the direction of purely theoretical considerations, in which Birkoff has recently made noteworthy extensions; on the other hand, they foreshadowed somewhat dimly methods which will doubtless
celestial
mechanics.
researches of Poincare are scarcely less revolutionary in character than were those of Newton when he discovered the law of gravitation and laid the foundations of celestial mechanics.
of three bodies in Acta Maihematica.
The
In 1896 Sir George Darwin published an extensive paper on the problem In mathematical spirit it was similar
essentially those of Hill, but the problem treated was considerably different. For a ratio of the finite masses of ten to one, Darwin undertook to discover by numerical processes
to the
work
the periodic orbits of certain types and to follow their changes with varying values of the Jacobian constant of integration. This program was It gave excellently carried out at the cost of a great amount of labor.
all
numerical results for many orbits in a particular example. The investigations contained in this volume were begun in 1900 and, with the exception of the last chapter, they were completed by 1912. Those not made by myself were carried out by students who made their doctorates
specific
under
my
direction.
in
iv
Introduction.
The
following chapters have been heretofore published in substance: American Journal of Mathematics, vol. xxxm (1911). I. Sections III and IV
II.
III.
IV.
VII.
VIII.
Astronomical Journal, vol. xxv (1907). Hcudiconti Matematico di Palermo, vol. xxxn (1911). Transactions of the American Mathematical Society, vol. xi (1910). Mathcinatisilie Annalen, vol. lxxiii (1913). Annals of Mathematics, 2d Series, vol. 12 (1910)
Transactions of the American Mathematical Society, vol. vu (1906). Transactions of the American Mathematical Society, vol. xin (1912). Transactions of the American Mathematical Society, vol. vin (1907). Transactions of the American Mathematical Society, vol. ix (1908). Proceedings of the London Mathematical Society, Series II, vol. 2 (1912).
XI.
XII. XIII.
XIV. XV.
The investigations and computations contained in the last chapter were completed in 1917. It was originally intended to publish only the first fifteen chapters, and if that program had been carried out they would have appeared in 1912. But as the work of printing progressed the ideas contained in the last chapter were being developed and the computations were begun. It was thought that an even more nearly complete and certain idea of the evolution of periodic orbits with changing parameters could be obtained in a year than were
obtained in five years. The difficulties and enormous amount of labor involved were not foreseen. No one can now read with better appreciation
than
I the following
far as I
his paper:
can see, the search resolves itself into the discussion of particular cases by numerical processes, and such a search necessarily involves a prodigious amount of work. It is not for me to say whether the enormous amount of labor I have undertaken was justifiable in the first instance; but I may remark that I have been led on by the interest of my results, step by step, to investigate more, and
again more, cases."
"As
The results which now appear had all been obtained when service in army made it necessary to lay them aside before the final chapter could be put into form for publication. After they had been laid aside for about two years it was not easy to gather up the details again and to arrange them
the
a systematic order. This explains the long delay in the appearance of volume. It is clear that it was in no wise due to the Carnegie Institution of Washington. Indeed, the patience of President Woodward with and long expensive delays has been far beyond what could reasonably have been expected. In the greater part of this work complete mathematical rigor has been
in
this
insisted upon.
the other hand, the developments have been in a form For example, applicable to practical problems in celestial mechanics. sections III and IV of Chapter I treat non-homogeneous equations of the
On
types which arise in practical problems; Chapter II is devoted to questions which have long been classic in celestial mechanics; Chapter III contains, among other things, a new and rigorous treatment of Hill's differential
IV
treats a
problem that
arises,
Introduction.
at least approximately, in the solar system; Chapter is developed in a form suitable for numerical applications; Chapter VI is an alternative treatment
of the
in
involving entirely
new types
of difficulties
methods of treatment are required; Chapter XI covers the same ground as Hill's work on the moon's variational orbit and Brown's work on the parallactic terms;
Chapter XII contains the corresponding discussion for superior XIV treats a problem similar to that presented by the planets; satellite systems of Jupiter and Saturn or by the planetary system. Chapter
and Chapter
XV contains a
and
it
discussion of limiting cases of periodic orbits, namely, closed It forms a basis for part of the work of the last chapter, orbits of ejection.
be found to have practical applications in the escape of atmostrace out the evolution of periodic pheres. In spite of the orbits as the parameters on which they depend are varied. fact that infinitely many families of periodic orbits were found, whereas only a few such families were previously known, the discussion remains in cerIt should be stated that many results were found tain respects incomplete. which have not been included because they did not contribute to the solution of the particular question under consideration. For example, a series
may
of orbits asymptotic to the collinear equilibrium points was computed. The amount of labor the last chapter cost can scarcely be overestimated. I should not be true to own feelings if I did not express the appre-
my my
collaborators.
The
association with
them
has been a deep inspiration. They are to be held accountable only for those chapters which appear under their names. Most of the computations on which many of the results of the last chapter are based were made by Dr. W. L. Hart and Dr. I. A. Barnett. Without assistance of such a high order the laborious computations could not have been carried out. F. R. Moulton.
source of satisfaction
and
CONTENTS.
Chapter
I. I.
Equations.
Certain Theorems on Implicit Functions and Differential (By F. R. Moulton and W. D. McMillan.)
page
Jolution of Implicit Functions. 1. Formal solution of simultaneous equations is distinct from zero at the origin 2. Proof of convergence of the solutions
3. 4.
5.
Generalization to
many
parameters
all its first
The
functional determinant zero but not = a* t (a, fi) p x Case where (a, a)
P2
(a,
/i)
2 5 5 6
6
6. 7.
G
10 10
1
The types
Formal
10. 11.
Type 1 Determination of the constants of integration in Type Proof of the convergence of the solutions of Type 1
11
11
III.
parameters 13. Generalization of the parameter 14. Formal solution of differential equations of Type II 15. Determination of the constants of integration in Type II 16. Proof of the convergence of the solutions of Type II 17. Case of homogeneous linear equations Homogeneous Linear Differential Equations with Periodic Coefficients. 18. The determinant of a fundamental set of solutions 19. The character of the solutions of a set of linear homogeneous differential equations with uniform periodic coefficients 20. Solutions when the roots of the fundamental equation are all distinct .... 21. Solutions when the fundamental equation has multiple roots
22.
12. Generalization to
many
14 14 15 16 16 19 21
22 24 26 30 32
The
characteristic equation
when
,
power
when a \ a^,
l
...
the coefficients of the differential equaa parameter n 0) are distinct and their differences are a,^
series in
when no two
mod.
0)
to zero
25. Solutions
when, for m = 0, the characteristic equation has a multiple root 26. Construction of the solutions when, for n = 0, the roots of the characteristic equation are distinct and their differences are not congruent to
.
a^
when
0)
'
a^
o[
is
congruent
.
34 35
zero
mod.
36
27. Construction of
the difference of two roots of the characteristic equation is congruent to zero mod. y/ 1 28. Construction of the solutions when two roots of the characteristic equation are equal
the solutions
when
40
43
IV. Non-Homogeneous Linear Differential Equations. 29. Case where the right members are periodic with the period 2ir and the a; are distinct 30. Case where the right members are periodic terms multiplied by an exponential, and the a; are distinct
31.
46 47 48 50
51
Case where two characteristic exponents are equal and the right members
are periodic
V. Equations of Variation and their Characteristic Exponents. 32. The equations of variations 33. Theorems on the characteristic exponents
VII
Mil
CONTENTS.
Chapter
34.
II.
Elliptic Motion.
page.
35.
of
motion
55 56 58 59 60 61 61 64 66
Chapter
I.
III.
The
differential equations
44. Transformation of the z-equation 45. First demonstration that the solution of (12) is periodic period are expansible as power series in n
46.
Second demonstration that the solution of equation (12) 47. Third proof that the solution of (12) is periodic
48. Direct construction of the solution 49. Construction of the solution
II.
periodic
71
73
73 76 78 78
81
III.
Case 1 54. Direct construction of the solutions in Case II 55. Direct construction of the solutions in Case III Solution of the X- and Y -Equations for the Spherical Pendulum.
53. Direct construction of the solutions in
83 85 87 88 90 93
56. Application to the spherical pendulum 57. Application to the simple pendulum 58. Application of the integral relations
Chapter
59. Introduction
60.
The
I.
66. Construction of periodic solutions in the equatorial plane 67. Construction of periodic solutions for orbits inclined to the equatorial plane. 68. Construction of periodic solutions in a meridian plane
II. Orbits
Re-eniraid after
many
Revolutions.
69. 70.
71. Special theorems for the non-homogeneous equations 72. Integration of the differential equations 73. Existence of periodic orbits having the period 2kw 74. Construction of the solutions with the period
2kt
CONTENTS.
Chapter
75.
76.
IX
V. Oscillating Satellites
librium Points.
Statement of problem
(First Method.)
The
differential equations of
motion
77.
P2 P
,
81. Outline of steps for proving the existence of periodic solutions of equations 82.
=e=
*
when
=e=
for the differential equations 85. Existence of periodic orbits of Class 86. Some properties of solutions of Class
Normal form
A A
87. Direct construction of the solutions for Class 88. Additional properties of orbits of Class 89. Application of Jacobi's integral to the orbits of Class 90. Numerical examples of orbits of Class
A
B
93. Direct construction of the solutions for Class 94. Additional properties of the orbits of Class
95.
96.
On
C
(Second Method.)
163 166 168 173 176 179 180 180 183 189 190 191
Chapter
97. Outline of
method
differential equations 99. Integration of the differential equations 100. Existence of periodic solutions
98.
The
101. Direct construction of the solutions for Class 102. Direct construction of the solutions for Class
A B
Chapter
103.
104.
105. 106.
The The
differential equations of
elliptical solution
107. Integration of equations (13) 108. The terms of the second degree 109.
The terms
110. General properties of the solutions 111. Conditions for the existence of symmetrical periodic orbits 112. The existence of three-dimensional symmetrical periodic orbits 113.
114. 115.
116.
Case Case
N odd
Convergence
The
CONTENTS.
Chapter VII
Continued.
page.
The The
general step for the x and j/-equations general step for the z-equation
.
Determination of the Coefficients of Z2/+1 when e = 0. General equations for Z, 130. Case when e ?* 0.
131. Coefficients of e 132. Coefficients of e 133. Coefficients of e
2
Application of the Integral. 127. Form of the integral 128. The integral in case of the periodic solutions 129. Determination of the coefficients of Z2j+\ when e =
and
e*
136. Coefficients of e
members
of the first
two
278 280 280 282
of the periodic solution of the coefficients of Xi 2 142. Coefficients of X 143. Induction to the general
Form
Chapter
144.
I.
VIII.
Statement of problem
285
286 288 289 290 294 296 296 297
Determination of the Positions when the Masses are given. 145. The equations defining the solutions 146. Outline of the method of solution 147. Proof of theorem (B) 148. Proof of theorem (C)
149. Computation of the solutions of equations (6) Determination of the Masses when the Positions are given. 150. Determination of the masses when n is even 151. Determination of the masses when n is odd 152. Discussion of case n = 3
II.
Chapter IX. Oscillating Satellites near the Lagrangian Equilateral Triangle Points. (By Thomas Buck.)
153. Introduction
154. 155.
156.
CONTENTS.
Chapter IX
158.
159. 160. 161. 162.
XI
PAGE
307 309 309 311 312 313 319
Continued.
The first generating solution The third generating solution The fourth generating solution The fifth generating solution The seventh generating solution
163. Construction of the solutions in the plane 164. Construction of the solution with period 2ir
Chapter X. Isosceles-Triangle Solutions of the Problem of Three Bodies. (By Daniel Buchanan.)
165. Introduction
I.
32
the Finite Bodies
Periodic Orbits
when
the
Third Body
is
Infinitesimal.
166.
The
differential equation of
motion
167. Proof of existence of a periodic solution of equation (1) 168. Direct construction of the periodic solution of equation (1)
II.
the Finite
Body
169.
is Infinitesimal.
The
differential equation
330
by a necessary condition
171. Existence of symmetrical periodic orbits 172. Direct construction of symmetrical periodic solutions of (23)
III. Periodic Orbits
when
the Three
173.
The
differential equations
340
341
174. Proof of existence of periodic solutions of equations (70) 175. Proof that all the periodic orbits are symmetrical
176. Direct construction of the periodic solutions of (70) 177. The periodic solution of equation (69)
178.
The
180.
The
187. Induction to the general step of the integration 188. Application of Jacobi's integral
189.
The
190. Applications to the lunar theory 191. Applications to Darwin's periodic orbits
XII
CONTENTS.
Chapter XII. Periodic Orbits of Superior Planets.
PAGE.
192. Introduction 193.
The
differential
equation
194. Proof of the existence of {>criodic solutions 195. Practical construction of the periodic solutions 196. Application of the integral
Chapter XIII.
A Class of Periodic Orbits of a Particle Subject to the Attraction of n Spheres having Prescribed Motion. (By W. R. Longley.)
389 390 396 402 404 410 414 415 417 419 420 422
197. Introduction 198. Existence of periodic orbits having no line of symmetry 199. Construction of periodic orbits having no line of symmetry
200. Numerical
Example
201
Some
n bodies
Numerical Example 3
206.
207.
208.
The undisturbed orbit must be circular More general types of motion for the finite bodies Numerical Example 4
Chapter XIV. Certain Periodic Orbits of k Finite Bodies Revolving about a Relatively Large Central Mass. (By F. L. Griffin.)
209.
210.
21 1. 212. Conditions for periodic solutions 213. Nature of the periodicity conditions 214. Integration of the differential equations as power series in parameters. 215. Existence of periodic orbits of Type 1 216. Method of construction of solutions. Type 1
217. Concerning orbits of 218. Concerning orbits of
Type Type
II
III
.'
219. Concerning lacunary spaces 220. Jupiter's satellites I, II, and III 221. Orbits about an oblate central body
425 427 427 428 429 430 434 438 442 447 450 452 455
two-body problem
225. Orbits of ejection in rotating axes 226. Ejectional orbits in the problem of three bodies 227. Construction of the solutions of ejection
228. Recursion formulas for solutions 229. The conditions for existence of closed orbits of ejection 230. Proof of the existence of closed orbits of ejection 231. Conditions at an arbitrary point for an orbit of ejection 232. Closed orbits of ejection for large values of n 233. Periodic orbits related to closed orbits of ejection 234. Periodic orbits having many near approaches
457 457 461 461 462 463 465 469 471 473 477 479 483
CONTENTS.
Chapter XVI.
XIII
238.
The The
bodies
239.
Cusps on periodic
orbits
240. Periodic orbits having loops which are related to cusps 241. The persistence of cusps with changing mass-ratio of the finite bodies.
242.
243.
Some
The The
triangular points
244.
analytic continuity of the orbits about the equilateral triangular points existence of periodic orbits about the equilateral triangular points for
large values of n 245. Numerical periodic orbits about equilateral triangular points 246. Closed orbits of ejection for large values of n
247. Orbits of ejection from 1 ft and collision with n 248. Proof of the existence of an infinite number of closed orbits of ejection of orbits of ejection and collision when ju J^ 249. On the evolution of periodic orbits about equilibrium points
and
517 519 520 522 524
250.
251. 252.
On On On
the evolution of direct periodic satellite orbits the evolution of retrograde periodic satellite orbits the evolution of periodic orbits of superior planets
U
-
PERIODIC ORBITS
BY
F. R.
MOULTON
IN COLLABOHATION
WITH
DANIEL BUCHANAN, THOMAS BUCK, FRANK L. GRIFFIN, WILLIAM R. LONGLEY and WILLIAM D. MacMILLAN
Chapter
I.
Moulton and W.
D.
MacMillan.
I.
1. Formal Solution of Simultaneous Equations when the Functional Determinant is Distinct from Zero at the Origin. In applying the condi-
tions for periodicity of the solutions of differential equations after the method of Poincare,* there will be frequent occasion to consider the solution of
P(a,,
for
a,
,
. .
,a; M )=0
t
(t-1,
n ),
(1)
a in terms of n, where the P are power series in the a, and n, = 0, n = 0, but not with a, = 0, n^O, and with a, vanishing for converging a j\< r j>Q, M < p>0. We are interested in only those solutions which vanish with p; that is, if we regard a, a, p as coordinates in (n+l)-space, in those curves satisfying (1) which pass through the origin. Equations (1) can be satisfied formally by the series
. ,
\ | I
= 2 aV,
(2)
where the
mined.
J)
a\
Upon
series in n, it is
which are to be detersubstituting (2) in (1), expanding and arranging as power found that
t
+[2f^"+ F
where the
tt,
"
-
"""">]*'
_1)
<i
-1
"
are polynomials in a"', a/ moment the for the that series (2) converge and satisfy Upon assuming Hence the coefficients of each (1), it follows that (3) are identities in p. are zero. of ix separately power
.
Ff
I,
chap.
3.
PERIODIC ORBITS.
From
first
power
of
m we get
a<"
<
dP,
tf-1,
,n).
(4)
Since
da
A=
dP
dcii
3a,
equations (4) can be solved uniquely for the a"'. If not all of the dP /dn are zero, then not all of the a"' " are zero; but if all of the dP /dn are zero, then all of the a are zero.
is
distinct
from zero
for a!
=
t
=a =
0,
Equating the
coefficients of the
second power of
/z
in (3) to zero,
we
get
2|?- -*?>(?,
**
tfo
0-i,
,n),
the right members of which are completely known. The determinant of the coefficients of the af is A, and the af are therefore uniquely determined, being all zero or not all zero according as the Ff are all zero or not all zero.
seen from (3) that the treatment of the general term is entirely similar and depends upon the same determinant A. Hence, under the hypothesis
it is
And
as to A, a formal solution
is
possible,
and
it is
unique.
2.
Q. (ft
for
/3j
.
.
,/3 B
M )=0
are
|
(*-l
n)
(!')
in terms of n,
where the
power
\
series in the #,
|
r vanishing for #, = 0, m = 0, and convergent for fij 0, where also the coefficients of all terms beyond the linear in
< >
n each
|
and n, and
are real,
positive, independent of i, and greater than the coefficients* in the expansions of any of the
The character of
the coeffi-
cients of the linear terms will be specified when they are used. Suppose the solutions of (10 have the form
(i-1,
...,>.
(20
*For proof of the possibility of satisfying these conditions see Picard's Traite d'Analyse, edition of 1905, vol, II, pp. 255-260.
IMPLICIT FUNCTIONS.
On
substituting
(2') in (1')
system
results a
W
ft
15
ft
(-l,
n).
(4')
It is necessary now to specify the properties of the linear terms of It will be supposed first that the dQ /dn are real and positive, and the Qt that dQ /dfi= =dQ n /dn = dQ/d/j,. It follows that for fixed values of
.
the dQi/dfis the values of the /3 satisfying The dQi/dj3j will now be so determined that
t l
"
(4')
greatest \dP /dn\ the ftj' determined by (4') shall be equal, positive, and at least as great as the greatest |a "| for all values of dPi/d/x such that dP,/dn ^ dQ/dn. This must be done in such a way that the determinant
!
aQi
aft'
A'
=
dQn dQn
aft
shall
n = 0.
satisfied in
j?i and many ways. A simple way is to choose dQ,/d^= 1 if n~ = = A' c is of determinant Then the (c+n), (-1)" (4') dQ /dp -(1+c). which can vanish only if c = or c = - n, and the solutions of (4') are
infinitely
t
t
dQ
(i)
fl"
_ =
(i)
dfx
(5)
(c
+ n)
"
.
|
by equations
of the
form
2^>
It follows
<
i/3
2)=
GW>
{%>)
(t-i,
...,).
(6)
of
from the properties of the Q together with the the G'f and the values of the 0?', that
t
,
explicit structure
G< 2)
G? ^ F?
|
(i
= l,
),
**
a<
I
()
;
(<-l,
n).
4
It is
PERIODIC ORBITS.
very easily shown by induction that for every value of the index k
GPtfTherefore
if (2')
= G
-tf
|
^ FT
|
0-1, ...,),
fc|,"|
then also
(2)
0-1
n).
converge for
^ p'>
converge for
imposed upon
by
functions
Q =-c 8
l
j
-(l+Jlf)((8
+.) +
M(^t
1
'
'
+ ^" +
',
'
0-5)
where
for 0i
ft+
+/3n
^,
/
. .
=0
(8)
,n),
r
(t'=l,
M
+
is
PM
0,+
+j8
p-p-yj\ ^n
p-/x/
p-m
(9)
0+5+*)
Since each
/3,
,
and therefore the sum &+ +/3 must be taken before the radical in
|p|
is
zero for p = 0,
(9).
The
p.
right
|
member
of (9)
series in
if
is
|
<p and
|/Gu)|
< |<p(p)|,
where
conditions which can always be satisfied, whatever may be the values of n, r, p, c, and Moreover, the coefficients of all powers of p. in the expansion
and
positive.
Hence
it
follows that
&+
where
R(p.) is
+(3n
pR(p),
all
(10)
positive.
a power
series in
It
all
the
ft
are equal.
Hence
n
For
|p|
(11)
power
this result
and
|
of these equations are converging moreover, they identically satisfy (8). It follows from the second set of (7) that p" exists such that the series (2)
members
>
converge for
< p".
*Picard's Trailt d' Analyse, loc.
cit.
IMPLICIT FUNCTIONS.
Many
.
Parameters.
(a,
a.
/*,
Mt )
=0
(i
= l,
n ),
(12)
and that the functional determinant with respect to the a., is distinct from = a = ii = = Mt = 0. Then the problem can be reduced zero for aj= to that discussed in 1 and 2 by letting fij = Cjix. After the solutions have been obtained the c} n everywhere can be replaced by n for the Cj m will occur
1
its
First
Minors
(a,
a;
p)
{i
= \,
),
(13)
where the P, have the properties imposed upon the P, of 1. Suppose that = a = n = 0, the determinant of the linear terms in the a, is zero for a! = but that not all of its first minors are zero. It may be supposed without
determinant of the terms remaining after the last row and column of the linear terms is distinct from zero. deleting 1 Hence, as a consequence of the theorems proved in 1, 2, 3, the first n
any
equations can be solved uniquely for a 1; ., a_i as power series in a a and = = 0. H, vanishing for a the solutions of the first n 1 equations for ot ., Suppose a_i are substituted in the last equation. It will become a function of a n and n which
.
fj.
may
P(a, M )
= S 2cw aV =
i=0 )=Q
(k
+ j>0).
(14)
Since the determinant of the linear terms of (13) is zero, this equation carries no linear term in a. Suppose the term of lowest degree in a alone is cto a\ Then, for each value of n whose modulus is sufficiently small there are k
values of a satisfying (14) and, moreover, the modulus of /x can be taken so small that the moduli of the solutions for a shall be as small as one pleases, f
Also, for each set of values of a there is one set of values a!
.
= a and
. .
a B satisfying (13). special discussion and the method of of these solutions the character to determine required finding them. These questions are taken up in the immediately following
.
whose moduli are sufficiently small a_, satisfying the first n 1 equations of whose modulus is sufficiently small there are
/x
,
.
.
is
articles.
case, in which all the first minors of the functional determinant vanish, does not arise has only recently (in 1911) been completely solved by MacMillan, in a paper which will appear in Maihemalische Annalen. fWeierstrass, Abhandlungen aus der Functionenlehre, p. 107. Picard, Traite d' Analyse, vol. II, chap. 9, Harkness and Morley, Treatise on the Theory of Functions, chap. 4. {7, and chap. 13.
in this work.
6
5.
PERIODIC ORBITS.
of (14)
p)
=a*P,(o,
m)~m
Pj(a, p).
p)
a*P,(a,p)
- ^PA*,n) =0,
| |
where Pi and P2 are power series in a and p which do not vanish for a = p = and which converge for a < r > 0, p < p > 0. Upon extracting the k' h
| |
a Qi (a, p)
V M* Qa (a, p)
0,
where Qi and Q2 are power series in a and p which do not vanish for a = p = and which converge for a < r' > 0, p < p' > 0, and where r\ is a A;" root of = v k this equation takes the form of those treated in 1 unity. If we let p and 2 and can be solved uniquely for a in terms of v for each 77. The k
1
|
by taking
for
17
Suppose
C,
P (a,
p) has the
form
(15)
in a
P (a, p) = 2
a*-' p'
+ Q (a, p)
0,
where c ^ and C> contains no term of degree less than k + 1 It can be supposed without loss of generality that c = 1. Then
1
and
p.
2c,a*
(-0
y=
(a,n)
(a
- 6, p)
(a
-6
p)
(a
b k n).
(16)
Suppose (a bjfi) is a simple factor of the homogeneous polynomial (a, p), and exclude the trivial case in which it is also a factor of Q (a, p).
Then dP /da
for a
= bj p.
(17)
divisible
by
p*.
After p*
is
divided out, P carries a term in P to the first degree whose coefficient is not zero, and Q carries no term independent of p, but has at least one term
in p alone, for
otherwise
for /3 as a con1 quently by = series in for 0. Therefore a can be expanded p p, vanishing verging power in with of the simple as a converging power series p, vanishing u, for each b k are all distinct, the roots of the polynomial P (a, p) = 0. If fe,
. . .
b } y).
Conse-
branches of the function P(a, p) which pass through the The actual determination of the coeffiorigin can be found by this process.
expansions for the k
cients
is
by the method
of
n=
7. General Case of Power Series in two Variables.* The method of treatment consists in reducing the equation, by suitable transformations, to forms of a standard type from which the solutions can be found. In cer-
The
analysis in
*This problem has been treated by Puiscux, Nother, etc. For references and discussion see Harkness and Morley's T realize on the Theory of Functions, chap. 4, and Crystal's Algebra, part 2, chap. 30.
IMPLICIT FUNCTIONS.
general, as well as the particular transformations required in to reduce the equation to the standard forms, is indicated
Newton's Parallelogram. In constructing Newton's parallelogram it is sufficient to consider only those terms c y aV of P(a, n) for which i <! &, j ^ X, cw a* being the term of lowest degree in a alone and cox m x the term of lowest degree in n alone. Take a set of rectangular axes and for each of the terms cw an c 9^ 0, lay down a degree point whose coordinates are i and j. Then the line passing through the origin and the point (k,0) is rotated around (k,0) as a pole so that it moves along the ./-axis in the positive direction until it strikes at least one
J
,
tj
other degree point (it may, of course, strike several simultaneously). Let the one of those which it first strikes having the greatest j be (ii,ji). Then the
line is rotated
around
(t'i,
in the
same
direction until
it
strikes at least
one
other degree point. Letting the one of these having the greatest j be (i2 ,j2 ), the line is rotated around (i2 j2 ) until another is encountered. This process is
,
continued until the point (0, X) is reached. evidently can not be greater than k or X.
The number of steps in the process The part of Newton's parallelocharacter of the function near the origin is made
to
(i2 ,
j2 ),
(w,)
to
(0, X).
For the
terms of (a, n) corresponding to each one of these segments there is, as will be shown, a transformation which throws P(a, n) into a standard form. In order to illustrate Newton's parallelogram, consider the example
P(a, n)
= CM a
-\-C 3i
3
fi
C22
2
fj.
+C
l3
an
+ C u a + Qx //+Q (a,
fi'
n),
where Q(a, m) contains only terms of the seventh and higher degrees in a and m- The coordinates of the points in Newton's Parallelogram are (5, 0), (3,1), (2,2), (1,3), and (0,6), and it (5,0) consists of three segments which are
shown
(i 2
in Fig. 1.
j3
M%
_ J2-J1 ti
x2
m
n
(18)
8
It follows
PERIODIC ORBITS.
(%,j,) to (i,jt)
with reference
(a, y) to every other degree point, that in the case of any term cti a y' of i and the j satisfy the whose degree point is not on this segment exponents
inequality
i
U2-J1)
+j (h - H) + itjt - hh >
0.
J Consequently, after the transformation (18) the term c,jay becomes c^py'",
where *"
.>
<T
This discussion proves that after the transformation (18) the terms belonging to the segment (ii,ji) to (i2 ,j2 ) contain y"' as a factor, and that every other term contains y to a higher power than a'. Since a' is not necessarily an integer the series will not be, in general, a series in integral powers of y, but
it
will
be
in integral
powers of
(}"
y'
y"'
the series
becomes
=
For
c,
c uh
c hh 0*
y'
P,
(0, m')-
(20)
n'
[>-''
j8''
+
=
is
+}]-
*?
(0-cd-
(/3-c _
fl
(!
0.
(21)
The
it
solution
belongs to the solutions obtained from the segments having smaller values of i. Suppose (8 c,, is a simple factor of (21) and let
=
Then the
from
back
zero.
c,
t..
(22)
right
y
ishing with
member of (20) becomes a power series in 7, and fi', van= = 0, and the coefficient of y to the first power is distinct y n'
y
Therefore,
by
is
1,
7,,
yJ
vanishing for
= 0. ju'
,
.
uniquely as a
and
(18), a
expressed in integral
.
are all distinct we integral series in n only if a is an integer. If c, c,,_ obtain at this step i, i2 solutions, and if a is an integer the number of them
,
(.
is
Since when a is not an integer y has more than one determination, and since the series obtained by the transformation (18) after removing the factor y,"' is not in integral powers of y, it would seem
precisely
i\
i2
that the
it
<r
will
is
number of solutions for the segment is now be shown that the number of distinct
not.
ji)/(ii
t,
ii
i2
i2 ,
But
?',
whether
prime
t
an integer or
Now
integers.
<t=
It
is
(J2
h)=m/n, where
-i2 equals
x
and n are
than
n,
relatively
clear that
n, or is greater
according
as^- j
do not have, or have, a common integral divisor greater than unity. Consider first the case where i i 2 = n. There can be no degree point
and i,
i2
IMPLICIT FUNCTIONS.
(i',f)
9
coordinates would
is
(t'i
and
i2 )
have to
i'
j')/{i'
= m/n, which
impossible
when
i2
<n.
0*+
x
+ A = 0,
x
whose solution gives i i2 values of /3, differing only by the i % 2 roots of unity. Hence the same final results are obtained by using the principal value of y." in (18) and all of these it it values of /3 as are obtained if any other
determination of //
is
used.
i2
Suppose now
than
q1
qn, where q is an integer. There can not be more on the segment (i j ) to (i2 j2 ), i. e. satisfying the degree points
ix
x ,
relation (j2 -j')/(i'-i2 ) =m/n. Therefore in this case (21) is a polynomial in 0" of degree q, and for each of the solutions for ft" there are n values of /3 obtained
one from the other by multiplying by the nth roots of unity. Hence in this case also the final results are the same as are obtained by using any other
of the
n determinations
of n".
all
It follows that in
every case
',
by
taking the principal value of y! and that the number of them for the In a similar manner the segment (ii,ji) to (i*,ji) is precisely ii i 2 solutions associated with each of the other segments can be obtained. The
.
in this
2)
way
is
N = (k-h) + & - i +
fj.)
(i
- 0) - *,
(23)
= for a which vanish with n = 0. which is the number of solutions of (P a, In this case the problem is completely solved. Suppose, however, that in treating the terms belonging to the segment =c = The analysis above fails to (i\,ji) to (12,^2) it is found that d
.
=
is
c,
+7
(24)
made, after which the right member of (20) is a power series in 7 and n'; and for /x' = 0, 7" = is a solution. That is, the equation is of the same form there are now as P(a, n) =0, only in place of having k zero roots for ju = = only p such roots. This number p is always less than k except when ii k, = = = = -*-*. But whatever the value of p j = 0, i2 0,j2 \, and d c2 a new Newton's parallelogram for the 7 /-equation is to be constructed.
l
depend upon terms of higher degree in the because the terms which gave rise to the p equal roots,
It will
original
c,
. .
.
concentrated, so to speak, by the transformations into the single one 7", and the parallelogram depends upon the term in // alone of lowest degree. By this step, or some succeeding one, the solutions will all become distinct
unless, indeed, the original P(a, m) are identical in n.
=0 has two
or
more
10
PERIODIC ORBITS.
II.
8.
types of differential equations will arise and they will be solved by processes adapted to attaining their solutions in convenient forms. It will tend to
and brevity of exposition of the actual dynamical problems to set down in advance those methods of solving differential equations which will be used, and to state, the conditions under which the results obtained bj them are valid. Consequently, this section will be devoted to these quesclearness
r
tions without
by being
independent and dependent variables and in certain parameters upon which they depend; and the solutions are considered only for those values of the variables and parameters for which the equations are all regular. In the case where the differential equations are linear, their coefficients are
either constants or periodic functions of the independent variable.
9.
Formal Solution
of Differential Equations of
Type
I.*
The
dif-
ferential equations
dx
-jt
will
- M/(*i
I
xn
m;
t)
0'=l
. ,
n)
(25)
when
x
x
the right
.
members have n
t,
as a factor
and when
the point
.,x,n and
Then
a ) and
t
= a,
m
as
for all
to^t^T.
t
the
/x
can be expanded
|x,-a,|
power
series in (x t
<r>0
and
t
(
|/x|
<p>0 for
^.t^T.
Suppose x
That
is,
suppose
(26)
XtiQ^a,,
in
which the letter under the identity sign = indicates the parameter in which the identity is defined. Equations (25) can be solved formally as power series in n which have the form x, = S x ? M '. (27)
{
where the
in
x\ are functions of n, it is
t.
On
and arranging
powers of
found that
(28)
+
*8ee Moulton's Introduction
to Celestial
convergent the coefficients of corresponding powers of /* members are equal. On assuming for the moment that are they convergent, the identity relations become
in the right
and
left
=
-^-
(*!,...,),
t
(29)
^=f
dt
(x?,0;t)
(j=i,...,n),
(30)
dx k
Xt
'
dn
31 )
^f =
These
d yJ 4+i y y
ay
(1)
>
ay
(1)
ay
rq2
sets of equations
From
(29)
we
get
(33)
x? = a
where the
a,'"
now known
functions of
are constants of integration. The right members of (30) are t, and their solutions can be written
*? = //.
where F*
1
(*?, 0;
t)
dt
+
0;
? - F? (0
t).
a ;>,
(34)
Then
(31), (32),
give in
order, similarly,
xf = F?\t)
a?,
x? = Ff(t)
a?,
(35)
may
be desired.
Determination of the Constants of Integration in Type I. At each of integration are obtained, and they must be determined in terms of the initial values of the x From (26), (27), (33), (34),
step n additive constants
(35)
it
{ .
,
. . .
follows that
af +[*?(fc)
Therefore
a 0)
(
afltt'ta,.
(i-1, ...
<*
a,
- - K(U)
oo).
(36)
By
in
these equations all of the constants of integration are uniquely determined terms of the constants of the differential equations and of the initial values
of the
dependent variables.
Proof of the Convergence of the Solutions of Type I. The of integrating differential equations as power series in parameters has been in use in more or less explicit form since almost the beginnings For example, in the year 1772 Euler published his of celestial mechanics.
11.
method
12
PERIODIC ORBITS.
second Lunar Theory, in which he used a process quite analogous to this;* and the method of computing the absolute perturbations of the elements of the planetary orbits is virtually that of developing the solutions as power series in the masses. But the actual determination of the conditions for the validity of the process was not made until Cauchy published his celebrated memoirs on differential equations in 1842. f The results of Cauchy were
extended by Poincare" in his prize memoir on the Problem of Three Bodies, t and were proved again, following Cauchy 's Calcul des Limites, in LesMethodes The theorem will be Nouvelles de la Mecanique Celeste, vol. I, pp. 58-63.
needed in this work in the form given by Poincare, viz. x m, ,x n ,n; t) of equations (25) are analytic \\inxi,. // thef (Xi, and t, and regular at x, = a n = 0, for all t ^.t^T, then p > can be taken so small that the series (27) will converge for all to^t^T provided m < PTo prove this theorem consider a comparison set of differential equations
:
t ,
I'^^.d/.r
where the
for all
to <p t
m;
-i,
n),
(250
,
and
/x
/x=0 it, t, and regular at y,= |a |=& 4 the coefficients of all of where, further, powers ^ 6, of all the are in the expansions <p, real, positive, and greater than the
are analytic in
ffo,
.
,yn
^ t^ T; and
moduli of the corresponding coefficients in the expansions of the /, for values of t under consideration. Then positive constants M, p,r exist such that equations (25') can be written in the form
y , .
all
. .
the
,
dy 1
Mjj.
"'"(>The
dyt_ "
dt
00-^)
by
(2/i-fri)
(i-^)'
+
+ (y- 6,) 1
'
(i_^ \ 1
r,
+
.
Mm_
(y,- h)
(25")
where
r is the smallest of
rn
Suppose now
series in
n of the form
fc2tfV.
Ju*0
(27')
There
will
be quadratures corresponding to
(29), (30),
....
Moreover,
by
on the
<p t
(i)
dt
> d^_
dt
Tisaerand's Micanique Celeste, vol. Ill, chapter 6. See Cauchy's Collected Works, 1st series, vol. VII. t Acta Mathcmatica, vol. XIII, pp. 5-266. ||The assumption that the /; are analytic in I is not necessary for the demonstration. {Picard's Traiti d' Analyse, edition of 1905, vol. II, pp. 255-260.
t
13
t
<3 $T.
tf
^ 4?
1
for
|
alU ^
T.
Then
dt
> dxf
dt
for
From this it follows similarly that yf > \xf\. This process to <t^ T. can be continued indefinitely, giving by induction for the general term
I^5|* /|
(
(37)
for
the right members of (27') are convergent for to^t^T, then likewise are the right members series when p <p> for the same range in t of (27) convergent when p < p >
t
<it^T.
Consequently,
|
if
It
is
(27')
integration of (25").
members
are the
same
,
for all
by a direct i, we have
= y -c =
2
= yn -cn
where
y
bt
cx
.,
By
.
it
follows that
t
c,
= Cj
bs
and y
b,= yt
of (25")
b}
Let this
common
value of
be y
b.
becomes
'
"
(i
Mjx
a) (i
_ n (y~ b
25 "0
On
integrating this equation and determining the constant of integration the condition that (y b)=0 at t = U, it is found that
by
0-$)
The
solution of this equation for (y
b) is
0-..
2n Afp
< -'-
,,
u>
J-^-)
<
38 >
Since (y
the negative sign must be taken before the radical. It follows directly from equation (38) that whatever finite values n, M,r,p, and T-t may have, (y b) can be expanded as a convergent series in p for to^t^T provided the condition
b)
at
= U>,
2nM\tx\
(T-kXl
0-5)
is
satisfied.
<
2nM(T-tp) ~1
r
"*
(39)
14
PERIODIC ORBITS.
for r>0, p>0, and for upon n, which can always be satisfied by \n\ < mo> the are conditions and T-tn finite. If these resulting expression satisfied,
for
(y-b) substituted
series for (y
in (25'") leads to
b)
satisfies
(25") and
is
(27') converges, and therefore also (27) if m < Mo of n satisfying the inequality (39), for all t in value where mo is the limiting the range t^^t^T. The theorem is thus established.
unique.
Consequently
12.
Generalization to
Many
mi,
Parameters.
,
The
"
differential equations
may
The The
involve
= m* = for = M* = /, are supposed to be regular for Mi discussion can be thrown upon the preceding case by letting
'
'
many
parameters,
M*>
<
5=
7\
Mi
"0tM
(t
= l,
,k).
ft m can be everywhere replaced by ju< This groups the terms of the same degree in /u m* together. The equations can also be integrated as multiple series in the parameters Mt without the use of this artifice, and then the constants of inteMi gration can be determined and the convergence proved. But the method is not essentially distinct from the other, and the details may be omitted.
Suppose the differential equations It a n. may happen that this parameter single parameter depend upon enters in two distinct ways. For example, it may enter in one way so that, so far as this way alone is concerned, the / can be expanded very simply as power series in n. It may enter in another way so that, so far as this way
13.
(
as power series in n are very complex, or even impossible without throwing the equations into an undesirable form.
is
alone
/,
Under the circumstances thus described it is sometimes of the highest importance to generalize the parameter. Where it enters in the first way it is left simply as the parameter Where it enters in the second way it In forming the solutions m is replaced by m to preserve the distinction.
.
is
regarded as a variable parameter in terms of which identity arguments are made, while m is regarded simply as a fixed number. The solutions obtained
are valid mathematically for any value of n whose modulus is sufficiently small, but they belong to the original (physical) problem for only one
But it will be observed that when the particular value of n, viz., for n = m. differential equations are regular for a continuous range of values of this
restriction is of
the literal
no importance, provided the solutions converge value of m has been retained in the solutions.*
this artifice see
for
n = m,
if
Moulton's Introduction
to Celestial
15
II.
Formal Solution
of Differential
Equations of Type
The
dif-
ferential equations
doc
(x,
. ,
xn ;t)
+ nf
if
(x l
,x n
jtj t)
= i, ...,),
(40)
will
be said to be of Type II
(a) the g t {x u
. . .
/x
and not
t)
identically
zero;
(6)
the g (x lf
t
.,
t)
and }
t;
{xi,
.,
xn xn
ft',
are analytic*
in Xi,
(c)
.,
,
.
x,
.
.
and
xn
;
the
g,(xi
t)
and f
{x x
n
i
t)
= xf{t), m=0, Xi
It follows
for
from these conditions that the g and f can be expanded as power series in (x, zf) and /x, which converge if \x xf < r > and n < p > for all t in the range t ^.t^T. Equations (40) can be solved formally as power series in n having the form x = 2z<V, (41)
t
t
1=0
where the
as^ are
functions of
t,
and where
x
t
(to)
=a
(42)
substituting (41) in (40) and equating coefficients of corresponding powers of n, it is found that
Upon
,(0)
dt
16
PERIODIC ORBITS.
c,
,
.
where
are the constants of integration which can be determined in terms of the a Substituting these xT in (44) and integrating, we obtain
. .
c.
t .
I?
where the
^4"'
(48)
found, equations (45) can be integrated, and this process can be continued to the kth step, which gives
.
xf = 2
^;V
<y
(*)
+/*?(<)
(49)
belonging to the complementary function are the same for each step, but the Ff (t), which depend upon the right members of the differential equation, are in general all different. The problem of finding the Ff\ the <p() and the Ff depends upon the explicit form of the differential equations, and
The
<p u (t)
Determination of the Constants of Integration in Type II. At each step there are n constants of integration introduced which can be It follows from equadetermined in terms of the initial values of the x
15.
t
.
that
a,
.
FT
Hence
(c,
c-
Q+
.
2 [ J 47 9u W + F? &)V f
)=a
t
(50)
F<
0)
( Cl
,.
.,cn ;t
(i
= l, = i,
,n),
-Ff (Q
. .
(51)
. .
.
(*
oc),
determined in terms of a
by the first set of equations of (51) Then the F become completely defined, and from the second set of (51) the A -" are uniquely determined since the
.
1
"
determinant A = <p (Q is the determinant of a fundamental set of solutions at a regular point of the differential equations and is therefore not zero ( 18). Then the Ff become entirely known and the Af are determined by a similar set of linear equations whose determinant is the same A. The whole process is unique and can be continued indefinitely.
|
(J
Proof of the Convergence of the Solutions of Type the comparison set of differential equations
16.
II.
Consider
dv
-g
<Pt(y l)
y;
t)
+ M ft (y,
yn
t),
(40')
where the conditions corresponding to (a), (6), (c), and (42) of 14 are satisfied, and where, in addition, the coefficients of the expansions of the <p and the ft as power series in (y yT) and m are real, positive, and greater than the moduli of the corresponding coefficients in the expansions of the g, and the /, for all t in the interval t ^ t g T. Suppose y = a, = 6, at t = t
t
17
= 2 tfV,
It will
(42')
where the yf are functions of t to be determined. the yf are real and positive, and that
be shown that
|?>|a#lio*4i'*3f.
The xf and yf
are defined
(52)
by
(53)
Since,
by hypothesis, the integrands of the second set of equations are real, positive, and greater than the maximum values of the moduli of the integrands in the first set of equations in the whole interval yf > \xf\.
in
the interval
fc
<! <
T,
it
follows that
The
x'"
and y
d
are defined
by
^ +/<>
+ * ^"
<
:*?.0;o;
(54)
,
"?.Sfc^
It follows
.
'
2/
'
0;
= Similarly yf xf = at t t at < = to Equations (54) can be solved by Picard's approximation process.* and Let x2 y t be the k'" approximations to x and yf. Then
from
(48)
and
(51) that
*S -J?/* (*?
'
'
(-i. ...;,
),
(55)
(56)
*s
=.f,' [
2 S6*B- +
* ox*
>
<*
*)]*,
(57)
*7Vai
1905, p. 340.
18
It follows
PERIODIC ORBITS.
from
*o
(55)
and the
relations
between the
Vu
y
>
^
x?l\
^
^
T.
and the <p ^ T; and from the method of forming the successive
t ,
and the ^ that the relations between the it follows from (56) that
/,
4
approximations it is seen that, y{" > x[ t for t ^ t ^ T, for all k. Now Picard has shown* that \im x"l = x for a sufficiently restricted range of values of t. But equations (44) being linear, the range of values for / is precisely that for which the differential equations are valid. f Therefor t ^ t ^ T. The corresponding relation fore we conclude that y > x[ between yf and x can be proved in the same manner, and the process can be continued step by step indefinitely. Consequently the inequalities (52)
l)
| |
l)
are established.
series
Hence, if the series (42') converge when n (42) also converge when m < p' f or U ^ t ^ T.
| | |
< p',
then the
Since
<Pi\yi
1
,
>
i),
it
follows
(40')
from the reference given in 11 that the conditions imposed upon can be satisfied by equations of the formj
"
61
(58)
O-SMg^*.]}
As a consequence
Cj
y )=(yj
follows that
< )
+c
where the
are constants.
Since y
= yf
at
it
= 0. Now let c,
(59)
z=
(</,-</;>)
+M
i,
we
get
Mnz(l+z)
On
of integration
by
log (1+m)
%(! + )-
Mn(t-Q
AM
(61)
*Loc.
cil. \Traitt d' Analyse, vol. Ill, edition of 1894, p. 91. tSee Lei Milhodes Nowelles de la Micanique Ctlesle, vol. I, p. 60.
19
=n
at
,--
q+)'
2/x
"^"(j+io 86
k-w
2
(62)
(1+m)
where
#=
It follows
Mn
(-5)
from
(62) that
if
\p\
< p,
|m|
< 1, and
4 M e K(r w
^j=y
<
1,
then
can be expanded as a converging power series in m for t ^.t^T, and that in this range for t the values of z are such that the expansion of the right member of (60) as a power series in z also converges. Consequently, the y, and x,
satisfying (40') and (40) respectively can also be series in /x for all t in the interval t ^. t ^ T.
expanded as converging
The point
tions are
when the differential equaof the Types I or II, as defined above, and when the interval T t has
to be noted in these results
is
that
been chosen in advance and kept fixed, then the parameter /x, in which the solutions are developed, can be taken so small in absolute value that the series in which the solutions are expressed will all converge in the whole
U ^ / ^ T. As in equations of Type I, there may be many parameters, Mi>Mi> >M*> instead of a single one. The treatment can be reduced to the case of the single
interval
obvious that
if
generalized precisely as was explained in 13. It there are many parameters they may all be generalized.
made
in
an
infinite
is
number
of ways, a great
secured.
17.
Case
of
tions are included in those already treated, they deserve some special attention for the reason that in their solutions the values of m are not restricted by so
many
conditions.
^=id
at
i=i
(1
(t)Xj
(i=l
n),
(63)
where the
6^
= S
t.o
da n
20
PERIODIC ORBITS.
t
which converge if \p\ < p for t*^. t ^ T. Suppose x = a, at solutions can be developed as power series of the form
x<
U-
Then
the
- 2 a? v\
,
(64)
comparison set of
dy
dt
= 2+ >-i
(t)
y>
(63')
series in
p of the form
which converge provided p < p for t ^ t^ T. Suppose also that to^.t^T. Develop the solutions of (63') in the form
| \
\f/*)
> |0*|
for
</,=
It
2</<V.
Jr.O
(64')
in (52) that
can be shown by the method used in proving the inequalities given if y {Q - \a = b then yf >\xf\ for U 5 t ^ T. The conditions imposed on (63') are satisfied by the equations
t
t ,
dy,
i"
in
v 2 M
1
_M^'
p
tj
(63")
which
(
is
the
bj).
(y
b,)
= (yj
for
\
to
^ 7\
Then
It follows that
(y
b).
(63") becomes
ifcfi.
P
<- + -*P
jv
is
The
_M
P
(64")
series in
p.
con-
< ^ 7\ That is, w/ien ^e differential equations are P for to verging for m linear the realm of convergence of the solutions in the parameter p is precisely the
|
<
same as
simple
that of the coefficients of the differential equations. Therefore, in those cases in which the original equations are polynomials in p, the solu-
p. 212.
21
18.
Fundamental Set
: = ke
tl
of Solutions.
Suppose
. .
x' i
{t)x l
(-i,
is
,*),
(65)
where
x\ is
t,
*>(0>
Z(2
<Pa(t)
,x tn =
<p in {t)
('-l,
...,),
(66)
its solutions.
of solutions
A
It will
=\<Pij\.
(67)
6 n are all
t
be shown that A can not vanish for any t for which the In the applications which follow, the 6V are analytic in regular.
general regular for
all finite
and
in
values of
t.
The
result of differentiating
A with
respect to
is
A'=
where the index k denotes that
derivatives of the
<p IL
Sj^|,
column the
in the k'"
t.
with respect to
But
it
<p<t
Hence
A' can
be written
A'=2
22
PERIODIC ORBITS.
is
where A lt
<p lk
in A.
But
it is
known from
and equal
the
to
when j = i. Therefore*
AS9,
<-i
whence
*-*/*',
coefficients of the
n
(69)
where Ao is the value of A at t = 0. The initial conditions are taken so that Ao^O. Thus A can vanish or become infinite only at the singularities of the
main diagonal
If
2
.
If the differential equations were originally of the second order, constant A having the form usually arising in celestial mechanics
n
x[ =
6 tJ Xj
(i
= l, ...,),
= y
( y'
= 2
e tJ Xj
(i=i,
n),
which has the form of equations (65) Since every 6U of this set of equations is zero, the determinant of any fundamental set of their solutions is a constant.
.
19.
The Character
of the Solutions of
a Set
of Linear
Homogeneous
Linear difDifferential Equations with Uniform Periodic Coefficients. ferential equations with simply periodic coefficients were first treated by
memoirs on the lunar theory. About the same time HermiteJ discovered the form of the solution of Lame's equation, which has a doubly periodic coefficient. Starting from the results obtained by Hermite, Picard|| showed that in general a fundamental set of solutions of a
Hillf in one of his celebrated
linear differential equation of the n' order with doubly periodic coefficients of the first kind can be expressed by means of doubly periodic functions of the
h
second kind.
In 1883, Floquet published a complete discussion of the character of the solutions of a linear differential equation of the n'"order which has simply periodic coefficients. In this memoir Floquet gave not only the form of the solutions in general, but he considered in detail the forms of the solutions when the fundamental equation has multiple roots. The forms of the solutions being thus known, the efforts of later writers have been directed
Equation (69) was
vol. IV, p. 403.
first
developed by Jacobi
in
W.
Hill, vol. I, p.
136;
also published
tComptes Rendus, 1877 et seq. \\Comptes Rendus, 1879-80; Journal fur McUhematik, vol. 90 (1881). lAnnales de I'Gcole Normale Superieure, 1883-1884.
23
when
2
dx
-jjf
(oo
+ a,cosJ + a
cos2
-)x
0.
(70)
Different methods for constructing solutions of this equation have been proposed by Lindemann,* Lindstedt,f Bruns,J Callandreau,|| Stieltjes, and
Harzer.*|f
coefficients
In what follows there will arise only equations with simply periodic having the form
*,'
= 2e
tj
Xj
(i=l,
,n),
(71)
where the dtj are periodic functions of t with the period 2tr. It will be assumed that the dtJ are uniform analytic functions of t and are regular for0^^27r. Let
Xa
<Pu
(Oi
z<2
=
it
<Pt2
(t),
,X in
<p in (t)
(t
= l,
. ,
n),
be a fundamental set of solutions which satisfy the initial conditions = if ir*j and <p (0) = 1. It is clear that n solutions can be <p (0) constructed with these n sets of initial conditions, and since their determinant is unity at t = 0, they constitute a fundamental set of solutions.
Now make
where a
is
the transformation
*,-"*,
an undetermined constant.
y:
(72)
differential equations
The
become
(73)
ay = 2d4J yj,
t
any solution
of
in the
form
d'=i,
. .
j=i
ZA,*,
,n),
(74)
The question arises whether it is possible to determine a and the A s in such a manner that the y shall be periodic in t with the period 2 t. From the
t
form
tions
of equations (73)
t
(
it is
of the
(0)
(i=l,
n).
On
upon
- S A, [<p
j=i
(J
(2tt)
-e^
2
,,
(0)]
(i=
1,
n).
(75)
Either
is
all
trivial
the Aj are zero or the determinant must vanish. and we therefore impose the condition
The former
case
24
PERIODIC ORBITS.
e
2ar
where ^ (J (0) = if it*j and ^,,(0) = 1. Putting simply by <p ih this determinant becomes
=s and denoting
tp
(2ir)
<P2\
<P3l
}
<P22~ S
<P32
<Pi3
^3.1
&
0.
(76)
<Pn3
<Ptm
This is an equation of the n'" degree in s, the constant term of which can not vanish since it is the determinant of a fundamental set of solutions. It is known as the fundamental equation for the period 2w. Its roots can be neither zero nor infinite, because the coefficient of s" is unity and the term
independent of
s is A.
the Roots of the Fundamental Equation are all Distinct. s n of (76) are all distinct. Then Suppose the roots s s 2 least one of the first minors of is at distinct from zero when s is put (76)
20. Solutions
,
. . . ,
when
equal tos t and therefore the ratios of the A, are uniquely determined by (75). s t a set of yt is determined by (74) involving one arbitrary constant. Since this solution depends upon s it will be designated by y ik and the
,
For each
corresponding A by A
t
jk
Since
2ar
,
the a
v
is
additive constant
Vl,
where
uniquely determined in terms of s except for the In every case the principal v is an integer.
value of a can be taken, for its other values simply remove periodic factors from the y, Consequently, for the n values of s there are n values a, and
.
from equation (74) there are n solutions, one for each k from
i/ rt
to n,
n),
= e- a*' i A
jk <p
(t)
(t-1,
(77)
Jk
are determined
by
(75).
k,
From
(72),
solutions
x, k
a e
"'y ik
(-l, ,,.,),
(78)
where the y ik are periodic in t with the period 2ir. These solutions (78) form a fundamental set, for, would exist linear relations among the x lk of the form
if
2 Ck x lk
1
(0
t
m
it
(t'
= l, ...,),
(79)
where not
all
the
Ck = Q.
x,
by 2t,
2 Ck x ik
*-i
(t
S C + 2tt) = i-i
sk
x lk
(t)
0,
25
S Ct xik
{t
+ 4tt) =
S Ck si xtk
(t)
0,
I
2 Ck x lk
Since the
[t
(80)
+ 2(n-l)ir]
= 2 C.s^Xn
(t)
0.
Ct are not all zero it follows that the determinant of these equations
is,
,1
,
S2
2
n>(o
Si
S2
26
zero.
PERIODIC ORBITS.
If these solutions are taken, the identities (82)
satisfied.
when the Fundamental Equation has Multiple Roots. the case where the fundamental equation has only two roots equal. The notation can be chosen so that s 2 =Si. There are two cases = according as all, or not all, of the first minors of (76) vanish when s Si. Suppose first that all the first minors vanish for this value of s. Since s =Si is only a double root, not all of the second minors can vanish Hence two of the A t
Consider
.
can be taken arbitrarily and (75) can be solved for the remaining (n 2) of Since the <p u them. Then equations (74) and (72) give the corresponding z are linearly distinct two linearly distinct values of the y can be obtained by
t
. t
taking
first
to zero.
one of the arbitrary A } equal to zero, and then the other equal Therefore in this case there are two linearly distinct solutions of xn
the form
a e ''yn
xt2 =e ait y i2
(t-1,
,n),
where the y n and y i2 are periodic in t with the period 2ir. = s 1} there is but a If, however, not all the first minors of (76) vanish for s single solution of this form belonging to the root s of the fundamental equaail tion. Let x n = e yiX be this solution; it will be shown that the other one belonging to this root has the form
x
'
Xf 2
(y t 2+tytl )
(i
= l, ...,n),
(83)
with the period 2x. Before proceeding to the demonstration a lemma pertaining to a certain type of transformation of a fundamental set of solutions will be proved. Suppose the <p u constitute a fundamental set of solutions. Then define new
t
functions
\f/ ik
by the
relations
if/ (k
= 2 A Jk tp
i-k
(i,k
= l,
,n).
(84)
The
a also constitute a fundamental set of solutions provided no the determinant of the \f/ ik is
\f/
A kk = 0,
for
n
|
ft,
\-L
t
lk <p lt
= \A*\
|^
,|,
where \A lk and \<p tl are the determinants of the The determinant <p (J is distinct from zero and
\ \
|
\
Jk
and
<p ti
respectively.
An A
> ,
2i
22
Jk
=
\
27
which is distinct from zero unless some AM is zero. A special case, which will be used first, is that where all the elements except those in the first line and
in the
zero.
return to the point under discussion. By hypothesis not all the first minors of (76) vanish for s = s Let the notation be chosen so that one of those which is distinct from zero is formed from the elements of the last
x
.
Now
trivial case in
Then A must be distinct from zero in order not to get the which all the ^are zero. Now in place of <pu (i,j=l ,ri) as a fundamental set of solutions we can take, as a consequence of the lemma,
1
columns.
a,<
Vn,
<P<j
(i
= l,
,n;j = 2,
,n).
(86)
Any
fi 1 e
0l,
j/( ,
+ 2 5^
e
a,t
i,
(t
= i,
,n).
Now make
the transformation
xi2
(y (2
+ ty n );
e"
whence
.
V
Since
that,
if
= -
tyn
+ B lVil e
ait
(a '- a
'+
w 2 A j*
by hypothesis the x a = e y n
n
satisfy (71), it is
found by substitution
+ a^a tJ
dtjyjz- y
n).
(87)
Since
t2
shall
2ir
are
Vn
(2tt)
- ya (0) = - 2tt y
tl
(0)
+ 1=2 2B,
s,
(2tt)
e~^- Vt
(0)]
= 0.
On
-2tt
t,
yn
5 1(
(0)
2 Bj
[<p tj
(2x)
- s, it,] =
0,
(88)
where 6 U =
if
jVi and
=1.
is
The
A=
where
|y(0),p
(2*)
-*!,
(27r)-s
5 lB
=
|
0,
is
is satisfied, for if the fundamental equation fundamental set (86), it is found that
the determinant formed from their coefficients. This equation is formed as usual from the
D= pendent of the fundamental* set from which it is derived, and since = 0. Therefore (88) can be has the double root 8 - i, it follows that (s,) B n These equations determine the yi2 solved uniquely for the B 2 and through them the x i2 in the form given in (83).
D=(s-s )D = 0.
1
Since
D is
inde-
LXVI
(1866), p. 133.
28
PERIODIC ORBITS.
a triple root of the fundamental equation, but that If all its minors of the first and second order it is not a quadruple root. = vanish for s Si, three of the A s can be taken arbitrarily and three linearly
Now suppose
s=Si
is
form x l2
, ,
xu
'
ytl
a ''
yi2
x i3
a,t
yt3
can be determined, where the y n y t2 and y t3 are periodic in t with the period 2ir. If all of the minors of the first order of the fundamental determinant vanish, but not all of "those of the second order, then two of the Aj can be taken arbitrarily, and two linearly distinct solutions of the form Xn
will
'
i/n
x {2
2/*2
be obtained, where the y n and y t2 are again periodic. In order to obtain a third solution associated with the root a new fundamental set of solutions
s,
take as
l^'fn,
so that
e *'y i2
<p ti
(i
= l,
n; j = 3,
n),
in the
a ''
form
B,
a >'
y tl
+B
y l2
+ +y
e
B, <ptJ
(-l,
n).
Now make
whence
y t3
the transformation
x, 3
il
t2 )];
-^u-^< + B
2
(a
-a
'),
^/)1
^-
ca '- a
),
2/e+e
S
tml
B 1<Pil
In a manner similar to that in the case just treated the periodicity conditions on the y i3 lead to the equations
0=-2*- Sl
As
t/ (1
(0)
-27rs
t/ (2
found that for s=s determined and that the x, 3 have the form uniquely
in the preceding case, it is
the
.,
are
x l3
, ,
a e ''[y i3
t(yn
+ ?/)],
(89)
where the ytl ya and yi3 are periodic in t with the period 2ir. Suppose now that not all of the first minors of the fundamental determinant vanish fors=s,. Then there will be one solution x,i = e ai 'y and another x l2 = e a,< (y a + t y (l ). It will be shown that in this case the third solution belonging to s, is of the form
i i
x i3
> l
[y i3
+ ty a + \e yil
\.
(90)
29
Take
e
a,,
2/..
as a
,
new fundamental
e a >'
set of solutions
(y l2
x,
+ ty n
by
a
),
*,
(,>it
. ,
n;
j=3
n).
(90')
=B
'
ll
+B
e >'(y t2
+ty n
Then
yn
2
J
=3
B j<PlJ
(;=i,
),
make
yi 3
=-ty -%t
t2
il
+B
(a >- a > )l
+B
1 '- a
a >'
2 Bj<p
tJ
If the
(71), the
yi3 must
+ a, y =
l3
6 ti
y J3
y t2
y n satisfy (73) and the y t2 satisfy (87). that the y l3 shall be periodic are that
since the
Hence
sufficient conditions
yt3
(2ir)
yi3
(0)
(t-1,
n).
0= -2TS
y l2 (0)- 2 7r
in the
sl
t/,
(0)+2
rfi 2 s l2
second column of the determinant of the coefficients may be suppressed. Let this determinant be denoted by In order that these equations shall be consistent it is 2 = that D 0. This condition is satisfied; for if the fundamental 2 necessary equation be formed from (90'), it is found that
of these equations evidently
root.
is
Therefore
Since
s = Si
a simple root
,
.
D (i) = of D not
2
2
,
minors are zero. Therefore the B3 determined, and the x i3 have the form (90). Suppose s=Si is a root of multiplicity I. There
tions,
will
I in number, attached to have the following form
are uniquely
this root.
xn
= =
e ''y n
a
a
0'
= l,
,n),
*x i2
e ''[y t2
e
'
[y t3
+ ty + ty + %?y u
tl ),
i2
],
(91)
x= e^ly^ + ty,^ +
a^
tyi}.
30
If all
PERIODIC ORBITS.
= up to the order the minors of the fundamental equation k-l(k^.l), but not all of order k, vanish for s=s,, then there are A; solutions
t. e.
of the
,
x ti
If
a,
'y ti
t =e ai 'y
ik
now
y(i
a,<
yu
<p t
t+l
formed,
it is
found that
(-,)*#- 0.
D
l
Since the roots of the fundamental equation are not changed by adopting = has s =s as a root of multiplicity the new fundamental set of solutions, t
lk.
Suppose
all
the minors of
of order g
1,
but not
all of
order
g,
a,i
[y + +1
t
2 y i}
~\
...,
+(
= ea ''k t+,+<S|/J.
it
If
k+g<l, by a
will
be found that
D=
Now Dk+e =
admits
s
(s- Sl )
t+
'A + =
0.
as a root of multiplicity I (k+g) and there is a certain number of solutions of the third type of (91), depending upon the order of the minors of Dt+<l which do not all vanish for s=s Continuing, there is obtained finally I linearly independent solutions associated with s, and in a similar way the solutions associated with the other roots of the fundamental equation can be found.
t
.
= Si
Characteristic Equation when the Coefficients of the Differential Equations are Expansible as Power Series in a Parameter /u. In
22.
The
the preceding discussions no explicit reference was made to the parameters upon which the tJ may depend. It will be assumed now that the 8 tJ are
expansible as power series in n whose coefficients separately are periodic in It will be t, and that the series converge for all finite values of t\i \n\ <p.
assumed further that 6u = a iJ where the a u are constants, for ^ = 0. Under these conditions, which are often realized in practice and particularly in the applications which follow, the discussion of the character of the solutions can be made so as to lead to a convenient method for their practical construction. The discussion will depend upon the principles of 19 and the
,
"
= S exj = ~
J
1
2
j=i
[au
S flV]x, + *"'
d = l,
n),
(92)
31
27r,
*}
are periodic in
and
efjfj."
converge for
0)
values of
if \n\
<p.
For n =
equations
(92)
admit
xj
depend upon
as a solution, where the c are constants whose ratios the coefficients of the differential equations, and a (0) is one of
(
= c,e
a "'
On
a2
i
32
Since the A)
PERIODIC ORBITS.
all
A =
|
=
|
0.
(98)
number of solutions, for if a = a, is a solution, + V 1, v any integer. The fundamental equation iar = s. If the values to (76) is obtained by the transformation e corresponding
This equation has an then also is a = a,
of s satisfying the fundamental equation are distinct, the corresponding values of a are distinct but not the converse, for if two values of a differ by an
Only those imaginary integer the corresponding values of s are equal. 0) values of a will be taken which reduce to the a , for m = 0, the aj being
(
uniquely determined by (93). Suppose now that two of the roots of the characteristic equation, say o? and a i\ are equal. Then the solutions of (92) have in general the form
(
= Ajcn e <U
e^'
+ 2 a/|
(99)
SA [*V* +5<y]
J=3
The exception
second term for
to this general
i 1
,
. . .
n,
form is that tcn may be absent from the and this possibility must be considered at
t
those places where it makes differences in the discussion. After making the transformation Xt=e at y,, the solutions for the y are
yi
= Ajcn e {a
y
'
a)t
+
The conditions
the determinant
^A
a
[c ()
e^ -^+e-lx^^].
y,(0)
(100)
=0, lead to
A=
|[c
(1
(l-e< -<>) + e2
<
'2^H2
2
)
r)M
[c
12
(l-e
a - a<,
'
= 0,
(101)
where the elements which are not written are of the same form as those in (98) = 0, the characteristic equation has a root of higher order of If, for m multiplicity, the fundamental equation is formed in a similar manner.
.
23. Solutions
when
of, of,
( 0)
V^l.
The
'u(l
_j(-r>
^=1
\c (J
is
\
unity.
33
its
ft
in
be simply
/
,
any particular case, (98) were an identity = 0* In case it is not an identity, let
in
solutions
would
and A becomes
"?+&
p)
(103)
A = Ao+m ftiw
= (1 - e^)
+p/^(ft,p)
]J
=
(l
-e
a "- a
)
(j*k),
(m)
ft and p, converging for |ft| finite and |p| <p. no a ; ) is an imaginary integer, the expanSince, by hypothesis, (af sion of (104) as a power series in ft and p contains a term in ft of the first 1 and degree and no term independent of both ft and p. Therefore (see 2) it can be solved uniquely for ft as a power series in p of the form
where F*(ft,
p)
is
a series in
( 0)
ft
= pP*(p).
l
. .
(105)
Substituting this value of ft in (103) and the resulting a in (97), n homoA n are obtained whose determigeneous linear relations among A nant vanishes, but for p sufficiently small not all of its first minors vanish,
,
since the roots of the determinant set equal to zero were all distinct for p = Therefore the ratios of the t are uniquely determined as power series in p, for When the ratios of the Aj have been converging |p| sufficiently small.
.
determined, the y* are determined as power series in p, and the coefficient of each power of p separately is periodic in t. A solution is found similarly for each af. The origin of the singularities which determine the radii of convergence
of the final solution series
is
known.
If p is
Consider the fundamental equation, A(s, p)=0, which is a polynomial in s of degree n and a power series in p converging if \n <p. From the algebraic character of A it follows that the only singularities introduced by solving for s in terms of p are branch-points, which are determined by the simultaneous equations
| | \
convergence of the original solutions (95) as t varies general, the final solutions will converge only if p < p.
from
to 2r, then, in
A(s,p)
= 0,
^=
lie
0.
(106)
The
variable $ can be eliminated from these equations by rational processes and the eliminant will converge if |p| <p. Its zeros are branch-points for s
as defined
The
by
which
within
|
=p
can be found
in
any
method* proparticular numerical case by = zero then the solutions is a at there If all are p p zeros the vided simple. If if there is no po the < |p |. for s as a power series in p converge only |p| limit remains p.
Picard's extension of Kronecker's
,
7.
34
PERIODIC ORBITS.
Now
'
its
relation with
s,
viz.,
has a branch-point for M = m, then ft also has a branch-point at the same place since d/3/ds = 1/2 tts is distinct from zero Therefore the series for ft converges only if \n\ < !mo| for all finite values of s.
St
=e
=e
2ai
"r
^T e
If Sl
The
If for
root
mi
is 8,
= s?+ 2
|
st
0)
2
ft
8?
//].
any
such that
m,
< P we
=
|
2 4
m!
then
has an
essential singularity at m
= mi, and
it
converges only
if \n\
< |ah|
zeros determining these singularities can also be found in a special When \n\ satisfies the inequalities numerical case by Picard's method.
The
convergent
are equal but when af af is Congruent two roots of the characteristic equation for l. to Zero mod. V Suppose H = 0, say a'"' and af, differ only by an imaginary integer, and that there Then the equation corresponding is no other such congruence among them. to (104) becomes
when no two af
(1
- e^Y
II (l
- e'^*- 8
Wi
m)
= 0.
(107)
The term
in
ix
The term of lowest degree of lowest degree in ft alone is 4tt P] in general be precisely will second and of the alone is at least degree,
.
This follows from the fact that every term in every of the determinant (98) contains in this element of the In order to get the terms in it alone, special case either ft or /x as a factor. and then the conclusion follows from the are those involving ft suppressed, fact that every term in the expansion of the determinant contains one term from each of the first two columns. In a similar way if p of the a are congruent to zero mod. V l, then the term of lowest degree in ft alone
of the second degree.
first
two columns
is
exactly of degree p, and in /x alone it is at least of degree p. Consider the expansion of (107), which may be written in the form
tf
+ 7n&M + Yo2M +
2
=0,
factored, giving
where y n
702,
are constants.
(ft
6iju)(ft
&2M)
If &!
and
= 6.M + M
Pi(M),
ft 2
= 6 2M + M
i2(M),
(108)
where P, and
series in
Vm
If bi = b 2 the solutions are power are power series in fx. If 7 02 is or n, depending upon the terms of higher degree.
35
zero at least one of the solutions starts with a term of degree higher than the first in /* If the first term in m alone is and if yu is zero, then the solution has the form
/3ii
=c
m'
/3 I2
=- ClM +
v,
....
But
will
in general the solutions are of the type (108), and no other special cases be considered in detail; they can all be treated by the principles of 6 and 7. Thus, starting from the root of of the characteristic equation,
two solutions are obtained. But it follows from the form of equations (98) and (104) that if the start were made from the root of the same values for ft would be found.
,
The
solutions
other ft (k
is
3,
n) are found as in the preceding case, the in the same way, and their realm of
convergence
by
same
types.
when, for n = 0, the Characteristic Equation has a MulRoot. tiple Suppose only two roots are equal, say of = of, and that there are none of the congruences treated above. Then, for m = 0, equation
25. Solutions
(101)
a
becomes
fi
Ao=Ca(l-e
which
A
v
a ~ a T)'\
>>),
c I2
/1
(l-e
( v
a-
W
i '
)' '
\ )
lo +2ttc,.
.,ctJ
{\-e
/i
(a-tt K '
;V\
),-
easily reduces to
5K
in
(109)
is
\
unity.
= af +
ft
is
made
(109)
and the
result
.
2 expanded, it is found that the term of lowest degree in ft alone is 4 t 0[ If the determinant A for of = af is of the special form (98), the term of lowest degree in alone is at least of the second; but if in this case A is of the general form (101), the term of lowest degree in n alone is in general of
fj.
the
first.
Except
af are
of the
form
fc,-yjy)',
where
ft^-^-M
is zero in the
),
P is a
power
series in
m-
When the coefficient of fi expansion of (101), the first term in /x alone is of at least the second degree, and the problem is of the
type treated in the preceding article. = 0, p roots of the characteristic equation are equal, then for If, for n these roots the expansion of (101) starts with $[ as the term of lowest degree ft alone, and except in special cases the lowest degree of terms in m alone is
36
the
first.
PERIODIC ORBITS.
Consequently
in general for
a?-F(j
= <C
the solutions
of (101) are
/3
= e' M 'iVM')
= l,...,p),
where e is any p lh root of unity. Another case is that in which A = the conditions for which are
A(a, M )=0,
for all
|
m,
|^(a,
c^
.
M )=0
If,
ju|
sufficiently small.
Suppose a 2 =
for a
= ai
all
the
first
minors
of
are zero, the solutions of (97) for the ratios of the A, will carry two arbiIf not all traries, and the two solutions associated with a! will be obtained.
minors of A vanish for a = o, then in this way only one solution is found. But it is known from the general theory of 21 that the second solution has the form
the
first
,
x a = e a, '(y a +tytt )
(i
= l,...,n).
On
substituting these expressions in the differential equations and use of the fact that e ait yti are a solution, it is found that
making
V
If the left
+ *iV-
21
l
aa
2
-i
(*-li
...,*)
of these equations are set equal to zero, they become Consequently precisely of the form of the equations satisfied by the yn yt2=yn plus such particular integrals that the differential equations shall be
members
satisfied
when
the right
members
are retained.
The method
of finding the
up
in
29-31.
when,
for
ju
= 0, the Roots of
the Char-
acteristic Equation are Distinct and their Differences are not Congruent to Zero mod. The knowledge of the properties of the solutions and 1.'
their expansibility as
power series in fx leads to convenient methods for = the roots of the constructing them. Under the conditions that for n characteristic equation are distinct and that the difference of no two of them n is congruent to zero mod. V l, it has been shown that there are exactly
distinct values of a expansible as converging
akt
.
.
such that xa = e y ik {i = l, ,n), where the y* are purely periodic, constitute a fundamental set of solutions. It will be assumed that for ft = no two o'"'
power
series in n,
are equal
mod.
the
a*
and that the difference of no two of them 1, and it will be shown that the coefficients
is
congruent to zero
of the expansions of
and yu are determined, except for a constant factor, by the conditions that the differential equations shall be satisfied and that the y tk shall be periodic in t with the period 2w.
37
series
yik
- ly^n
00
1
,
where the
converge for
small.
It follows
(*,
*-l,
).
an identity,
it
follows that
(i,k
i#(2*)-tff(0)
= l, ...
,n).
Therefore each yik separately is periodic with the period 2t. Now the original differential equations (92) after making the transformation x = e at y become
i
tf+fcr 2) [oo+itfv'V
For n
are of, of that for n
P-i
,n).
(no)
the roots of the characteristic equation belonging to these equations It has been shown of. Consider any one of them, as af
,
. . .
but sufficiently small in absolute value, at , in expansible converging series of the form
a a*
and the
i/, t
are
<**
M+
2j
>>=o
at n
y=yS? + y2V+
= s
(in)
yV.
On
substituting (111) in (110), arranging as power series in m, and equating coefficients of corresponding powers in n there results a series of sets of
,
equations from which at and the yiK can be determined so that the ylk shall be periodic with the period 2x. The determination is unique except for For simplicity of notation this an arbitrary constant factor of the y a constant factor will be determined so that y k (0) = c u provided c lk 9^Q, and it can be restored in the final results by multiplying this particular solution by an arbitrary constant.
-
Terms independent of n.
defined
The terms
/z
are
by the
differential equations
(j)'+afi#- 5/wff-o
the general solution of which
is
n
/
(o)
0-i,
n),
(ii2)
(o)\
jC- ltficjF'^
where the
j - k,
iffi
(-i, ...,n),
tjf k
(113)
Since the
of
every
ij
if
therefore
C=
1.
38
If
c,
PERIODIC ORBITS.
were zero the initial condition would be imposed upon another y, not all of which can be zero at t = 0. The solution satisfying the conditions laid down is then
|-A.
Coefficients of n.
(114)
The
terms in the
first
power of m are
The
homogeneous
a'
}t
,
in
is
(116)
same
and the
c tJ
2
where the
g\"(t)
m'c eW- ^
a
ti
(117)
of
The determinant
A=M'
which can not vanish
equations (117) for
(
fci
-e* Vj
i.
*'
for
any
are
finite
value of
(ij '*)'
C,2>'-.-W->%S,
where the
A," are periodic functions of
(118)
27r.
The
j^k
#--<'*-*">'P2+*S
where the
constants.
tfp4),
119 )
(0 are periodic with the period 2t, and the For j = k equation (118) becomes
B%
are arbitrary
0tf)'-A8--tf +
where
(120)
lh A"t after the terms a^z/l" have been omitted from the k column. It is a periodic function of t with the period 2x, and has in general a term independent of t. It can be written in the form
}
is
f%
= d? + Q?(t),
(t)
where d"
is
is
a constant and
Q"
is
mean
value
zero.
Then
()' =
-0 + W).
39
of this equation
clear that
if
ijg
is
member
Therefore
(121) (122)
and
where
= C,
nS-PSJ + flg,
and B is the constant of integration. and substituting (119) (122) in (116), the general solution with the value of a* determined by (121) becomes
Fg
is
Upon
2^
/<o)
Clj
(o)\
e^
-*>>'
+ JCCvF*.
=
oo
(123)
In order that the y tk shall be periodic with the period 2w, all the Bjk must vanish From the condition that y u (0) =c n for all \n\ sufficiently small, except Bu
.
it
follows that
dition that
= 2/i"
fj
1,
).
From
the con-
B2--7-2cP?(0).
Therefore the solution satisfying
i/2
all
the conditions
c c
is
2 [* p * W - f
*2 ()1
124 )
remains to be shown that the integration of the coefficients of the higher powers of n can be effected in a similar manner. Let it be supposed that ai", af a*"" and the \g, y ,... y%~ have been uniquely = 0, determined so that the yf k (t) are periodic with the period 2t and that yf k 1=1, ,m 1. It will be shown that the y can be determined so as
It
l)
, .
. .
to satisfy the
same
conditions.
it is
From
equations (96)
found that
(125)
tJ
y jk
J.
Omitting the terms included under the sign of summation with respect to p, these equations are identical in form with equations (115) except for the Obviously the integrations proceed with the index superscripts (1) and (m). (m) just as with the index (1), and the character of the process is no wise altered by the inclusion of the terms under the sign of summation with respect to p, for they are all periodic with the period 2x and do not change m) the essential character of the g^\t) Therefore a k and the y can be uniquely determined so that the y shall satisfy the differential equations and be = 0. (0) periodic in t with the period 2w, and so that at the same time y The induction is complete and the process can be indefinitely continued. The solutions associated with the other a'j" are found in the same way.
.
40
PERIODIC ORBITS.
27. Construction of the Solutions
when
of the Characteristic
Equation
is
1 and that this be supposed now that of af is associated of The solutions other relation is not satisfied by any pair af. 0) with ?*...., a are computed by the method of 26 without modification. It has been shown that in general a a 2 and the y n y i2 can be developed It will be assumed further that as converging series in integral powers of juis an not the case under consideration exceptional one. The general solutions of (96) for the terms independent of ju is in this
(
l.
It will
x ,
case
m v yn
v _ ^
L tjV Vji c
<o)
Mj
~a
,
be periodic with the period 2w and 0) these equations become, since af ai is an imaginary integer,
shall
(t-1,
...,),
(126)
where
t} 2 l
is
so far arbitrary.
/*
Coefficients of
It follows
from
n must
The
when
4'
their right
'
members
= !,...,
are zero
is
}
(*
|C-Z/<*/
On
n).
(128)
considering the coefficients j as functions of t conditions that (127) shall be satisfied, it is found that
^vW
On
c,/
ttj
yn
1
-+-
z^(f
iJ
yn
(t-1,
n).
and
(*)'- -
l)
(i
<) +<ag
2l
+#
(t),
),
(*)'-
(129)
where the
A'^
and
DJJ'
d
a',"
and and
W_
(o)\,
'
;
.
In the
first
t$>
41
Equations (129) are to be integrated and the results substituted in (128). In order that the y shall be periodic the conditions must be imposed that
( }
,[
42
PERIODIC ORBITS.
The general solution of these equations when the right members are neglected
is
the
On
as (128) except that the superscripts are (2) instead of (1). varying the *, the equations corresponding to (129) are
same
#M
Z?^"
(134)
(J -8,
n).
are exhibited explicitly in the first A",' are precisely the same
In order that these equations shall lead to periodic values of the y the undetermined constants must satisfy the conditions
.<
2>
o,
(\ ^1
T) 2l
?j 21
<"
aj
R J 21
(1 >
_L
-f-
fc
k<
n # 21
R> J./) +d u
_ n U=
<
2 >.
Oj
R< a, iJ 21
<D
+0
_1_
J><
21
(135)
= B%
The
first
= 3,
n),
linear in a ,
( 2)
and
is
zero.
The determinant
is
A=On
eliminating
,<> -~ rl2i
j<"_i_
,
c
(
>
On "T
""
21
a-i
= -6 2
,(1)
1
+ a-+^[6 A +
<
^]
121
>
j?
of (130),
it is
found that
A=V5,
where D is the discriminant of (131). Since by hypothesis D is not zero, " the determinant A is not zero. Hence af and -B 2 are uniquely determined by (135). Having determined JS", and af, equations (134) are integrated and the results are substituted in the equations corresponding to (128). Then the conditions that y (0) = are imposed and the final solution at this step becomes
tf
(0)
(o)\
n
|
(136)
(*-l,
,),
where
is
43
The next step is similar to the preceding and all the equations are the same except the superscripts are (3) and (2) in place of (2) and (1) respectively.
The determinant of
In fact
(
the same.
all
the equations corresponding to (135) is precisely succeeding steps are the same, and the whole process
can be repeated as many times as is desired. The solutions associated with a 0) are found as they were in 26. af, For congruences of higher order similar methods can be used, and in the cases which are exceptions to this mode of treatment the existence
.
. .
the Characteristic
Equation are Equal. It will be supposed af = af and that all the remaining af are mutually distinct and distinct from af. The solutions af can be computed by the method of 26. It depending upon af
,
.
. .
has been shown that the two solutions proceeding from a^ are in general The detailed discussion will be made expansible as power series in VJi. only for the general case, where
a1 a2
=af + a|V + af M +
=af-aV + af M (137)
Terms independent
oj
ju.
of n are defined
by
(138)
(^'+ftC-J ,4C0
l
(*-l,...,).
The
is
= v^c
tl
+v?!(ci2 +tcil )+
ifZc/*?-?)*'.
initial
(139)
In order that the t/f shall be periodic with the period 2ir and the cn of yfi shall be obtained, the V must satisfy the conditions
value
Vn=0
The
solution satisfying all the conditions
is
(j
= 2,...,n).
then
(i
|C-*.
Coefficients of
= l,...,n).
(140)
The
by the equations
.
.
+ ai
iff-
(t-1,
).
(141)
44
PERIODIC ORBITS.
neglecting the right members, the general solution of these equations
,
On
is
|tf
"
(142)
The method
Wl)'c
-a%
(t-1,
n).
it is
found that
(j
?,
G$)'0
= 2, ...,n).
Consequently
$-*-?,
On
y" = (B\-a?t)c
(l
$-3
)+ J=3
(i-2,
...,).
(143)
becomes
'
+B
2l
(c{2 +tc(l
2#X
to
'
'
To
must
and
make y
(0)
= 0,
the
Bn
<?
= <*>,
BSCu+BSfCa-O,
all
<
fJ-8,
...,).
(144)
Then
the conditions
c
become
>
y
where the constant
not
all itself
(145)
a"' remains as yet undetermined. It is to be observed that the coefficients of a"' can vanish, for otherwise the determinant \ctJ would vanish.
\
Coefficients of
p..
The
coefficients of
0-1.
.>
d46) and by
-af
C(l
^ ^it is
same form
as (142),
found that
,
(-^C+
Cfi
)(ar>)
C<
(147)
45
The
(148)
r
(n?f)
-+e-<
^%()
the type
c os
.
(,'-8,
),
of
t.
The
first
equation
a )J(t l
s[n
cos
itdt= x JIM + -^
s in
i/41 J
'
-
sin
cos./
1-
-aJZJtdt=p*jt
When
type
1
Hence
at this step
1=3
(149)
J=3
where the
terms in
initial
P jl(t)
(
t,
dfl
tions (147).
D Equations (149) are the general solutions of equaIn order to satisfy the conditions for periodicity and the condition ^'(O) =0, the constants a"' and B' must fulfill the relations
Af,'
and
(150)
#21 -o-i -r
n
still
a u>
0-B*
(J
=3, ...,n).
The constant af
remains undetermined.
c
The
solutions
now
are
(151)
having the period 2ir. After found that of is determined making a choice as to a"', provided bfi^O, The process can be conuniquely by the periodicity conditions for t/Jf. determined constants are the tinued indefinitely and uniquely. The other solution associated with a is obtained by taking the other determination of a"', and the solutions depending on af, af by the method of 26. The chief types of cases have been treated, and the exceptions to them are developed similarly according to the forms indicated in the existence proofs.
periodic functions of
<
known
it is
46
IV.
PERIODIC ORBITS.
29.
2w and the
are Distinct.
Take the
x' t
i^x^g^t)
...,),
2tt.
is
(152)
left
members
are periodic in t with the period set equal to zero the form of the solution of (152)
t
{t)
For the
2,A,
(153)
where the jjj are arbitrary constants, the a, are the characteristic exponents which are supposed distinct, and the y t} are periodic in t with the period 2ir. By the method of variation of parameters, it is found that
a
iv',e
% =g
}
(t).
(154)
of the coefficients of the is the determinant of the fundamental set of solutions. Since the 6i} are assumed to be regular for all finite values of t, it follows from 18 that this determinant can not vanish for any This determinant is finite value of t.
rj'j
The determinant
s ajt
where A
when
If A,
t
A becomes
jj,
th
(t),
are
',=
|'ea
",
(155)
and consequently
;-/^e- *dH-B,.
:
(156)
finite in
the
0^^27r.
Therefore
it
series
= + ^ A <# m
If
=l
[ojf
mt
+U
sin mt]
a j+m 9^0
(i =
l,
n;
m = \,
L
oo),
a,
a'+ra
f
a'+m
i
,
so that
t,--
wF (0+B
where theP,(0 are periodic with the period 2t, and the B, arc constants of
t) }
= J B,e*%+
jtPffllks
-l.
,n).
(158)
47
Now
suppose
at
=kV
je~
l,
where k
'[a k
is
an
integer.
dt
iVZ::rT
cos
= 5
B,e
a*
yu
+P
(0
+ |(a, - 6, v^T)
a >'
yu
(159)
where the
characteristic exponents are distinct and none of them is Therefore, t/ congruent to zero mod. l, and if the g (t) are periodic with the period 2r, then the particular integrals are also periodic with the period 2w. But if some
Vi,
ticular integrals in general contain, in addition to periodic terms, the corresponding parts of the complementary function multiplied by a constant times t.
Case where the Right Members are Periodic Terms Multiplied by an Exponential, and the a t are Distinct. Consider the case where the g (t) have the form
30.
t
the f (t) being periodic with the period 2t. imaginary, this form includes such cases as
t
When
= V^T
l
is
a pure
(t)
4-ie
are transformed
x,~<?'ft (t),
(160)
u by x = e z
t
( ,
they become
(lei)
in
29.
If the character-
zi
+ ^-iy^m,
of (161) are a,
and are
distinct, then the characteristic exponents also distinct. Applying the results of the pre-
if
no
a,
is
V^i,
then
where the
(160) are
Q (i)
t
2ir.
= iB.e^'y.j+e^Q^t)
(*-l,
n),
(162)
48
PERIODIC ORBITS.
But
is
zy
for the x t
become
(t)
+ct
y{t
(163)
be stated as follows: // the g,(t) have the form e^f^t), where f,(t) =f,(t + 2ir), and if none of the characteristic g,(t) exponents is congruent to X mod. V i, then the particular solution has the form
These
results
may
x^e^QS)
where the
the
Q,(<) are periodic
,
(i=l,...,n),
but if one of the characteristic then the particular solution has
;
exponents, at
is
2w
V^,
form
x^e^Q^+cJe^y*
where the
31.
Q,(t) are periodic
(i=l,
,),
2ir.
Case where two Characteristic Exponents are Equal and the Right Members are Periodic. Suppose o 2 = a, Then the solutions of
.
x[
6 (J X j
(t'
= l,
n),
in general
*-% e a
+r, 2
(y a +ty (l )+
2 Vj e a*yf}
(i=l,
n),
(164)
-2 8
iJ
xJ
=g
(t)
(165)
v[+e
a"
(y i2 +ty il )r,' i
2 e a* y + J=3
rj' }
,
(J v'j
=g
(t)
(t-1,
)'.
On
= \g&)
(Va
+ ty a
), 2/
...
. . .
-a,(
t
\e-
*v 2=\ya
&v'j=\y<x
gt (t), y {3
y,\eajt
tom+tVa),
V, '.
\ytj
.
\
ytn\e-
(j
= 3,
n),
The expansions
of these determinants
^e-^P^D-e-HPAt),
9',-e-<*J>(*)
^ = e- a "P (0,
2
(i-3,
.
n),
where the
Pj(t)
(j=1,
n) are periodic
49
it is
V^T.
Then
easily
Vl
r, 2
= e-^R, (0 - e~ a H R = e- a *R (t)+B
2
2 (t)
+,
(167)
where
v (t), (t) are periodic with the period these values in (164), the solution becomes
. .
2ir.
On
substituting
x
n
=B
e "y n
+B
(yi2
+tyn
+ 2 5,e a'y + 2 R
Hence,
if
t (t)
y ir
exponents are equal but none of them is congruent to zero mod. 1, then the also solution is with the particular periodic period 2ir. The case where one aj(j = 3 n) is congruent to zero mod. v^17! is a combination of the present case with the second part of that treated in = 29, and that where a 2 l does not differ a! is congruent to zero mod. %/<
,
.
.
in
any
essentials
a2
= a, = 0.
to
Now
(166)
f
suppose a 2
= a = 0.
1
become
v' 2
=P
(t),
fae-^PM
2-k
(i
= 3,
n).
in the
form
= aj + 2
[a kJ
coskt+bkJ
Hence the
t],
are found,
by
integrating, to
iI
tl 2
+2&1 (0+<M--iatf--iS,(0,
=+R (t)+a
2
j
2 t,
li-tT-RM+B,
where
tf-3,
,n),
R,{t)=R
These values substituted
x
t
{t
+ 2ir)
n
(j=l,...,n).
complete solutions
n
1
= B y n +B
i
2 2* ]
y(l
+a ty + 2 R y u
2
(2 t
Hence, when
acteristic
exponents are not only equal, but are zero, then the particular integral involves not only t but, in general, also f outside of the trigonometric symbols. Of course it may happen that a and a 2 are either or both zero.
l
in
Those particular cases have been treated which will be most useful in the applications which follow. Any others which may arise can be
50
PERIODIC ORBITS.
V.
32.
The preceding
considerations find
immediate application
known
periodic solutions.
dynamics in the study of small variations from Suppose there are given the equations
^=X
where the
A',
-i,
...n),
(168)
t ,
satisfied
by
(169)
as,-*(0,
the
of t with the period 2w. <p,(t) being periodic functions the generating solutions. Let the initial conditions be varied slightly and put
(0)
= <Pi<P)+P
t ,
(170)
(
for
any t
will
be
*-*(*)+{,(*),
the
,
(171)
being functions of t which for at least a short interval of time will On substituting (171) in (168) and expanding the right remain small. it is found that members as power series in the
_,
% = 2 If^
t
*>
(t=
1,
n),
(172)
Since the
,
<pt (t)
are
are expansible as the coefficients of (172). The power series in theft which, by the Cauchy-Poincare theorem, 16, converge for any preassigned interval of time provided the |ft| are sufficiently small.
all
The
terms in the
ft
terms of (172), or
=f +!+
+f
*-*
'
<
i73 >
These equations are known as the equations of variation. Suppose the solution (169) contains an arbitrary constant c, that one not contained in the differential equations (168). If c = c + y, the <p arc expansible as power series in y of the form
is,
(t)
*(0 =*?()+*
Obviously
*'
y+h
2
+
{
-i,
,..,)
. .
dc
J^
dc
^
by
l,
,n)
Methodes
The subject of this section and other related questions have been treated Now>clles dz la Mecaniqus Celeste, vol. I, ehap. 4.
Poincar<5, Les
51
41
&^t
(il,.../)
(174)
a solution of equations (173). One such constant is always present when the do not contain t explicitly and the <p (t) are not mere constants, for then the origin of time is
is
arbitrary.
Hence
in this case
*-jg
is
<<-i,.
....1*
aw
a solution of (173). Another such constant usually present is the scale factor which determines the size of the generating orbit. If there are p such arbitrary constants in the generating solution, the equations of variation
have p solutions
If
of the
form
(174).
is
independent of
t,
x.)
= c,,
t
t . ,
where
an arbitrary constant, the x can be replaced by <p (t)-\- in the The result is integral and the integral can be expanded in powers of
c, is
t
7=
dF lt ,dF lt
dx~^
is
'
'
'
dx~^
,dF +^+
.
The constant 7
a power series in the /?, and therefore the linear terms are
,
^*+*+
;
+g*''
x t
(I76!
which
therefore an integral of equations (173). periodic functions of t, the x having been replaced
is
t
Si.
Theorems on the
Characteristic
Exponents.
The
existence
of
arbitrary constants in the generating solutions and the existence of integrals of equations (168) have an intimate connection with the characteristic
in general of the
form
(177)
fc-*"/(0
o;=i,
...,),
the/, being periodic with the period 2ir. The solutions (175) have the form (177), but since the <p are periodic so also are their derivatives, and the There is an exception only characteristic exponent of this solution is zero.
t
are constants, in which case the solution (175) disappears. solution obtained by differentiating with respect to the scale constant, which will be denoted by a, will, in general, have the form
if
the
<p t
The
^'=^(0+^(0,
\f/ t
The characteristic exponent of this solution is being periodic. zero. If the generating solutions have p distinct arbitrary constants, the equations of variation will have at least p characteristic exponents equal to zero.
and
52
PERIODIC ORBITS.
From
it
= e"%(0
,n),
(176),
we
get
%-tj
The
left
i'
= e ~ ait
U=h...,n).
(178)
of these equations are periodic with the period 2ir, except, It follows, therefore, perhaps, for coefficients which are polynomials in t. that either the a,=0 mod V^l, or all the 7^ = 0. In this connection a
members
congruence has the same properties as an equality, and they need not be If all the a } are distinct from zero, then distinguished from each other.
7 =
</>
= 1,
n)
and
(178)
becomes
2^/^
Since the determinant
|
(j-l,...,).
(179)
if
5* 0,
a*f
f0
<-'
>
Therefore, unless the integral (176) vanishes identically at least one of the and that 7 ," ?* 0. characteristic exponents is zero. Suppose that a x =
(
possible then to solve the equations corresponding to (178) uniquely for the dFJdxt in terms of yf and the jtl
It
is
.
Suppose now
there
is
a second integral
(x t
x)=c
Then
2|g^
On
for the
= const.
n fundamental
solutions
Sf/
If
= 5<-
2|f;/o
= 2,
.
^^
. .
(j
= 2, ...,n).
(180)
0,5*0,
and therefore 6^ =
can be
solved uniquely for dF^/dx, in terms of ftJ and from a constant factor,
8[
dJ\^d_F\ dx dx
'
'
But if F and F2 are integral is identical with the first. then there must be at least two characteristic distinct, exponents, say c^ and
x
which are
zero.
it
follows that
if
the equazero,
tions of variation
admit of p linearly distinct integrals not identically then there are p characteristic exponents equal to zero.
53
d% = dV
dt
2
dx,
(-l, ...,*),
the case usually in celestial mechanics, they equations involving only first derivatives by writing
is
which
may
be reduced to
dx,
_ dt~ Vt
is
dy,
'
_ dV
{
~dt~~dx
(t-1,
n).
If
of variation are
formed by putting
y
t
* d*
= *>
(0
+6
= <p[{t) +Vt
~dt~ vi
-n
tt-1,
n),
dm_
eft
d*V
dx,
d*V
S1
x
'
a
2
' '
F
'
(181)
dx
dXi dXi
dx,
dx
members of these equations (considered as contains a determinant matrix) only zero elements. Therefore, by 18, the determinant of any fundamental set of solutions of these equations is a
of the right
constant.
of the
fundamental
ait
set of solutions
i,-**&(0,
s
ajt
y<j=e
gtj (t)
u=i,
n).
follows
-of P (t) This must therefore be a constant, from which A= e that the sum of the characteristic exponents is zero since P (t) has
.
the period
2ir.
Suppose d
dt
#",
nf and
f
d
nf
(i=l,
are
of
equations (181).
>
Then
2
dn?
(1)
(1)
F
(182)
dx,dx n
'
and
also
dtf
dt
^
"'
2)
drfl
dt
it
dx dx
i
M
l
2,
+ #L^ +
,
dx,dx 2
+ dx dx
l
at
(183)
From
these equations
follows that
2/(* dF
The sum
of these
"dfJ
a)
'
A/A6
is
d<
6 dt)
(184)
two equations
dt
dt
2 [(<
which can be written
+tl
l fc)
drit
dt
)-bnt^di)]=>
fi
(185)
,7 at
i=i
^ (m
*?i
J~ u
54
PERIODIC ORBITS.
Consequently
2 The
(gV-SV) = const.
(
(186)
relation (186) between any two solutions leads to important concluU) sions respecting the characteristic exponents. Suppose the {$" and ij are
e=
ftl (t),
v? = e^g (1 {t)
0-1
,n;j =
,2),
where / and g u are polynomials in t with periodic coefficients, and that they constitute a fundamental set of solutions. On substituting any two of these solutions in (186) and dividing through by the exponential, there results
(=1
M/,^-/,,?,;)^/-"^.
of the left
(187)
It follows
a.j+a t
0, or
= 0.
of this equation that either It will be shown, however, that y Jk can not be zero
.
member
for every k.
2n. Suppose j is kept fixed and give to k all the values from 1, = = 2 Then one equation of (187) is an (k 1, n). Suppose y Jk others in the are linear the and The determinant identity ftJ and the gu of linear is the of of this set determinant the fundamental set equations Hence they can be satisfied only by ftJ ^g tj =0. But and is not zero. this also is impossible since / and gu are a solution of the fundamental set. Therefore not all the y jK can be zero. Hence for some k
. . . ,
. .
a,+a = 0,
But
t^0.
(188)
since a, is any one of the characteristic exponents, it follows that corresponding to each characteristic exponent there is another one which differs
from
it
only in sign.
of the a j are equal but not equal to zero, then there are two others are also equal and which differ from the first two only in sign. In
. ,
If
two
which
order to show this suppose a, = a >+1 = am Then a m = a m+l because from = If t,* = (188) it follows that a,+a m 0, y Ju i^0. (k=l, 2n, k*m), then (187) can be solved for / and gtJ uniquely in terms of f,k and git
. .
.
Now the corresponding equations for ftJ+1 and g Uj+l 2n). from (187) only in that j is replaced by j+1. They can be solved uniquely for /M+1 and g ]+l but this solution will differ from the solution for / and g tJ only by a constant factor. Since this is impossible it follows that Yj+i.m+i^O, and In the same consequently a J+l + a m+l = 0. manner it can be shown that if p of the a. j are equal, then p other a.j are also equal and differ from the first set only in sign.
l (k will differ
.
t,
Chapter
II.
ELLIPTIC MOTION.*
34.
The
Differential
Equations of Motion.
in
Then they attract each other as material points, their orbits are plane curves, and the differential equations which the motion of one relative to the other must satisfy are, in polar coordinates,
dt
2
\dt)
~u
'
diVTtJ'"-
In writing these equations the origin has been placed at one of the bodies and the variables r and v are measured in the plane of motion. Equations (1) are easily integrated, and the integrals show that the relative motion is in a conic section for any initial conditions. If the initial velocity is not too great the orbit is an ellipse, and the discussion will be
limited to this case.
(1)
certain conditions the orbits are ellipses, it does not express the coordinates The explicit developments are obtained explicitly in terms of the time.
means
through solving Kepler's equation, generally by Lagrange's method or by of Bessel's functions. In treating elliptic motion as periodic motion the expressions for r and v in terms of t will be derived directly from the
differential equations.
integrating the second equation of (1), and eliminating dv/dt from the first, it is found that
T
2
On
by means of
this integral
dv
dt
=C
d
>
F(m,+w _ dT~? +
r c
*~
2)
(t}
= the major
=the
= the mean angular velocity in the orbit; = T the time the body passes its nearest apse; t =w(t T) =the mean anomaly.
It is
(2)
i!
thatf
2
1
coV = A;
Vol.
K+0,
to Celestial
= (m
A;
-|-m2 )a(l-e
it
)=coV(l-e
).
(3)
was published
XXV, May,
1907.
fMoulton's Introduction
56
PERIODIC ORBITS.
On making
second equation of
becomes
_<
(It
+r
.e.
(4)
Equation
(4) is satisfied
by the
r
circular solution r
= a.
= a(l-pe),
7t,
(5)
where p =
fore p
dp/dr =
at
t = x.
at t
= a(l+e).
Theremotion,
values of r in elliptical
and therefore
+ 1 ^p^
Upon
*
The second term
all
+ -2Z* =0
(6)
in this
if \e\
the values of p
e for
f?
and the
first
+P=!
(2)
(*'+!)[*'-
(H-2)p>'-V,
(7)
equation of
35.
Form
of the Solution.
it
The
first
be
considered.
After
determined from
(8)
by a simple
quadrature.
14-16, and therefore Equation (7) belongs to the type treated in can be integrated as a power series in e, and \e\ can be taken so small that the series will converge for ^ t ^ 2ir. Since the body moves so that the law of areas is satisfied and completes a revolution in 2t, p is periodic with the period 2ir. Consequently, if the series converges f or <! t ^ 2ir, it converges
indeed, possible to find the precise limits for |e| within which the series will converge for all values of t, and outside of which they will diverge for some values of r. The problem was first solved by
for all real values of
t.
It
is,
Laplace,*
who found
all r if
e< 0.6627
which
is
far
any
planet or satellite in
ELLIPTIC MOTION.
57
The
P=2
where the
Pj{t) are functions of r. tions, the constants of integration
J=0
P ,(tK
(9)
conditions
p (0)
= l, Pl (0)=0
(-l, ...oo)j
it
^(0)=0
-l,\..oo).
(10)
As
is
whence
(11)
i
o
p,(r+27r)e
=i
o
p j (r)6
is periodic. at its nearest apse when r = 0, and the orbit is symmetrical with respect to the line of apses. Therefore it follows that p is an even function of r. Since p is periodic in r identically with respect to e, each
Since (11) is an identity in e it follows that Pj(t-\-2t) = Pj(t) simply the definition of periodicity. Therefore each p t separately
.
But
this is
The body
is
Pj (t) is expressible
as a
sum
t.
changed, then the body is at its farthest apse = when t 0. Consequently changing the sign of e and increasing t by ir does not change the value of r. Since r = a(l pe) it follows that
If
the sign of
e in (6) is
p/ry-^p/T+TrM-e)'and when j
If
1
.
(12)
t;
Therefore when,/ is even, pj(r) involves cosines of only odd multiples of is odd, p } (t) involves cosines of only even multiples of t.
we
we
get
=0
J=0
(13)
x=vi-e
where
2J(i+i) 2) c*
At+l
c,
orr~n
t,
.
t2
(*++
+**=*)
14 )
t,
then there are two cases to be and even, (6) when i is odd: considered, viz. (a) when t must be odd, and (a) When i is even an even number of t\ must be even. Therefore the odd odd the number of jx i\ multiplied by number of odd ix multiplied by even j\ must be even. All those factors p in (13) for which i x is even involve only even multiples of t, and those for which jx is even and i x odd involve only odd multiples of r. Since there must be an even number of these terms involving only odd multiples of
Suppose i,i,+
+i.j,
+i
is
even;
i is
r, their
t.
58
(6)
PERIODIC ORBITS.
are
When i is odd it follows from (14) that an odd number of t, +ij,+i is even odd, and from the hypothesis that i,Ji+
. .
i,
it
follows
that an odd
are odd. only if Since the whole whether raised to an odd or even power. of r multiples number of odd i x is odd, and an odd number of them are multiplied by an odd j x it follows that there is an even number of terms p'; where j x is even and ix is odd. Therefore their product will be cosines of even mul, ,
of iA
jt
term ix jx can be odd ij, are odd. When j x is odd the term involves only even
. .
tiples of t.
That
is,
in the right
member
involve only even multiples of t. easily proved in a similar way that the coefficients of odd powers of e in (13) are odd multiples of t.
powers of
It
is
Upon
in e of the
integrating (13),
it is
found that
v is expansible as
a power series
form
v
= CT+Zv
e\
t
(15)
even, and of odd The coefficient c is unity because, the ellipse * is odd. multiples of in v increases fixed space, being by precisely 2ir in a period.
where
vt is
sum of t when
sines of
even multiples of
when
i is
and arranging
in
powers
of
e,
Upon we obtain
2
substituting equation
P ;e'+
2
2
Pj e>
+ [-6 Pop +3p (l-p -6pD+2pS(3-5pD]e + [-6p(p,+3p?+3p p -2p )+3p (l-2p )-5p2(8p -2-3pDK
1
(16)
.
where p*
is
coefficients of like
the second derivative of p } with respect to t. Upon equating powers of e, the differential equations which define the
several coefficients
become
(a) po "'+Po
(6)
(c)
= 0,
(d)
The only
is
= cosr.
(17)
Then equation
(6)
becomes
Pi+Pi=
13 ~ COS2t.
2
is
The
Pi=-f4 c
s2r.
(18)
ELLIPTIC MOTION.
59
.
and
(c), (d),
=
g
(-cost+cos3t),
p3
2t-cos4t), =-^(cos i
(19)
The
is
defined
by an equation
of the
form
(20)
p'j+P^A^+A^
where the
cos
t+
+^coskt,
J)
Since p s is periodic, A[ is zero for all values of j, and since p s involves only even or odd multiples of r according as j is odd or even, all the A^ with even subscripts are zero if j is odd, and all the A with odd subscripts are zero if j is even. On putting Af equal to zero, the solution of (20) satisfying the initial conditions (10)
is
*
-|
A ^'
1
are
known
constants.
2 [JL
-K
n
A'D
11J
...
cos T ~
2 x^and
(5),
AV)
cos Xt
1
(21)
On
for r
in (9)
becomes
= a-U
[cost]
+|[l
cos
2r]e
4
+ |[cos2t-cos4t]
On making
of v
is
+! [cost + j-
cos3r]e
(22)
the
same substitutions
in (8)
and
found to be
v
= t+
[2
sin
r]e+ [f
sin
2t]
(23)
It will be proved that no 37. Additional Properties of the Solution. It has been seen that it is true
it is
true up to j
1,
and then
it
will
be proved that it is true for the next step. The general term in the right member of (7)
1=t
is,
apart from
its
numerical
After substituting the series (9) for p, any term of degree coefficient, p p"V, where j in e arising from this term has the form plpl'pl'
'e'.
+ X, + X +
2
+\ = il,
K
X,+2X 2 +
it is
+^+{1 =
7.
After eliminating
from equations,
found that
X+2X, + 3X 2 +
where obviously
k
+(+l)X,=;l,
(24)
<j.
60
PERIODIC ORBITS.
p K are respechypothesis the highest multiples of t in p pj in the Therefore k+1. highest multiples p*' are pi, p\', tively 1, 2, Consequently the highest multiple in (k4-1)X*respectively X, 2X,, is ... the product +(/c-f-l)X K which is pi' X+2X.+ by (24). 1 Therefore the highest multiple in the expression for pMs j
By
p^
jl
A similar discussion of
multiples of t greater than
(8), (13),
,;'.
and
(15)
shows that
vt
In certain cases where the motion convenient to suppose the body moves in an This conception ellipse whose position and form are constantly changing. is at the foundation of the theory of perturbations originated by Newton, and has been essential in the work of most writers on celestial mechanics. One of the historically interesting and important problems has been the theory of revolution of the line of apsides of the moon's orbit. When Clairaut first made a computation of the rate of this revolution under the
38.
of the Rotating Ellipse.
is
Problem
not strictly
elliptical, it is
supposition that it was due to perturbations of the moon's motion by the sun, he obtained an amount about half as great as that furnished by obser-
Later work by himself and others has shown that the discrepancy but at first he sought to relieve the that difficulties by supposing gravitation does not vary simply as the inverse of the but that it also depends upon a term in the inverse distance, square We shall solve the problem of the motion for third power of the distance.
vations.*
was due
law of force as a further illustration of the power and simplicity of the methods which are being used here. The steps in the solution are almost exactly parallel to those used above, and it will be noted in the course of the work that they would not be fundamentally different if the added term were not such as to make the peculiar simplicity of this problem. The differential equations of motion in this case are
this
*r
df~
T ( r
dv
\dt)
\ + fcVi+K)
i
_ "
F (m. + m
?
2)
'
d( r dv\_ dA Tt)~>
2
1
(25)
where, for simplicity in the final formulas, the constant coefficient of the term in the inverse third power of r is given the formA;2 (m The p. is an 2 )a/x.
+m
is
r*f t
by means
of
=c,
(26)
which the
first
equation reduces to
2
A;
dV
__
d7^7
_ ~ + Ar^+ra.,) ?
Ill,
(m,+m2) aM ?
,
'
{27)
24 and 27.
97
chap.
4,
and particularly
articles
61
The
by
Circular Solution.
areas,
2
We
c,
an arbitrary constant of
defined
for
which
c
r is constant.
Let a and u be
(28)
k2 (m Let
r
+m )=w
2
3
,
= kVm
+m )a.
2
= a(l+p), where p
1
is
a constant.
1
'
Then
/I
I
(27)
1
becomes
.
a+py
or,
ft _L<v^2
(i+pY
(i+p)
3 \3
M>
expanding as a power
2
series in p,
z
p-3p +6p By
1
=- M (l-3p+
(29)
can be solved for p as a power series in n, conIn fact, the additional term has been verging for sufficiently small. chosen in such a way that the power series reduces to a single term, but it is evident that this condition is in no way essential to the It is process. found at once that
this equation
| |
and 2
jut
P=-MIn this case the solutions of equations (25) and (26) are
(30)
r-(l-),
.-^ypS-yfl-ri-S*!^-!,,
a, v
,
(31)
and
T.
We
shall
now
derive a
solution of equations (25) corresponding to the elliptic solution in the ordinary two-body problem. It will involve four constants of integration which are arbitrary except for the restriction that the orbit shall not deviate too
widely from a
the series.
circle,
a condition which
is
is
The
let
c
solution
Now
{m^m
= wV, w(t-T) = Vl + 5
2)
(32)
r,
\
the time of passing the nearest apse, and a(l /* e) is the arbivalue of r. Therefore the initial value of p is unity. The trary constants a and e are defined by the first and third equations at t = T, u by
is
initial
the second, and 5 is a parameter to be determined later.* There are so far three arbitrary constants of integration a, e, and T; the fourth is introduced
in integrating
equation (26).
With
(33)
Les Methodes Nouvelles de
pe)
this.
la
Mecanique
61.
62
PERIODIC ORBITS.
Equation (33) admits a periodic solution, as is known from the fact that the orbit is a rotating ellipse. However, it will be shown directly by forming
the
first
')
'+
**}">> *>-9<r). r
<p(r
,
(34)
Suppose the
initial
r2 two roots, viz., r e) For p small it also has two roots, r[ and r2 near r, and r2 respecThen <p{r) <0 if r<r[ or r>r' 2 Consequently it Suppose r[<r2 tively. it will increase until follows from dr/dt = V<p(r) that if r is increasing at / = r = r 2 when the radical changes sign, after which it will decrease until the
= a(l
must vary.
r[.
The
is
-*r V^Jrj
If r is periodic
'
35 )
The
then p is periodic also. existence of the periodic solution can be established directly from a proof made of the possibility of a construction similar to that used
Equation
(33) can
2
2
p''
(36)
-]
+[-6p+6e+
-]p +-
-J,
where the right member converges so long as |p+pej<l. this equation can be integrated so as to express p as a power and e of the form
p
By
14-16
5,
series in
p,
= P(d,n,e;
r).
(37)
The
series will
converge for
all t in
the interval
0^r^27r
if
\e\,
jp|,
\S\
are sufficiently small. now avail ourselves of the arbitrary parameter 5 to determine the Since will determine d so that the period shall be 2-k in r. period.
We
We
equation (36) does not involve t explicitly, sufficient conditions for periodicity with the period 2t are
P (2ir)=p(0),
p'(2tt)=p'(0).
(38)
These equations are not independent, for (36) has an integral corresponding to (34), which is a relation between p' and p that is always satisfied.* This integral has the form
'
p '4-(l
(39)
I,
p. 87.
63
Suppose the particle is projected from an apse so that p' = at t = 0. Then C is a power series in e, p, and the initial value of p, which is unity. Let the general value of p be 1 +<r. Then the value of the integral at any time minus 2 its value at t = will be p' plus a power series in a, e, and p, vanishing for a = whatever e and p may be. The terms coming from the right member all involve e or p as a factor, but there is a term coming from p 2 which involves a alone to the first degree. Therefore (39) can be solved
Hence uniquely for a as a power series in p' e, p, vanishing with p' = 0. if p' = at t = 2it, then will the first equation of (38) necessarily be satisfied, and consequently it may be suppressed. This result can also be shown from a certain symmetry which is par2
,
Equation
Suppose
l, p'
at
t=0
P=/i(t), P'=/2 W.
consider the differential equations obtained formed by the substitution p=Pj p'=-p[ t=-t
. ,
Now
transin
the
new
the initial
Pi=fi(r )=J(-T)=p,
1
p!=/a(Ti)=/,(-T)= -p.
of r
Therefore p
is
an even function
and
p' is
an odd function
of r.
Since it is an odd function it must also have Suppose p' = at r = r. = t. t as Since p is even in t it has the same value at r = been zero at t = is the same at r ir as it the was it has at r = at system Consequently = with the 2w. Hence if period p' t=tt, and the motion is periodic at t = -k in order to at t = 0, it is sufficient to satisfy the condition p =
ir.
secure a periodic solution of (33) with the period 2ir. It will now be shown that the second equation of (38) can be solved = It is found uniquely for 5 as a power series in p and e, vanishing with p 0. that 1, p by integrating (36) and imposing the initial conditions p
=Poo+Pi M+Poie+
"
'
'
where
P00 = cos\/1
+ 5t,
t sin
P.o=
- f Vl +
i
,
Vl + 5r,
(41)
64
PERIODIC ORBITS.
substituting these results in the second equation of (38)
On
in
and expanding
powers of
5 also,
we
get
p'(2x)-p'(0)
=6
[-!*+ (terms in
+m[
3
7r
5, p,
e)]
p, e)l
+ (terms in
5,
= 0.
(42)
There are no terms in e alone, for when p = 0, the orbit becomes a fixed = satisfies the periodicity condition. Hence (42) can be ellipse and 8
power series in p and e vanishing with p = d obtained from (42) is substituted in (37), p becomes periodic in t with the period 2ir, and is expanded as a power series in p and e which converges provided |pj and \e\ are sufficiently small. It can be written
solved uniquely for When the value of
5
as a
p=
p ,/iV,
f
8= 2 2
tj
3,,/iV.
is
(43)
From the reasoning of 35 it follows that each p The range of convergence of (43) is limited
place the inequalities
5
1 \
separately
periodic.
in
two ways.
In the
first
<5
<p
e
|
Then the inequalities |p|<p, (37) may converge for 0^t^2tt. must be satisfied in order that the solution of (42) shall converge
for
5
1
1
|e|<e,
and give
p and e satisfy both of these sets of inequalities the convergence of (43) is assured for all t. After the explicit development of equations (43) has been made the results can be substituted in (26), when v will be determined by a quadraa value less than
8
.
| |
When
ture.
The
final
form of
v is
v= S
The constant
found that
<-o ;=o
Si;mV
f or
(44)
v=r
it
was
4 1 . Direct Construction of the Non-Circular Solution. In carrying out the practical construction of the solution, we shall make use of the facts that = 0, (6) p is expansible in the form (43), and (c) each (a) p^l, p'^0 at r ptj separately is periodic with the period 2ir.
On
of p
and
substituting (43) in (36) and equating coefficients of equal powers e, it is found that the several coefficients must satisfy
(A) Pm+Poo = 0,
(5) Po,+P<n = (l-3p2o),
( 5 u)
+ 3)Poo>
1 1
(E) (F)
= Po2+Po2 3poo(l-2p01 -2p^), Pn+P= -8 p00 -5 oPoi + 5 o(l-3p2 )-6p00p -3p ,+3(l = _ &20POO ^10Pl0~35, o poo 3p P20+P20 6P00,
11
10
4p^,),
10
65
The
solution of (A) satisfying (a) is p00 = cos r. The solution of (B) is in given equation (18). The solution of (C) is not periodic unless the coeffi= 0. cient of Poo is zero. Imposing also the condition (a), 5 10 = 3, p 10 The term p02 is given in equation (19). Equation (E) becomes explicitly
Pa
+ Pn =
(c),
n cosr 3/ 2 cos2r. Upon imposing conditions (a) and the solution of this equation is found to be
<5
3 /2
*n
= 0,
The
explicit
form
of (F)
(45)
cos
r,
whose
and
(c) is
= 3,
and
8
p20 = 0. as
(47)
series in e
Hence the
P
power
and
p.
are
= cost+
cos2r]pe
2
+ |[
5=-3/i+3M
where, from (32), t
2
cost
+cos3r]e
(48)
-|,
+
to
be replaced throughout by w(t T)/Vl + 8The differential equation defining the general term is
is
,
Po+Pi>= -8,jPoo+Fij(8K\
Suppose
all
Pkx)
(k=1,
z'-l; X
= l,
1).
the pKX and 8 KX for which k<?', X<,;' have been found, this equation can be written in the form
pr ;
Then
<y
-Mi'* COS
kt,
(49)
and
(c) is
UJ)
df>
(A
A
-\
-++
A \ cost ffrjww-
rwi-
cos2t-
(50)
cos
KT.
The
Therefore the the coefficients are uniquely determined. far as be desired. may process can be continued without modification as After the transformations (32) equation (26) becomes
5 (;
and
all
^=
Upon
that
t;
vT+^jl+2p+2p +(-i+3p
)e
+6p M e-f-3p
+
J.
it is
found
2
J;
=[l+2p-f-3M
+
and
v
]T+[2sinT]e+[Jsin2T]e
(51)
>
where
t is to
be replaced by
e,
w(tT)/Vl + S. The
.
of integration are a,
T,
66
42. Properties
of
PERIODIC ORBITS.
It has been proved that each p tl each pit is periodic in t with the that 0, an even function in r. Hence each p tJ involves only
the
Solution.
(except
Poo)
period
2rr,
vanish at t p' tJ
is
It
depends upon no
the differential
There
is
when
equations are satisfied for the same initial values of the variables by p = 0. Or, it is seen, from the development of the second term of (33), that the only terms which are independent of e involve p to the first degree alone. Conse-
quently the right members of the equations corresponding to (49), which 5 w -|-/(p)] cos t alone. The periodicity condition makes define p(0 will be [ and the initial conditions then make p<0 = 0. it necessary to put 5 (0 =/(p)
,
expression for the right member of (51) has no term independent of p, except unity, in the non-periodic part, for when p = the ellipse is fixed and this part reduces simply to r. On making use of all of these facts
The
artifices,
period in t=2x, or = 2ir\/l-|-5/w. It follows from (51) that the longitude of the radius has increased in this interval by 2t. Therefore the line of apsides has moved Vi 5 [l+2p+3p 2 ]
forward
in this interval
Hence
d&
its
2t
2 it.
vT+M1 +2 m +3m +
w
2ir -2ir
r,.
-i
fKtn
where d> is the longitude of the nearest apse. The parameter p can be determined so as to secure any rate of revolution of the apsides not too great.
Chapter
III.
43.
The
lum
falls in
Differential Equations. The problem of the spherical penduthe class of those which can be treated by the methods of
periodic orbits. Its simplicity makes it particularly well suited to illustrating these processes, and its value as an introduction to the subject is increased
it is easy to verify the results experimentally. It is doubtwhether there is a problem which is superior in these respects. Let us take a rectangular system of axes with the positive 2-axis directed upward and with the origin at the fixed point of the pendulum. The pendulum is subject to gravity and the normal reaction, N, which we shall take with the positive sign when directed outward. If we represent the radius of gyration by I, the motion of the pendulum satisfies
+y +z =
2
2
mx" =
Nj,
my" = N^,
mz'^N^-mg,
(1)
where the accents indicate derivatives with respect to the time. The last three equations of (1) admit the integral
m(x' +y' +z'
2
2
=mv = mg{-2z+c
2
),
(2)
where
c t is
The normal reaction exactly balances the centrifugal acceleration of the pendulum due to its motion and the component of mg along the normal to
the surface; hence
z mv N=--r -FN =-r +mg J
,,
mv
where
FN
is
On making
Ci)
.
use of
(2),
we
get
(
3)
become
+y +z = t
2 2
2
,
y"=faz- Cl )y,
(4)
x"=
The
last
$z-c^x,
z"=f
(3z-c )z-g.
t
equation
first.
After
it is
independent of the others and is therefore solved first. solved the second gives x in terms of t, and then y can be found
is
67
from the
68
44.
PERIODIC ORBITS.
Transformation of the z-Equation. It will be convenient to transform the last equation of (4). It admits the integral
)=f(z),
(5)
where
a constant of integration which is independent of c,. If subtract this equation from (2) and reduce the result by (2), we find
c2 is
we
g(c,-c2 )
= (x*+y")
z'\
(6)
Now z ^ f
pendulum
cl
c3
= 0.
In the case where the spherical pendulum reduces to the simple In this case takes the value = I, and at the same time z' = = In the spherical pendulum z> I when z' 0; consequently in
.
this case c l
c 2 >0.
Hence
problem
C,
r2
c 2 5;0.
Now
is
consider equation
If
=f(z
zero or positive
and
/(_) = -oo,
/(z,)=0
/(--?(*-<*,)
2
0,
(-**+,
l
f(+l)=-9(c -c
)m
/(+oo) = +oo.
Therefore the equation /(z) = has always three real roots. Let them be a, a 2 and a 3 where the notation is chosen so that a! ^ a 2 ^ a 3 Then
, ,
equation
(5)
becomes
z
'2
"
(*
- a,)
- a,) (z - a,)
seen that
On comparing
2 (a
1
this equation
3
l ,
with
(5), it is
+ a +a )=c
2
a,
a 2 +a 2 a 3
+a
at
f,
2a
a2 o
=-c
<!
f,
(7)
that a 3 satisfies the inequalities I follows from the last of (7) that c2 is negative if a 3 is positive.
It will
z'
now be shown
3
a3
0.
It
On
and
=a
putting
in (5),
we
get
c2
= 2a
+p(-2a +c ).
3
1
By
hypothesis the
first
is
positive,
positive,
Therefore
must be
:
implication from the last of (7). Therefore a 3 ^0. Some special cases may be indicated
(1)
It follows
,
from
2a 2 = c,
= c2
(5), (6),
and
(7)
that a 3
= -I
a,
a 3 = 2l.
suppose
it is less
than +1.
lum making
finite
=c = c,
a
is not determined, and we shall This case is that of the ordinary simple penduoscillations. In the sub-case where a 2 = -I, we have
The constant
a2
21
of (4) are
x = y = 0,
z=
I.
69
from the same equations that a 3 = -I, 2a1 ~C1 *eL V The constant a x is not determined. If a, > I, we have the case of the simple per dulum swinging round and round, and a, - a 3 > 2 1 In the sub-case where a = l, we have c^c^+21, and the solutions of (4) are x=y=Q, z = +l. Therefore the second of (7) can (3) If a 3 = 0, it follows that a 2 ^0. not be satisfied except by a 2 = 0, a = oo Then, from the first equation we
.
If a 2 = +1, it follows
the case of revolution in the xy-p\a,ne with infinite and of course can not be realized physically. Excluding this case speed, and that of the simple pendulum, the constants a, a 2 a 3 satisfy the
get
c,
= oo.
This
is
inequalities
-Z<a3 <0,
l<a2 <+l,
z
^> +1.
a3 )u
2
.
a3 =(o 2
(8)
(5)
becomes
u'
2
= ^"i
&I
aa) (1
-u
(l
V
at
^^ u
a3
) /
(9)
Also
let
where
is
an arbitrary
/x
2
initial
time and
8 is
The constant
0^/x^l2
Then
(9)
becomes
(11)
ii
where u
variable
is r.
the
first
derivative of
to the
new independent
(12)
The
first
derivative of (11)
w=(l+6)[-U+M)w+2 M w
45. First
].
Demonstration that the Solution of (12) is Periodic, and that u and the Period are Expansible as Power Series in ju. It will first be
shown
except
that, for
any
when
ix
= \,
conditions belonging to the physical problem, the solution of (12) is periodic. By the fundamental
initial
existence theorem of the solutions of differential equations* the solutions of For real initial condi(12) are regular in t for all finite values of r and u.
tions the coefficients of
u and
its
derivatives expanded as
power
series in t
are real, and by analytic continuation they remain real for all finite real values of r provided u does not become infinite. Now consider the curve
and that u is positive. Then u is Suppose u = u at t = increasing at a rate which is proportional to the square root of F(u ), and it It can not increase beyond + 1 for then u continues to increase until u = 1 would become a pure imaginary, and it has just been shown that it always remains real. It can not remain constantly equal to 1 unless n = \, for
F = F(u).
otherwise u
Therefore, unless n =
111.
1,
will increase
70
to
1
PERIODIC ORBITS.
;
= 1 1 and then decrease to then, in a similar way, it changes at u 1 the funcfrom a decreasing to an increasing function. That is, at u= 1 tion F(u) changes sign and u varies periodically between + 1 and This result follows, of course, from the fact that in the present problem u is the sine amplitude of t, one of whose properties is that of having a real period, but the argument given above applies to much more general = F(u). It may cases, and the result can be read from the diagram for F be mentioned in passing that the imaginary period of the elliptic function is associated in a similar -way with the portions of the curve between + 1 and + 1/Vjl, and between 1/ V/I and 1 The period of a complete oscillation is found from (11) to be
.
p=
which
well
is
/"+ r+
du
2 2
(13)
finite unless
n=
We
It is
known
that
p=
In
t
vfeE1 +(i)"*+G3)W
is
+G:l:::S
tl)
)>+
<>
the period
That
the period is expansible as a power series in /x, and in the present simple case the series converges provided /x < 1 The constant 8 has so far remained undetermined. If we let
is,
.
|
v^r-[i+(i)V+(a)V+
the period in t will be simply 2w. the required value of 5 is
5
],
On
2
we
find that
|m+^m +
(16)
which
a power series in /x. consider equation (12). By 14-16, this equation can be integrated as a power series in n, and |m| can be taken so small that the series will converge for all t in the interval O^t^t, chosen arbitrarily in advance.
is
Now
greater than P and the constant of integration is chosen so that the first integral of (12), that is, so that the period of the motion is (11) P, then it follows from the periodicity of the solution that the series will converge for all finite values of r. That is, both u and T arc expansible
If t, is
is
as
power
series in n.
71
of
Equation (12)
is
Peri-
be instructive to give another demonstration of its periodicity. In many problems the former methods can not be applied. By 14-16, equation (12) can be integrated as a power series in the two parameters m and 8 of the form
will
(17)
where the utJ are functions of r to be determined by the conditions that (17) shall satisfy (12) and the initial conditions, and where |5| and \n\ can be taken so small that the series will converge for any interval O^t^t^
chosen arbitrarily in advance.
values w(0)=0, solution as in to the same from we see that order and (11) get u(0)=a, The a= at the end. before we must put subsequent steps of this Vl-\-8 demonstration would not be essentially modified if we took general initial
initial
We may
take the
conditions.
From
we
a
get
=
(=0 1=0
utJ (0)
V>
=
<=0
1=0
tt(0)
v,
from which
it
follows that
(i,i=0,
oo),
u(0)=0
On
powers
of 5
M O0 (0) = a,
<(0)=0
(i+j>0).
(18)
and
n,
we
get
Woo
+ ^oo = 0,
first
+ W =-W
10
00
(19)
is
The
solution of the
u00 = a
sin
r.
(20)
substituting this result in the right member of the second equation of (19), integrating, and imposing the conditions (18), we find
Upon
w,o
=
2
sirXT
rC0ST
21 )
Hence we have
u = a sin r + %
powers of
and
m-
(22)
Since the right member of (12) does not contain r explicitly, sufficient conditions that u shall be periodic with the period 2x in r are
u (2 tt) - u (0) =
0,
u (2 x)
(0)
0.
(23)
72
It will
PERIODIC ORBITS.
now be shown that the second one of these equations is necessarily Let w = 0+v, u = a-\-'v, where v and v satisfied when the first is fulfilled. vanish at t = 0. Then (11) becomes
{a
+ vf=
(1
On making
).
There are two solutions of this equation for It has the form at t = must be used.
v,
v=vp(v),
where p(v)
is
(24)
Since u(0)=v(0) = 0, it follows from a power series in v. the first equation of (23) that w(27r)=0. Then, from (24) we have v (2w) =u(2tt) virtue of the existence of the That is, by m(0)=0. integral (11), the second equation of (23) is a consequence of the first. Now let us consider the solution of the first equation of (23). Upon
substituting
u from
(22),
we
get
= Tab + terms of
from the theorems uniquely in the form
It follows
higher degree in
and
/j..
(25)
5
of
= mP,(m),
where
p,(/x) is
(26)
if \n\ is
a power series in
/x,
On
sufficiently small.
U=lu {r)
i
| | 1 1
(27)
which converges for all 0^t^2it for \/x\ sufficiently small. It is sufficient that fx and 5 satisfy the conditions necessary to insure the convergence of These conditions can both be satisfied by (17) and the solution of (25). values of m different from zero because the expression for 8, given in equation (26), carries m as a factor.
Hence, the periodicity conditions having been satisfied, we have proved is periodic. It has been found to be expansible as a power series in it, and the period, which in t is
that the solution
r=
is
27rZV2(l
+ 5)
V<jf(ai-a3 )
It is clear that this mode also expansible uniquely as a power series in n. of demonstration applies to a wide class of equations, for the explicit values
of only the first terms of the right member of the differential equation, the general properties of its convergence, and the existence of a first integral have been used.
73
47. Third Proof that the Solution of Equation (12) is Periodic There is a certain symmetry property of the solutions which can be used to
simplify the demonstration that the motion is periodic. It will be shown that the motion is symmetrical in t with respect to the value w = 0, which, by (8), corresponds to z = a 3 or the lowest point reached by the pendulum. and that the solution of (12) for these Suppose u = 0, ii = a at t =
,
initial
conditions
is
w = /i(t),
/,(0)=0,
*-/.(*),
/2 (0)=a.
(28)
Now make
Then
the transformation
u=
(12)
v,
u = v,
t=<t.
xv l
is
3
(29)
becomes
v
= (l + 8)[-(l+iJL)v+2
(30)
Hence
if
= 0,
=a
v-M*),
where
/,
/.(0)=0,
#-/,(),
/,
M0) = a,
and f2
,
(31)
t.
and /2 are the same functions of <r that On substituting (28) and (31) in (29), we get
of (28), are of
w = / (r)=-/ (-r),
1 1
*-/,(r)-+/.(-r).
at t = 0, and hence it Therefore u is an odd function of t when u = follows that sufficient conditions that the solution of (12) shall be periodic with the period 2ir are
w(0)=0,
!*(*)= 0.
initial
(32)
The
condition w(0)
= 0.
Hence
=
The
of 6
and
/x-
(33)
solution of this equation for 5 and the further discussion are precisely like the treatment of (25), and lead to the same results.
It has
= 5 iM + 5 m*+
2
(34)
is
u=u
+u h+u
1
2 2
ij.
(35)
74
with the
only of
period
initial
PERIODIC ORBITS.
conditions u(0)=0, ii(0)=a, u will be periodic with the In fact, the value of 5 is given in (16), but we shall make use
period 2t.
its
explicit value
2ir,
would
not, in general,
we have
u(2t+t)-u(t)= 2
and
[t*,(2T
+ T)-u,(*)]/i' = 0,
Therefore
l,
. . .
u (2t+t)-u ] (t)=0,
j
(j
= 0,
oo).
(36)
is periodic
Instead of determining the solution by the initial conditions u(0) u (0) - 0, we may use u (0) = 0, u(w/2) = 1. Or, by (35)
,
= 0,
2 u,(0V=0,
Therefore
u,(W-i.
we have
o(0)
= 0,
*(g)-l.
w,(f)=0,
= 1,
(37)
2,
. .
,(0)=0,
oo).
determine u and h by the conditions that (12) shall be satisfied identically in /x and t, and that the conditions (36) and (37) shall be fulfilled. By direct substitution and equating of coefficients, we find
^0
Now we
+ ^0=0,
+w =-(1 + 5 )w
1 1
+2m*,
I
(38)
1
+25
The solution of the first equation of (38) Then the second equation of (38) becomes
satisfying (37)
is
u =
sin
t.
^i+w^-Q + SjsinT-l-
sinr-i
sin3r.
(39)
In order that the solution of this equation shall be periodic we must set the Then the solution satisfying (37) is coefficient of sinr equal to zero.
,- j,
of (38),
!,--
[anr+anSrl.
(40)
+8
27
) siiit+
sinr- - sin3r1
sin5r.
(41)
75
Upon
setting the coefficient of sinT equal to zero, as before, subject to the conditions (37), we find
52
and integrating
=
,
u2 =
[7
sinr
+ 8 sin3r+ sin5r].
(42)
The induction to the general term can now be made. We assume that u u _ 5j (_! have been determined and that it has been found that u is a sum of sines of odd multiplies of r, of which the highest is 2j + l. The differential equation for the coefficient of \x is
,
. .
u
t
+u = -S
t
+F
(8 K
MX )
(k,
X = 0,
t-l),
.
(43)
in the 8 K
and
The
where
ra
or
1,
"i+" 2 +"3=lor3,
ritK-\-v l
(44)
(v l
\ 1 -\-v2 \ a -\-v3
'\
= 3
or
i1
+v 2 +v 3 = l
or 3).
It follows
of
odd
Vj
of these equations that there are an odd Tt is a sum of sines of odd multiplies of
number r. The
highest multiple of r
tf( =
(2A
+ l) +^
(2X 2 +1)
*>
(2X 3 +1).
On reducing this expression by the third of equation (44), it N = 2mic+2i+l, the greatest value of which is, by (44), N = 2i+l.
f t
is
found that
(45)
Hence
(43)
i
may
t
be written
i)
+4h an(2t+l)r,
be periodic, the condition
'
(46)
where the A 2 \ +l are known constants. In order that the solution of (46)
l
shall
*-A?
must be
satisfied,
47 )
8t
Then
the solution of
,-?
a ^'
sin
r+a?
sin
3r+
(X
l
=_
=
X= l
4<S
'---' 1h
(48)
4X(x+T)
a,
Zl(- l)
^~ Z
A-i
4X(X+1)
was assumed
for
The
ua
,
solution at this step has the same form as that which rt _, and the induction is therefore complete.
.
76
PERIODIC ORBITS.
On
collecting results,
we have
M = [sinr]+
5
^r 10
(49)
0+|M+gM +
2
And
2
we
1
=a
+ S5(o -o
1
3)
(50)
we have made no
it
In the direct Solution from the Integral. use of the existence of explicit
shall show that the integral (11). furnish the or to solution itself. tions,
We
Equation
(11)
/x)=0.
series in
fi,
Since
it,
u,
and
we have
(51)
>o+^iM+*>2M
'
=-
is
an identity
,5 2
,
. . .
in n, it follows that
= <P (u ,u
u
.
u
. ,
6,)
=0
(*-0,
...),
(52)
where
is
a polynomial in u It follows from (11) that <p, 6, of the second degree in ii\, and ie\,, and of the second or fourth degree in u\ and wx,. Therefore <p is a sum of cosines of even of t. It is seen without multiples difficulty that the highest multiple of
<pt
is
linear in the 5 X
t in
<pi
is
2i+2.
<pi
Hence we have
+fl*?+1 cos(2i+2)T = 0.
values of
(f=l,
r, it
= B?+B?co&2T+
all
(53)
follows that
B&=0
oo;
X=0,
,t+l).
(54)
77
The B^lare functions of the a$ +1 and 5 and equations (54) constitute a searching check upon the computation of these quantities. If by some numerical accident an error were not indicated at any particular step, it would be revealed by the failure to satisfy (54) at some later step. But the c4x+i can be computed from (54) as will now be shown. Suppose ua m_ 52 Then we find from _, have been computed.
,
, .
. .
<
.,
Pi
= 2uo u +2u
t
u- 5,(1 -ul)+^ (u
t
ux
5 2X )
(X
t-1),
the
\J/ (
being
tion of (48),
first
equa-
^ = 2ar>+0-i5 +C =
2
1
0,
(55)
2x
2x
(X-2
i),
constants.
On
solving these
last,
we
I
find
(o
C 2<+2
n>
2<+l
_ a 2X-l~
(o
,
X-f- 1
A
a 2X
,
C 2x
2C\
1 )
*
'
' "
'
'
''
0,-oa 3
-\-ZL 2
o1
--d
{
(56)
J
-L
which uniquely determine 8 and the a 2 x +1 Let us apply these equations to the computation of the first terms Suppose M = sinr and take i = 1. Then we find from (11) that
.
of u.
<p t
whence
<
- 2ai" - 1 5l + = 0, 1
(1)
B? = 4a;" - 1 1, = 0,
*
B? = 2a -
= 0.
Therefore
a3
i.
n (1)
ttl
16'
16
'
The
computing the solutions have been developed. They are both reduced by the general discussion to the mere routine of handling trigonometric terms. When they are both used each serves as a check on the other. There is little advantage with either over the other, so far as their convenience is concerned; the integral has a slight advantage in that when using it the computations are made with cosines, and the disadvantage of involving u to the fourth degree instead of only to the third.
different
Two
methods
of
78
PERIODIC ORBITS.
II.
50.
The x-Equation.
is
was given
x+[a +0 lM +02 M
where a
is
+
,
]x = 0,
, . .
.
(57)
are periodic functions a constant independent of /x, and 0, 2 in this case the it. Since the of t, having period period can always be made 2-k by a linear transformation on the independent variable, we shall suppose for the sake of uniformity that the period is 2-k. This problem belongs to the class which was treated in Chapter I, and can be transformed to the form considered there. But it is now in the form used first by Hill, and later by Bruns, Stieltjes, Harzer, Callandreau, etc., and because of its historical interest and for the sake of comparison with this earlier work, it will be treated directly in the form (57)
.
51.
The
ditions
x(0) = 1,
Characteristic Equation. Suppose that with the initial conx(Q) =0 the solution of (57) is
x = <p( T ),
*-*(t),
v(0)
l,
#0)-0,
initial
= 1,
is
*-*(t),
i*(r), A=
*(0)=0,
= 1.
The determinant
<p(.t), v?(t)
<Kt),
f (r)
being equal to unity at r = 0, these solutions form a fundamental set. In fact, A is independent of t for an equation of the form (57), as was shown in 18. It follows from the initial conditions that its value is unity. Hence
the general solution of (57)
is
x^^tO+c^t),
where
c,
x = c <p(t)+c2 \},(t).
1
(58)
and
let
c2
Now
where a
us
x = e",
is
(59) (57)
an undetermined constant.
Then equation
J
becomes
(60)
+2a+a
The
5
2 +[a +0 M+0 2 M +
1
1-0.
is,
by
2
(58)
and
(59),
= e~ aT [cMr) +C,*(t)],
(61)
79
the question whether it is possible to determine a in such be way It follows from the form periodic with the period 2tt. of (60) that sufficient conditions for the of with the period 2ir are periodicity
We now raise
that
shall
(27r)-(0)=0,
!(2)-i(0)-0.
after
On
ip,
we get,
initial
values of
<p,
\f/,
and
fa
2a
[e-
>(27r)
- l] Cl +e- 2a V(27r)c = 0,
'
e-
2a
V(27r)c 1
+ [e- 2a V(2x)-l]c = 0.
2
(62)
J
In order that these equations may have a solution other than the trivial one = c2 = 0, the determinant of the coefficients of c, and c must c, 2 vanish; or,
2a
D = eSince
<p{2ir)-e
<K2tt),
*(2x)
2a *
+(2ir)-e
is
(63)
is
= e ^)^_^(2 7r )+^(2x)]e 2a +l = 0,
e
2a,x
(64)
of
44.)
(1)
of
When
ratio of
c,
the value of
to c2
is
is
not the
is
a,
(64)
determined.
Then equations
(2>
give
and
(I)
periodic with the period 2x. the other root e~ 2a '\ The
We
(1)
and
1,
<2)
by using
and
We
shall
<2)
(0)
and multiply
the solutions
if
by
Consider equation
<2, a
a, is
and
if
all real
<p(2ir)+^(2ir)< values of r;
becomes
x=
infinite as r
becomes
infinite
= oo except for special initial conditions. It is found from (57) = = cos a w for = 0. Therefore the part of a, which is indethat <p(2ir) \j/(2ir)
oo for r
/j,
aVi. Suppose a is not an integer; fi pendent then a, is a pure imaginary for all real values of n whose modulus is sufficiently If a is an integer, the value of c^ for real values of n whose modulus small. is small may be purely imaginary, real, or complex according to the values
of
is
of <p(2w)
and
^(2ir).
and will be derived. There are two particular solutions of (57) of the form x = e a,T such that a, a V l for n = 0, and such that is periodic with the is a constant reducing to a,T The coefficients of (57) by hypothesis are and e -a,T period 2ir viz. e of multiples of t. sums of cosines all real, and the 6, are Therefore, if the
of the
Some
of
(1)
<2)
(1)
<2>
sign of
will
be a solution.
Suppose
80
PERIODIC ORBITS.
o, is
a pure imaginary.
is
(2)
Then
it
follows that
"
(\^TY)
0>
(
l).
the sign of t,
it
follows that
= (r)
- r)
unchanged by changing
finally, since (57) is
t,
it
And
l
and
follows that
r(v^=T,T)=r(-v^T,-r), |(v^t,r)-{*(-V=T,-r).
and in the expressions for sine terms of the real and pure imaginaries,
the sign of
ll> <2)
and
(1)
and
<2)
differ
only in
l.
series,
we have
(2,
=2 =2
[a j [a j
cos jr cos jt
lbj sinjr],
It follows
where the
a,j
and the
bj
(,)
from
this that it is
a,T
(1>
sufficient to
compute
Any
e -a,r(2).
solution of (57) can be expressed linearly in terms (1) It follows from the initial values of <p, <p, and ,
and
(2)
and the
Since
ll)
and
C2)
are periodic with the period 2 x, and since their initial values
are unity,
we have
I
(e
2a.x
(27r).
But by
(64), e
t
2a
''+e-
2a,T
= ^(2T)+^(27r).
When a = V
/3
l is
a pure imaginary, as
we have
cos27r = <p(27r).
This equation has the same form as that developed by Hill in his memoir on the motion of the Lunar Perigee, Acta Mathemetica, vol. VIII, pp. 1-36, and It is also derived differently in Collected Works, vol. I, pp. 243-270.
Tisserand's Mecanique Celeste, vol. Ill, chap. 1. Equation (64) or (65) furnishes a means of computing the transcendental a, because, under the hypotheses on (57), <p can always be found, for example
as a
power
series in n,
Though
this
constitutes a complete solution of the problem and there are no difficulties in carrying it out except those of the lengthy computations, we shall find it convenient to make use of more of the properties of equation (57), and
to find both ai
and
otherwise.
is
When m =
z
>
known
aV [a l e
to be
iT
= a,eoV= r +a
a,
e- av/:r
'
T
,
x = aV^l
-a2 e- a
'
vr=:iT
]l
(66)
where
81
series
of (57) that x
0<t^2tt
if
is
sufficiently small.
Hence
(67)
cos2a7r-e
2a,r
2 ^(2*V,
w
<p x
-sin2air
2a,r
+
OS
&,(2it)m
D = e"
-osin2ax
= 0.
x
+2
(68)
(27r)M\cos2a7r-e
4-
2 ^ x (2ir)/i
This equation expresses the condition that (57) shall have a periodic solution form (59), where is periodic with the period 2w. If it is satisfied = where v is any integer. These by a a it is also satisfied by a = a +v different values of a do not, however, lead to distinct values of x. We
of the
Vl,
which reduce to
l for
n = 0.
are
0)
o,
Solution.
0)
ci2
For n = the principal solutions of (68) There are three cases depending l.
of (68) that
D = P(a,
where
It is
M ),
/x
(69)
a power series in a and n which vanishes for l that also easily found for /x = and a = * a
P is
= 0,
a=
l.
<>P
da
4T
^azT
2 sin2a7r e* "^^*,
which is distinct from zero under the conditions of Case I. Therefore it follows from the theory of implicit functions that (68) can be solved for a in the form
a1 a2
=aV
=+aV^T+< M + aiV+
,
'
*
i
+C4"
ju
+ <4
2>
'
'
'
(70)
,
where the
is
converge for |/x| sufficiently small. 2 J) a reciprocal equation in e ", it follows that a[ =
series
- 0^0 = 1,
).
82
If
PERIODIC ORBITS.
substitute either of the series (70) in (62) we shall have the ratio If this result and equations of c, and c2 expressed as a power series in mwill be in are substituted (61), expressed as a power series in n, (67) converging for \n\ sufficiently small, and carrying one arbitrary constant shall omit the superfix and adopt the notation factor.
we
We
= o+?,M + ^M
+
,
(71)
is periodic and it Since the periodicity conditions have been satisfied, is Hence follows from its expansibility that each periodic. separately = 1 that it follows from this property and the initial condition (0)
(4*0,1, ...
00),
= l,2, (t
(72)
.
oo). j
be shown when the solutions are constructed that these properties uniquely define their coefficients. a V^T, that Case II. In this case we find from (68) f or /x = 0, a =
fa~
=0
'
10=^(2^)^(2^-^(2^)^(2^,
if?
Hence
if
=^
= aV
47r 3
J
^
2 2
= (-1)^2^(2^(2.)].
we
let a
+c n /JM+c02 M2 +
'
where
in general c n and c 02 are distinct from zero. In order not to multiply cases indefinitely we shall suppose that c n and c02 are distinct from zero and 2 2 c 02 16nK that the discriminant of the quadratic terms of (73), viz. 5 = c ll
l
is
distinct
from
zero.
(/3-M 03-M,
Then by the theory
as converging
,
where now
fe,?^6 2 since
is
5?^0.*
/3
of implicit functions
equation (73)
,3 1
solvable for
power
2
series of the
form
(74)
= 6 lM +#V+
solutions for
/3 2
=6
m+&V+
for a, as
/3,
and consequently
l
+aV
a
for
l
/x
= 0.
for
power series in There are two similar = n 0, but they do not lead
to distinct solutions since they differ from the former values by purely imaginary integers. Then, by means of (62) and (61), we obtain the final solutions as before.
for
example
b1
=b
all
of
which can be
they are
by the theory of implicit functions, but they will be omitted. *Since = 0, and 0= 01 are the roots of (63), it follows that in this case &, = &,, and that therefore distinct unless 6i = 6,=0.
83
for n
of
this case
we have
= 0,
instead
= a r+a
l
<Po=l,
^o
= T-
(75)
Hence equation
(68)
becomes
2tt+ 2 * X (2*V
X=l
=0.
x=i
(76)
0+2 X=l
<p x
(2w)\
l-e2a *+ 2
X=l
^(27t) m
As before, D can be expanded into a converging power series and for /z = the principal solutions of Z) = are a, = a 2 = 0. H, from (76) for p = a = that
in a
and
find
We
3D
da
,
d
0,
da
+8x
dD =
,
-2*rp
(2ir).
dfi
In general v (2t) is distinct from zero, and when it is we know from the theory of implicit functions that (76) can be solved for a in the form
1
=
1
+ ai VM + af M +
1
)
a2
J)
= 0-a\
ti
+ a?
ti+
(77)
are all zero. After a, and a 2 have been determined, we obtain the final solutions as before, except now the series proceed in powers of VJl instead of in powers of ii.
Since
a,
and
Case
I.
On
substituting
the
first
of (70)
and
zero,
(71) in (60)
powers of n to
we obtain
=0,
(78)
1)
+2av/^l
+2aV^T =
2
-2af [^+av^Ty-2ai
[j
+av^iy
the
left
members
and the
on
<*!
.
differ
first terms on the right There is a similar set of from these only in the sign
same except for the subscripts, being the same except for the superscripts equations defining the other solution, which
of
l.
Consider the solutions of (78) subject to the conditions general solution of the first equation is
(72).
The
where
2a
is
and b 2 are arbitrary constants of integration. not an integer, it follows from (72) that
6,
0)
0)
fc-1.
(79)
84
right function of t.
PERIODIC ORBITS.
The
member
When
is
of the second equation of (78) now becomes a known the left member of this equation is set equal to zero, its
general solution
^ = &" +6"
%
_
>
e-
2a /
_lr
.
(80)
as variables, according to the method of the regarding o"' and b 2aV ~ lT b e~ =0 variation of parameters, and imposing the conditions b^ be we of shall obtain and that the second equation satisfied, (78)
Now
^ = -['
-^]'
1
fti"-
81 )
u Let the constant part of 6 X be d v Then in order that b[ shall not contain a term proportional to t, which would make (80) non-periodic, we
2a
(82)
The integrals
of sines
sines respectively,
2
and
sin A
T e cos J
it
2oVr"lT j
ar
2aV
fia
(82)
f
e
2a
4a
cos sin
2aVZi T
Therefore
follows,
when
is satisfied,
that
bf-^W+Bf,
/
^~lT
&?-Q,(t)
+ P<,
(83)
where P, and Q t are periodic functions of r having the period 2x, and B and P4" are the constants of integration. Since 2a is not an integer,
4a2 can not vanish and there are no terms with infinite coefficients. On substituting (83) in (80) and imposing the conditions (72), we get
*,
(0)
(84)
It is easy to
show that
entirely similar.
,
The
succeeding steps of the integration are differential equation for the coefficient of /x* is
[|
4
,
P, (r) is an entirely known periodic function of r after the preceding have been taken. The general solution of the left member of this steps
where
equation set equal to zero is the same as (80), except that has the subscript The equations analogous to (81) are t, and 6, and b 2 have the superscript i.
6l
-~L ai
2a~J'
t
^- + [ a
F by
2a
'
a~J
If
we
we must impose
the condition
-^
be periodic. Then integrating, substituting in the equation analogous to (80), and imposing the conditions (72), we get
where
are periodic with the period 2ir. Thus the general step in the integration is in all essentials similar to the second step.
P,(t)
and Q
&-P,(r)+Q,]r)-P
<
(0)-Q,(0),
(r)
85
54. Direct Construction of the Solutions in Case II. Since in this case the solutions are also in general developable as power series in n, we start from equations (78). The general solution of the first equation is
Since 2a
is
an
integer,
it is
is
periodic for
all
values of of and b 2 \
initial
In this case
condition
(0)
instead of $(0)
= 1, whence
,(0)=0
o(0)
= l,
(t-i, ...co).
Hence we have
o=&r+^V ^,
2
(85)
where of
is
now becomes
1
0,e
_2aVzlT a ,,_2aVZiT
.
(86)
=+ V -
(1)
nUlOl
(l)
_2o\/^ix
(87)
'i
^T ^y=r-?+K^'-^[^+^>,].
In order that x shall be periodic the right members of these equations must contain no constant terms. Hence we must impose the conditions
[2a
V=T?+iJ &r+-o,
V^T
,
^? -^+^-0i
,
(88)
the constant part of 6 1} and where 5 2 and 5 22 are the constant T 2aV=lT 8 1 & aV If ^ is /2a respectively. /2a and parts of v^^T 61 e~ an even function of t, then 5 2l = 8 22 and if it is an odd function, 5 21 = 5 22
where
a\ is
Equations (88) express the conditions that the right member of (86) shall contain no terms independent of r, or which involve r only in the form -2 a v=T7This is only an expression for the fact that in order that the e solutions shall be periodic the right member of the differential equation must not contain terms of the type obtained by integrating the left member
set equal to zero.
Upon
of
we
get
V=l af+dj
[2a\/^T
af-dJ-40
5 21 S 22
= 0,
<=^yjs
The two values
of
2l
S 22
+^ V=T.
2
(89)
2
af
5 22
+d /4a = 0.
2
They
will
not
86
equations, and case treated in
PERIODIC ORBITS.
assume here that they are distinct. This was the 52, Case II, and when the two values of a[ are equal the After o"' has been solutions may be developable in a different form. once from either of at obtained equations (88). There is found, b can be = = a difficulty only if S 21 5 22 0, when one solution for b becomes infinite; but in this case we impose a different initial condition on After satisfying equation (88), the integrals of (87) have the form
we
shall
l)
bF-PM+B?,
1>
= Q (r)+
1
where P, (r) and Q, (t) are periodic functions of r, and P"' and P"' are undetermined constants of integration. These results substituted in equation
(80) give, after applying the condition
,(0)
=0,
[P, (0)
^%^() +
^
1)
( )
+0,()J
^
1
(90)
where P"'
It
is
so far undetermined.
is necessary to carry the integration one step further in order to that the general term satisfying the periodicity condition and the prove The differential equation for the coefficient initial condition can be found.
of
2
ju
is
m)
(r),
is
where
2)
ai
an entirely
known
periodic function of
r.
i)
l)
(),
!)
2,
where d2 and 8 2 are known constants depending on P2 The unknowns the and enter means of determinant of and, (92) linearly, (88), P'," by af 4a their coefficients becomes v^-T a', which, by hypothesis, is distinct from zero. Therefore a'," and B[ are uniquely determined by these equa.
,
l)
tions.
initial
equations (92) are satisfied, the solution of (91) satisfying the condition is
iT
a
When
i,
= B<?+P (r)+Q
a
(T)e-*'*=
-{QM+ ^T
2oV
T
)
+4]}e-
(93)
which has the same form as (90). Therefore the next step can be taken in the same manner. Thus it is seen that the process is unique after the choice of the sign of a"' is made, and in this way two solutions which satisfy the periodicity and initial conditions are obtained.
87
III.
Case
(1
= o+iVM+fcM+
= +a
V+a V+
<2
(94)
VJ1
u=o,
of
(72)
is
(95)
coefficient of
The
differential equation
which the
Vm must
satisfy
is
of this equation
i
which
satisfies the
conditions (72)
is
= 0,
2>
(l)
= an undetermined
which defines the
constant.
coefficient of
/x
(96)
is
The
differential equation
2
then
+2a<
= -2a m
^-(O'g,-^- -()-*,
When the left member of this equation is set equal to zero, its general solution
is
found to be
2)
6,-&r+*
(97)
where 6f and o 2 are the constants of integration. On making use of the variation of parameters and imposing the condition that the differential equation shall be satisfied, we find
2,
6i
-[(a
a,
)'+0,],
of =
+[(O
,
+0>]
r.
(98)
In order that, when the first of (98) is integrated and the result substituted 2 in (97), there shall be no term proportional to r the condition
(")+ d,-0
x
l
.
(99)
This equation determust be imposed, where d is the constant part of 6 mines two values of a which differ only in sign, and they are reals or pure
(1>
imaginaries when the coefficients of 6^ are real. Since there are no more arbitraries available in (98), no more conditions can be satisfied. The first equation of (98) gives rise to integrals of the type
n ~a
TdT jj cosJ
C sin
It.
J*.
~ -t *
7sinJ J
ai C0S 4* r-
The second
and associated
&] sin
-
integrals
ai T
sin
= F (r)-P
2
(0),
where
(t) is
periodic
and a
(2)
is
as yet undetermined.
88
PERIODIC ORBITS.
In determining the coefficient of
v
',
\j?
and &f
corre-
ir--.[2. >+^(T)],
a,
(T)] r,
<p 3
we
a) which uniquely determines a if a is distinct from zero, as it is in general. the expansion -may be of another form, for this is an exceptional case If a = in the existence discussion, and it is necessary to go to higher terms of the But differential equation to determine the character of the solution. a to case where is distinct from here the the solution ourselves zero, limiting
(I>
is
All the succeeding steps are the carried out as in the preceding step. same except for the indices and the numerical values of the coefficients.
III.
On transforming from
(4),
to
second equation of
of (50),
we
get
2
x+ia'+d, M +0 2 m
a
2
+
0,=
]x = 0,
Oil
2(2o,+o,) v 3Z
'
'
[0,
3
,, + (0,-0,)
>
cos 2t],
....
>
101 )
Cli
Obviously a will not in general be an integer. It will be shown that the only integral value it can have in the problem of the spherical pendulum is In this case the second of (101) gives unity. Suppose a equals the integer n.
)o 3
was shown in 43 that in the problem of the spherical pendulum a, is Therefore n- must be unity or zero. In the positive and a 3 is negative. = a 3 which, because of the inequalities satisfied by former case we have a, This is the special case of a I. and can be true a, only if a, = -f /, a 3 = 3 = = 2a 3 which, because of the the simple pendulum. If w we have 4a, which a are can not be satisfied. Therefore a to and 3 subject, inequalities a, is not an integer and the equations can be integrated by the methods of 53.
,
, ,
Upon omitting
a,
,
ou
the subscript on the a in e aT so as not to confuse o 3 defined in 43, it is found by actual computation that
,
it
with
_
?0
,.
+3V2 aiV^I
*
>
4V(o,-o,)
"
(2 a,
+a
3)
(1)
(a,
a 3)
4(a, + a
,/
\ (C0S2t v
1)
V/ 2V/ (a
^
1
a 3 )(2a,
,
'
+ a - V^-l Sin2r,
3)
,
3)
4(0,4-0,)
89
the sign of
The other solution is found from this one simply by changing V 1. Hence the general solution is
a:*=4e
aT
(1)
(r)+Be,
C2)
(r)
= fv ^V^at + a,
1
V ai -a V^^
___3v 2 a,
1
+a )^
3
V_i
J
*'
vWUi-aaH^ + as)
4(ai+a3 )
v^ sm 2r~L +
J
(102)
. . .
+ V2V(a
It
- a,)
4(,ai+a3 )
of
and constructing the solutions, that the coefficients of mj in and sines of even multiples of r, the highest multiple being 2j.
It follows
tions,
(1)
are cosines
and
2A=A
,
x
3^
=A
=1
[x 1 cos\t
'
x 2 sinXT]+^ 2
;c i
sinXr
+x
2
cosXr]
* ,
\ (cos2t-1) m + (cosines) m + 4^ + aJ
,
(103)
.
. .
^VK-aH^a^a,)^^^)^^^^ 4(a, + a
3)
are arbitrary constants. Since the second and third equations of (4) have the same form, the solution of the latter can differ from that of the former only in the constants
l
where
A and A
of integration.
?/
Therefore
=B
[a; 1
cosXT
The constants A
equations
(2),
the
first
are subject to the conditions that equation of (4), and the relation
,
and
xx+yy+zz=0
This leaves one arbitrary which may be used to dispose Let the axes be chosen so that of the orientation of the xy-axes at t = 0. = 0, we have from this equation = = t z at vanishes also Then since x 0. at r
shall
be
satisfied.
of
x and y above
V
[x 1
cosXT-a; 2 sinXr],
=B
[x 1
sinXr+z, cosXt].
(104)
90
PERIODIC ORBITS.
Making use
of (104),
it is
found from
(2)
2
and the
first of (4),
that at t = 0,
A'-r-n*
R2
21
(l+8)(2at +c
1)
Therefore, the solution is completely determined when the positive directions on the x and y-axes are chosen. The well-known properties of the motion can easily be derived from equations (104).
The
variables x
and y always
oscillate
around their
initial
values since
they are made up of terms which are the product of two periodic functions Since the period of x l and x2 is ir, the solutions are that are always finite. periodic and the curves described by the spherical pendulum are re-entrant
provided X
Let the period of x and x2 be and that of sinXr and cosXt be P2 =2ir/\. Then, when P and commensurable,
is
a rational number.
P = r, P are
1
Pg Pi
2 X
= 2j
V'
Hence the
least
integers.
common
multiple
P = pP = 2qP =
1
^
A
=2qir.
(105)
In the period
the variables
z,
oscillations,
In the independent and sinXr and cosXt make p complete oscillations. r the of and x is variable z, z, period 2 independent of n, but P2 is In the original independent variable t both a continuous function of ix.
,
But in either variable the period periods are continuous functions of t. is a discontinuous function of n, being finite only when the ratio of P, to
is
rational.
It is seen
which Pj
is constant with respect to fi, the linear continuum, and therefore that onby exceptionally
it
rational.
Simple Pendulum. Since the problem of the a simple pendulum special case under that of the spherical pendulum, it can be treated by the same methods. Of course, it is not advisable to do so in practice, for x and z must satisfy the relation
57. Application to the
is
+z
z
=l2
(106)
has been determined. Before discussing the properties of x we must find the expression for z in this case. Since in the simple pendulum z always passes through the
91
z'
it
follows that a 3 =
c2
I.
we
= c and =
z'
fact that
when
I,
Hence,
in
(50)
a,= +i,
z
a2
l+2lft,
=
2
I,
=-Z+J(l-cos2r) M +! Z(l-cos4r) M
t
;,
*-l,
...;
-]x
-|(H-2eo82T),
(107)
+ [l+|(l+2cos2T)M+|(5 + 16cos2r+4cos4r) M +
= 0.
In the expression for z the coefficient of each power of vanishes at r = and is a sum of cosines of even multiples of r. x
contains sin r as a factor.
relation m(1
2 2
it
separately Therefore
=r-z
The parameter
2
/x
is
also a factor.
From
the
cos2t)
=2
a*
sin
r, it
follows that
2
z=Vp-z
is
(108)
expansible as a power series in V/x, containing only odd powers of Vju. 2 is a sum of sines of odd It is easy to show that the coefficient of (V/I)
'" " 1
1
multiples of r, the highest multiple being 2t the second line of (107) and from (108) that
+ l.
We
V2
find directly
from
=
^|[2sinr]M
1/2
+|[-5sinT+3sin3r]M
last
(109)
It follows
the form
(110)
x,
and where
0.
It is
(110)
is
equation of (107) is linear, (110) multiplied by any power series in n having constant coefficients is also a solution. For example, we have for the determination of x and x3
t
+x = 0,
l
^+^--|(1-|-2c<2t)x
l
i ,,
x^Cisinr,
where
c,
x 3 = c 3 sinr+
c,
sm3r,
and
c 3 are
is
undetermined.
as the solution
92
PERIODIC ORBITS.
x at t = is an infinite series in vm whose general term is not easily obtained; but, from the fact that z(ir/2) = a 2 = l+2ln and from equation (108), we get
The value
of
x(|)
=2^Vn^ =
c,
2Zv^{l-},x-|M
we
find
(HI)
On
determining
and
c3
by these
m
v,
conditions,
x=
j[2"sinT]
| [-5
sin
r+ 3
sin3r] M
v*+
\>
agreeing with the direct computation (109). We may also consider the last equation of (107) from the standpoint of the general theory of linear differential equations having periodic coefficients.
From
/x
is
the fact that the part of the coefficient of x which is independent of unity, it follows that the solution of this problem belongs to Case II.
is
Since there
of a are
We independent of /x and which the two values of a not only differ by an imaginary integer for m = 0, but for all values of m- It follows from 21 that in this case the second
solution
is
periodic with the period 2w, the values are simply have here the case in 1.
is
*V
either r times a periodic function or, for special values of the In the problem coefficients of the differential equation, a periodic function.
of the simple
and
is
l,
i(0)=0.
If
we make
the transformation x
aT
,
(107)
becomes
2
+ ^(5+16cos2t+4cos4t) m +
We shall
]*
= 0.
shall be periodic integrate this equation and determine a so that with the period 2ir. The chief point of interest will be that a will be independent of n so far as the work is carried.
to (85)
and
(86) are
(113)
- (1+2 |
cos2r)
(2C+ v^e-'^).
93
The
conditions that the solution of the second of these equations shall be periodic are
-2V^fai 6'-|6-|v'^T = o,
1,
-a; '1
&>- fV=I0,
(114)
of
=
,
6<>
= -Iv^T.
(115)
Hence, we see by direct computation from the differential equations that in this problem a is independent of n up to \x at least
.
58. Application of the Integral Relations. now return to the consideration of the problem of the spherical pendulum. Since z and x have
We
been determined the value of y can be found from the first equation of (4). But it will be noticed that in this work no explicit use has been made of the Now x, y, and z must satisfy the differential equations, given integral (2).
in the last three equations of (4),
and the
integral relations
ttl
0-3
(HO)
xx+yy+zz=0,
xy-yx = c
where the second equation is determined from (2) by changing the independent variable from t to r; the third equation expresses the fact that the motion must be along the surface of the sphere; and the fourth equation is obtained from the second and third equations of (4), and expresses the fact that the projection of the areal velocity on the xy-plane is constant. The solutions of the second and third equations of (4) have been shown to have the form
x=a
,
aT
1
+a
e-
aT
Z2
,
y = b,
aT
^+b
e~
(117)
and 2 are power series in ju where a,, a 2 &,, and& 2 are arbitrary constants, and are periodic with the period t, and a is a pure imaginary which is also a power series in n, but is not an integer. Moreover, ^ and 2 are conjugate
,
complex quantities.
If
we make use
{a\+b\)
e*
becomes
+(a2 +b2 )
2
e-
,aT
+2( ai
a 2 +b b a )te> = V-*x
(U8)
Now
it is
has been shown that z is expansible as a power series in ju and that periodic with the periodic w. Suppose there is Before proceeding further we shall prove a lemma.
it
given
F(t)-i
aje *<pj(t)=0.
aj
Suppose the <pj(t) are not identically zero, that they are periodic with the period are equal or differ by an imaginary integer. 2tt, and that no two of the aj
94
PERIODIC ORBITS.
let e 2ajr
Then
from
zero.
=k
for
tl
that v>i(0
>
V. (O are distinct
jrfc+0r)-2| flt*
<
"ffi,G l )4$
=0,
F[t 1
It follows
+ (n
-l)2w]
=2
a J e'J, '<pJ (t
)kr'- =:0.
= a, =
or
1
,1
l
,
K2
n,_l
0.
ft,
/v 2
tin,
this
determinant
is
distinct
from
zero.
taking another t for which some of the remaining <p } do not vanish, we prove similarly that their coefficients are zero. Continuing thus we reach the conclusion that
a,j
On
=
we
2
(j'=i,
,*).
Upon
Therefore
oJ+ftJ-O,
6l
a 2 +b 2 =
v/^Ta
^iV^Toj,
or a l a2 -\-b l b2
(119)
= 2a
a2
l according as the same sign or the opposite signs are used in front of V in (119). It follows from (118) that in the former we have the trivial
Hence we shall take (119) with opposite signs. c, and c 2 of equations (7), upon which n depends, arise in energy integrals and are independent of the orientation of the x and y-axes.
case
Z.
z=
The
constants
take the axis so that y (0) = work up to this point, and from this it follows that
Consequently we
may
b^-b^V-la^V^l
As a consequence
of (119)
a2
(120)
and
becomes
(121)
4aJU = f-r.
95
Let us suppose that z has been computed and that we wish the coefficients of and 2 These quantities have the form
,
.
^-1
+K
1)
[a*'
1)
sin2T
+ V=i
1
V"
sin2ir] M
+
(122)
1}
&<" sin2r]
m+
>
>
- V-l
62
y)
sin2r
cos2r] M
=a
+[C+4n
+
n
[Co
+c
"
cos2r
c",'
+ +
>-,
where the
aj?
and
C+aJ"*
+<fr,
+#-'?
o=i,
. .
C+4+
The constant
we must
express (116) in
(123)
.oo).
and
ju,
and therefore
terms of the same parameters in order that it may be possible to compare these results with those obtained from the integrals.
We find from
(5), (7),
and the
2 l
first
of (10) that
=-a
x
(2a 1
+ a )-(a
3
-al)
fJi.
(124)
While the constant a is arbitrary in the solution of the differential equations, it must be subjected to the condition that the pendulum shall move on the sphere whose radius is I. This condition may very well make it a power series in n; hence it must be expressible in the form
In
fact,
we
a
t
find
from (121) at
a\
= 0, upon making
3
= 2
j=o
M '=
* | V-2a *
(ai+
a,)
[l+ u
VBM+
ai
ton
] -
(125)
If
we
we
form
F +F lM +F 2M +
2
=G +G lfi+G2lJ +
2
.
is
an identity
J
in n,
1
we have
(j=0,
F =G
By
hypothesis
It follows
...,).
fully
(127)
z has been determined; therefore the Gj are from (122) that the Fs and the G, have the form
known.
1)
1)
(128)
96
PERIODIC ORBITS.
where the
B^
/ are
known
constants.
identities in t,
we have
AJf-SJ?
(*-,
(129)
On
we
W+
When we
[**!+ life
1
"
+[<+%]
1 i
[a^ 2 -^] e2
"-2(a
(1
a 2 +6 6 2 )
1
[a
tf,
+ 6)
a3
^ ai
[-g + a. + a. + a,].
we obtain
1
4a;[a(f,f i
-g
,)
+ f,^l = a
(f-2 )-i
+ 8) [-g+a + a,+a,]. + 4m ai a3
we
(130)
substituting (122) and the series for z in this equation, expression of the form
Now, on
get an
H +H +H S+
1
=K +K L+K
lfJ
2 2
+
(131)
From
is
an identity
i )
in n, it follows that
H =K
The Kj
(a
(0)
are
known except
and the
The constant K
involves
= l, {j
On
referring to (122),
we
2
oo) involve the a linearly. the K, have the form see that the and t
#,= K, =
(T+C "cos2r+
+C ?cos2 Y,
2
<7
D+D
cos2r+
r, it
+0*?
= 0,
cos2ir.
follows that
(t
. .
.
C%-D&
It will
(133)
now be shown
y
uniquely the a 2\
of j
when
and
that equations (123), (129), and (133) determine U) &",' (j>0), in the order of increasing values a[ \ a are known. To do this it is necessary to develop the
J
,
explicit
It is necessary to eliminate a 2
forms of (129) and (133) by reference to equations (121) and (130). 2 and I from their right members by equations
first
0)
(124)
and the
of (10).
3
(121)
and
(130)
4(a[
=-2a
(a 1
2
+a
3 ),
0=-(a
The first of
(0,
[2a 3 (a
+ a )]- 4
3
ci3
(2 ai
(134)
ai
The sign of aj these equations determines a except as to sign. The second which end of the -axis. is taken as the depends upon positive
0)
equation gives
(0)
(a
= -a = 2
2(2ai+(fa)
,
(135)
ai
cu
97
When ./=1, we
find
= 8(ar) < +8ar<'=-(a^-aD-2a = B: = ii" = 8(aj )*a+0 0-2a,4 = ^ " = - - 4 a a (a, + a,) a - (a<>) ~S) "2 (*
,,
i)
,
1)
,)
,
C
^
(0 >
<0>
a 3 C?'
(136)
1]
+4K-2(a + a
1 1)
2 3)
]
+ a )3
3
'
=/)<",
)
)<",
The unknowns equation of (123) for j= 1. in these five equations are a"\ a?\ af\ a m , and b 2 which enter linearly. The second equation determines a'; then a is found from the first of (123) then the first of (136) defines a"', while a u> and b " are given by the last two
to
first
l)
, ;
1
equations of (136). We shall apply (136) in computing the We find from (49) and (50) that
first
V
Upon
,
2'
(137)
substituting in (136) and solving these equations l) a) a and 6 in order, we get (123) for a'", aj", a[
,
and the
first
of
,(i)
~~~~
(oi
eu)
4U.+0,)'
03)(2ai+os)
<=
w b = u
2
(ai
a3 )
V^+a,
3
V-2a
4(o,
.(i)
+3 V2ai V-l
4 V(oi
(138)
V2(dr-cto)"(2"ar+fa")
+a
3)
agreeing exactly with the results obtained in (102). For a general value of j the equations corresponding to (136) arc
iC = 8(a)*a<' +8a
>
a;'
+Z"
3)
>!
"0
A$=8(a>ya$+0
= C^= -4a - 16 a C<? = = a<' +a" +
2
+25? --B
(-l,
,
(0)
a 3 ( ai
(0)
>)
(139)
(t-1,
The unknowns
a\
k)
,
a[ \ a[
U)
,
(fc-0,
. .
j-1)
The have been determined, are a h% (t-1, J). a[ \ oj?, a known are quantities depending upon ,/) A%, CJ?, and D (t-1, the coefficients having smaller numbers for the superscripts. The j+2 equations of the first two and the last lines define uniquely the j+2 quantities The equation of the third line determines and a^t (i = 0, ,j). ,j). a U) and the equations of the fourth line, the coefficients b^ (t-1,
J)
}
.
98
Therefore
PERIODIC ORBITS.
the interesting result that in this problem the coefficients of the general solution can all be determined from the integral relations alone, the solution of the z-equation having been previously obtained from
we have
another integral in 49. The last two equations of (116) are unused integrals. Let us consider the last equation, which is the more complicated. By means of (117), we get
4 a\ a
V=l
& fc- 2 a
v^T fe k ~L Q =c
2
2
Upon
2a 2
1
(S 1 i 2
-k
i; 2 )
= a(l -z
series in
+ V-^ic
(140)
coefficients.
The constant
c 3 will
be a power
ju
power
series in
L lM +
where the
=M +ilf lM +M M +
.
(141)
involve linearly in the terms independent of r the unknown Since this equation is an identity in /u, coefficients of the expansion of c 3
t
we have
Lj
It follows
= Mj
(j=0,
...).
= #^ +#" cos2r+
,
+E% toaZfr,
+F$oob2jt,
J)
J)
from which
it
follows that
X?-n?
The
(t-0,
,j;j=0, ...
oo).
(142)
and the
F " (i= 1,
2
.
,
(
j) are
known
n involves the unknown coefficient of n in already computed, while the F the expansion of c3 Consequently equations (142) determine this constant for I = 0, and also furnish a check on the earlier computation of the
coefficients for
i=l,
j.
Chapter
PERIODIC ORBITS
IV.
59. Introduction.
The
an oblate spheroid
is
The motion of the particle is not, not, in general, closed geometrically. in from a therefore, general, periodic geometric point of view. But if we consider the orbit as described by the particle in a revolving meridian plane
which passes constantly through the particle, several classes of closed orbits can be found in which the motion is periodic. The failure of these orbits to close in space arises from the incommensurability of the period of rotation of the line of nodes with the period of motion in the revolving plane. When these periods happen to be commensurable the orbits are closed in space and the motion is therefore periodic, though the period may be very great. Indeed, it seems that much of the difficulty in giving mathematical expressions to the orbits about an oblate spheroid rests upon the incommensuraThe difficulty arising from the node can be overcome in bility of periods. the manner just described, but elsewhere it is more troublesome. Orbits closed in the revolving plane are considered most conveniently in two general classes: I, Those which re-enter after one revolution; II, those which re-enter after mam revolutions. The existence of both classes is established in this chapter and convenient methods for constructing the Orbits which re-enter after the first revolution are solutions are given. naturally the simpler and will be considered in the first part of the chapter. Those lying in the equatorial plane of the spheroid become straight lines in the revolving plane, and within the realm of convergence of the series employed all orbits in the equatorial plane are periodic. When the motion is not in the equatorial plane there exists one, and only one, orbit for assigned values of the inclination and the mean distance. These orbits reduce
r
to circles with the vanishing of the oblateness of the spheroid. In considering orbits which re-enter only after many revolutions the
differential equations are found to be of ever finding any of these orbits
proof of their existence and a method for the constructions of the solutions are given by the aid of theorems on the character of the solutions of nonhomogeneous linear differential equations with periodic coefficients.
100
PERIODIC ORBITS.
Those periodic orbits of many revolutions involve five arbitrary conOne, only, is lacking for a complete integration of the differential equations. The orbits are all symmetric with respect to the equatorial plane.
stants.
60.
The
Differential Equations.
The
differential equations of
motion
of a particle about
d?x
8 k'Mxr, ha 3,22 M x
-4z +^-42 +y
2
s i
,
.
-i
dV
dx
dF
R*
k2
'
3
L
1 1
'
io"
_..
,
e*
M ^
2
J
2 -
(Py__
<fc
2
My r
L
x*+y -lz
2
"*"
'
fl
io io"
ft* R*
i2
_ dV
-
-J A
"i
dy'
(1)
fz
k2 Mzr,,
3,2
3(:c
+?/
)-2z
dF
The symbols employed are defined as follows: The x, y, z are rectangular coordinates, the
of the spheroid
is
is is
p-
2i
k'Mf,
6V s +?/
2
-2z
2 -i
,
Since -
x dx
-r
-y by
we obtain one
dy
dx
integral of areas,
namely J
/flN
*Tt-*Tt-*'
That is, the projection
equatorial plane
integral
is
(2)
of the area described bj' the radius vector upon the have also the vis viva proportional to the time.
We
(3)
There are no other integrals which can be expressed in a finite number of terms, and for further integration we are compelled to resort to the use of
infinite series.
It will
differential equations to
cylindrical coordinates
by
v,
the substitutions
x = ar cos
k
2
y = ar sin
v,
M = n*a
3
,
c^cky/Ma,
= aq, nt = r,
z
(4)
R = a^^+f,
|j-<.
to Celestial
*Moulton's Introduction
Mechanics,
p. 113.
101
become
(b)
rv'+2r'v'
n" q
=
,
(5)
(r)
- ~q ~ (r +0*
2
3r*q-2q\ 2
r (^+g*)
lM
.,
where the accents denote derivatives with respect to r. The integral of (6) is r2 v' = c, by means of which v' .can be eliminated from equation (a), and the equations then take the form
w
(a)
(h) (6)
r"- & 3
r
r
(r
2
""
2 3
r'-lrq*
(r'
+?
q
+g
ft2
lM
2
7
.
'
'
'
'
a"?
3r*q-2q*
Jfr-4-tfV
^.
^+
2
'
(6)
w
The first two
q
of these equations are independent of the third, so that r and considered as being rectangular coordinates in a revolving plane
may be which passes always through the polar axis of the spheroid and through the The problem is thus reduced to the consideration of the particle itself. motion in this plane, for, when r is known, v is obtained from the last equation by a simple quadrature.
61. Surfaces of
Zero Velocity.
1
The
plane
is
r' -\-n'
2 2
202 -2g
2,
2
,
__
,~
the velocity equal to zero, the resulting equation represents a two-parameter family of curves. For assigned values of the parameters c and
If
we put
is
On one side of this defined a curve in the revolving plane. curve the motion is real and on the other side it is imaginary. For values of
c2
,
there
c2
<0,
this curve
is
closed
is
real
on the
inside.
As the plane
revolves this curve generates a surface of the general form of an anchor ring. 2 = 0, this curve belongs to the ordinary two-body problem and For
(j.
the motion
is elliptic,
is
negative,
zero, or positive.
Its
equation
?
is
-4-c =
q
On
we
putting
r
= p cos
p,
<p,
= p sin
<p,
find,
by
solving for
that
4[-W 1+ ^]-
102
PERIODIC ORBITS.
c 2 this
2 If c c 2
equation represents two closed ovals which do = 1 the oval shrinks upon the points
orbit
is
p=
1/Cj
<p
and
n.
The corresponding
c2
As
and
~
2 cos2
ip
'
For values of
c2
>0, there
is
for p,
which
is
hi^+f+M]
If c 5^0,
2
none
<p
= ir/2.
the circle p = 2/c2 inside of which the motion is 2 For values of m ?*0, but sufficiently small, we can put
r=(p+p)
and solve
for p as a
coscp,
q= (p+p)
2
p.
.
power
series in
l
We
a2
2
find in this
,
manner
3 cosV p(l+C2 p)
which
is
two-body problem.
I.
62.
(6)
Symmetry.
On
of (6),
we
r
into+r,
q into
q,
rinto
Hence,
q'
if
t,
at
some epoch
=t
=a
r'
= 0,
= 0,
= 0,
that
it
at the epoch t = t the particle crosses the r-axis perpendicularly, follows from the form of the differential equations that the orbit is symis, if
,
.
metrical with respect to the r-axis and with respect to the epoch t = t In other words, r is an even function of t r n and q is an odd function of = t r If now at some other epoch, r t T, the particle again crosses the
, .
r-axis perpendicularly, the orbit is symmetrical with respect to this epoch also. It is clear, therefore, that the orbit is a closed one, and that the motion in
T it must have been at the is periodic, for, at r = t -f T and at t = t q same point and moving with the same velocity in the same direction. Hence
it
)=q(r )=0,
v'
From
= c/rs
it
follows that
if
r is
periodic
t'
will
have
the form v = A (r
is
a constant.
103
case where q
(a)
In the
equations
(6)
reduce to
1
y2
r"-
&
n*3
^
m4
gaV
6
|
(8)
(*)
'-.2
first
of these equations is independent of the second and can be integrated separately. It represents motion in a straight line in the revolving plane. It admits the constant solution
The
r-v-1,
which represents a point
plane.
= c = l-H
2
+0
us put
= l+pe,
initial
=c
+ee,
value can be arbitrarily assigned, e is a parameter corresponding to the eccentricity in the two-body problem, and e is a parameter to be determined so that p shall be periodic.
p
is
where
a variable whose
On substituting these values in (8a) and expanding as power series in e, the terms independent of e cancel out, and it is possible to divide through
by
e.
p"+[l-0? M
-3^p +
.] e ]p=+[l-3pe+6 PV-10pV+ + [3-60?m -150 +---]p e + [-6+lO0?p +46^ M +---] PV + [+10-20e M -lll^p4 + PV
. .
(9)
We
can simplify this equation somewhat by dividing through by the cient of p in the left member and then substituting
coeffi-
T = rVl-^M -3^p^+
2
5=
1-0V-30 V +
2
J^+P
where
at
= [l-3pe+6pV-10pV+
2
]8+[3+a
3]
a
l
]p
(10)
V+
a2
==+
30 p
-(
30
40 p
2
a,=
- 1O0
+( 40 - (1O0
+ +280 )m + +810 M +
60 )m
2
2
104
PERIODIC ORBITS.
Equation
initial
(10) can
be integrated as a power
series in 5
and
with the
values
p=-l,
By
p'
= 0.
Poincare's extension of Cauchy's theorem, 14-16, this solution cone sufficiently small, and for all values of T in the verges for values of 8 and
interval
^T
<:
T,
where
is finite,
The
p
If
= 0atT = 2\
(11)
we choose T = t, an
is
solution for p
the inspection of equation (10) shows that for e = periodic with the period 2ir, whatever may be the value of 8.
e as
a factor.
is,
After integrating
we
x
explicitly,
-+a
Upon
\Trbe- [-
+ - a, +
a*+
a2
7re
+ higher degree
terms.
(12)
linear terms in 8
dividing out the factor e, there remains an equation in which the and e are present, and this equation can be solved for 5 as
e.
a power series in
5
We
find
2 2
=
8
[-l+20
+(f^-0
in
) M <+
-]e+
(13)
equation (10), it will then admit periodic solutions for p having the period 2ir for all values of e sufficiently small. Furthermore the solution as a power series in e is unique.
If this
value of
be substituted
Plane.
2 For p =
which are Inclined to the Equatorial the differential equations (6) admit the circular solution
c
2
=l,
=l,
= t,
<2
= 0.
(14)
In order to investigate the existence of orbits not lying in the equatorial 2 plane, but having the period 2t for p ?^0, let us put
r=l+p,
and take the
initial
= + c,
= 0,
=l+
(15)
conditions
= a,
p'
= 0,
(7
a'
j3iJL.
The
=Q
at T
= 7T.
,
We have three arbitrary constants at our disposal, a, 0, and and two conditions to be satisfied. will therefore let /3 remain arbitrary and determine a and e so as to satisfy the two conditions.
We
105
After making the substitutions (15) and expanding, equations (6) become
(a)
p"
+p =
+ 6p e-6p
2
-6p<r
+4p0,
(6)
ju
-f-
(16)
a"
+a = 3p a
6p
<x
?a
6\
let
us put
p= 2
The
put
p iM
e'ay,
a= 2
d*V.
(17)
ot can be found by successive integrations, the constants of In the integration being determined so as to satisfy the initial conditions. = series thus obtained put t t. The two conditions for periodicity give
or
and
(it)
= =a
+a
[6[ 6
2e
+ a + a ea + a
3 3e t
2
5e
+a
+a
8
2
3 2
+a ep
7
aM
+a
p
],
(18)
(6)
(ir)
=0 =
/3
62
&3 a
+6
ea-|-6 5/u
Equation (18a) involves only the even powers of p, while (186) involves only the odd powers. After dividing (186) by /3p, we can solve it fore as a 2 power series in a and p of the form
=C
+c
+c
+C
ajLr
+C
+
form
''
(19)
On
(a)
we obtain a
3
series of the
= d ap +d
2
l
+d
+d
i a 2 p2 +di a
'
'
'
20 )
If in this
equation we
make
the substitution
(6)
-(*-$)*
obtain
we
(c)
o=p
[/iT
+/
+/,7m +/47 p
+
.
I;
This solution y as a power series in p 2 This value of a subsubstituted in (206) gives a as a power series in p. 2 We thus have a solution stituted in (19) gives e as a power series in p.
for
=p
P, (p
=p P
2
.
(p
- arbitrary,
2 Newton's parallelogram shows that are power series in p equation (20a) has two additional solutions, but as they are imaginary we
where
and
we
shall not
develop them.
106
PERIODIC ORBITS.
65. Existence of Orbits in a
Meridian Plane. If in equations (6) we put the area constant c equal to zero, the motion of the particle is in a meridian plane; that is, the plane has ceased to revolve, and the orbit in this plane is the true orbit. After changing to polar coordinates by the substitution
r
=p
cos<p,
= psm<p,
v"-vWY+
e\
+
(21)
p?"+2pV =
For p2 = 0, equations
- sin2<p
2i
7 4
p'
sin
4^
^M +
2
(21)
p=
that
is,
= r;
= T + a,
<r'
circle.
For p2 7^0, we
values
p
will
put
<p
p = l+p,
with the
initial
= a,
p'
= 0,
<r
= 0,
j3,
/3 are two new arbitraries. By 14-16, p, p', a, and <r' are 2 series in with as and r entering the coefficients. p power expansible a, /3, = The conditions for periodicity are that at t x
where a and
p'
= = 0.
<r
integrate equations (21) and then put r periodicity conditions two equations of the form
If
(a)
(6)
a-
we
x,
we obtain from
4
the
(T)
= = a a+a 8+a
1
2J
a2 +a4 a/3+a 6
a 2 +& 4
/3
2
p'(tt)=0=
+&
a/3+&
5 /3
+ ap + +0.p +6 p +
+a
6
(22)
The
and p2
of the
(c)
.
first of equations (22) can be solved for a as a power series in /3 This expression for a substituted in (6) gives rise to an equation
form
O=c
1
0p +c
3
2 /3
+c,/3 p
+c
+
.
This equation has the same form as (20) and can be solved in the same 2 2 This way, giving a solution for /S as a power series in p vanishing with p in a for the for a for substituted value fi expression equation gives unique 2 2 a as a power series in p vanishing with p Therefore periodic orbits exist for pVO, which are analytic continuations of circular orbits for p = 0. We have thus proved the existence of the following three classes of periodic orbits which have the period 27r:
,
Orbits lying in the equatorial plane whose generating orbit is a circle. II. Orbits inclined to the equatorial plane whose generating orbit is a circle. III. Orbits in a meridian plane whose generating orbit is a circle.
I.
107
consider
66. Construction of Periodic Solutions in the Equatorial Plane. first orbits in the equatorial plane. take the differential equa-
Wc
We
tions (8), and by means of the transformations there given we proceed at once to the integration of equation (10). It was shown in equation (13) that 5 can be expanded uniquely as a power series in e in such a manner that the solution for p as a power series in e shall be periodic with the Since the series is periodic with the same period for all values period 2t. of e sufficiently small, it follows that the coefficient of each power of e is Since the solution exists and is unique, it must be possible itself periodic. to determine the 5 uniquely by the condition that the solution shall be In the existence proof it was shown that 8 vanishes with e. periodic. Therefore p and 5 have the form
P
= P +Pie+p
values
+p
= S e+5
1
+5
e -f
(23)
The
pj
are to be determined
by the
by the
(24)
initial
P(0)
= -1,
^=0.
The
8j
Upon
(a)
manner that the p, shall be periodic. and equating the coefficients, we find
ffS+Pb-0,
"Pi -j||j+P
I
(*>)
=
1
S
,
-]pI,
4
(c)
|p+p
tfpn
= 6 -3p
2
51
(25)
(d)
j*
+p = 5 ,-3p,A_ + [3-30?M*-(fl+6
l
2p.p.-
+/(Po,
These equations can be integrated
satisfies the initial conditions is
Po
Pn-^)y
in succession.
The
=-cosT.
e,
(26)
Since the
initial
every
p_,
except p must
vanish at T
p
= 0.
On
we have
(27)
-][|-|cosT-|cos2T]-
The constants
as yet, undetermined.
108
PERIODIC ORBITS.
On
p,
in (25c),
2
we
find
fp+P*= +[m-*i(3-3; m
2
1
-)
+ (3-60
2
)]
V
2
)
)/x<
.)]cost
)
+ [5
(3-30 M
2
/x
2
(3-180^*
]cos2T
+ [~3-40
of cos T
-]cos3T.
In order that the solution of this equation shall be periodic the coefficient must be zero. This is the condition that determines 5, and con,
sequently
51
=-1+2^ m +(|^-^)m +
2
4
With
this value of
5j
equa-
f^+P
The
= [5 +30
2
-]
+ [-90
2
2
]cos2T+[3-40 p
2
. .
.]
C os3t.
is
+ [30
+ [30
The constant
.]+[|_|^ M + ]cos2T+[-f + |0 +
. .
.]
cosT
I
(29)
/x
-]cos3T.
6 2 is as yet entirely arbitrary. It is determined by the in condition on the same manner that 6 S was determined by the p3 periodicity Without giving the details of the computation, periodicity condition on p2 its value is found to be
.
S2
=-60
,
.
+
,
This method of integration can be carried as far as is desired. In order to show this, let us suppose that p p_i have been computed and that all the constants are known except 8 n - t From (25 d) we have
. . .
ff+Pn=5-3p
where /(p
terms.
It
, . . . ,
5n _ 1
+[3-30
2
/x
.]**_+/.(*
x
p-
2 ),
(30)
is
a polynomial in the p } and contains only known easy to see that p_i depends upon b n _ in the following way,
p_ 2 )
is
p_i = 5_,(l
cosT)+ known
terms.
Equation (30)
n
may
n
therefore be written
2
|^ +P=5 +[3-3
+ [3-30
109
In order that the solution of this equation shall be periodic the coefficient of cost must be zero. This condition determines 5_,. The equation can then be integrated, and the constants of integration will be determined by
the conditions that, at T
= 0,
Everything
is
p=(l
On
p
cosT)5+known
81
terms.
and
8 2 in
it
has
= -cost,
2
4
-
Pi=[|+^iM +(^-4^) M
-]
+ [-^m +(-^+3^)m
2
-JcosT
4
-]cos2T,
=[-3^ M +
2 2
+ [f-|^M +
2
4
-
-]cosT
[30
2
-]cos2T
---]cos3T,
+ [-| + ^
51
M
2
=-1+2^ m +(|^-^)m
r=l-ecosT +
2
52
= O-60 M
2
From
(a)
'
+ [-0
+(-|0 + 30 )m +
4 2
4
-]cosT
e2
USD
3
-]
+ [3^ M
On
-]cos2T
+ [-f +}0
-]cos3Tje
substituting this value of r in the equation (86), transforming to the independent variable T, and integrating, we find
(b)
v-v
=+{[i+eis+(\e\+2(%)s+
]
2
) MH-
-]e
-}t
-JsinTJ-c
+[[2
+(j0}-6<)m +
4
-]sinT
'
+ [f + f^+(~ *+!)**+
-]sin2Tp
110
PERIODIC ORBITS.
Equations (31a) and (316) are the periodic solutions sought. If we return to the symbols defined in the original differential equations (1) by means of equations (5), with the additional notation
n a/I-0V-30^m4
'
= ">
we have the
/?
a|l-ecos
+ [|-|cos2j/f]e
+[|cos^<-|cos3j'f]e
+
1(32)
e2
+i) cos3j'0
e3
]*+
}>
<>
= vt+2 sinj^-e+Tsh^pf .e
+
(33)
+ (fsin^+|sin2^) e +
2
]m
+^[-
-]m
+
, -
Equations (32) and (33) contain four arbitrary constants,* a, e, v and J Since the differential equations of motion in the equatorial plane are of the fourth order, these series, within the realm of their convergence, represent
The expression for the radius vector, R, is always the general solution. At the expiration of this period v has with the periodic period 2-k/v. increased by the quantity
2
in excess of
*||*(5+S*+
)+
-1=2.9
(34)
2x; that is, the line of apsides has rotated forward by this 9 is commensurable with unity the orbit is eventually closed If 9 = I /J where / and J are relatively prime integers, geometrically. then v = 2(I + J)ir at t = 2Jw/v, and the particle is at its initial position
amount.
If
,
components of velocity. The particle has completed and the line of apsides has completed / revolutions. revolutions,
with
its initial
I+J
The
(35)
mean
sidereal period
is
P
Equation
=VWTW
(34) for the rotation of the line of apsides has an interesting in the case of Jupiter's fifth satellite. On the hypothesis that application
Jupiter is a homogeneous spheroid whose equatorial diameter is 90,190 miles and whose polar diameter is 84,570 miles, that the mean distance of the satellite
is
its
orbit
is .006,
t
and that
*The constant a
replaced by
(t
tt)
since
also contained implicitly in v through the constant n, and does not occur explicitly in the differential equations (1).
can obviously be
Ill
sidereal period is ll 22! 7, equation (34) gives for the rotation of the line of apsides 1440 per year. The values derived from observations are somewhat discordant, but are in the neighborhood of 883 per year. If we
still keep the hypothesis that Jupiter is homogeneous in density and of the same oblateness as before, we are compelled to suppose that the value adopted for its polar radius was about 9,000 miles too great. In reality Jupiter must be much more dense at the center than at its surface, and therefore it is not
its
Construction of Periodic Solutions for Orbits Inclined to the Equatorial Plane. By means of the area integral the problem has been
(6),
the
first
two
T
0,
of
which are
,
=3 r
(r
T +? )5
2
tf+q )*
2
(6)
tf-
L-<=2^/....
(f+q )*
2
(r
+g )*
After the solution of these equations has been obtained, the third coordinate is found from the equation
t>
ic)
*~ Str
2
We
= 1
+C l +p
p p
=g M+?
+C +p +?
+ p + M +
4 4 5
(36)
with the
initial
conditions
r'(0)=g(0)=0,
the constant
13
g'(O)=0M,
can therefore integrate the equations and determine the c s in such a manner
being arbitrary.
We
[p 2'+p2
|?i-c
9l
<z 3
+6p
q\
(37)
+(3c -40 )p
2 2
-Y^2
*y+
* '
+6 plq-3p&+
2
1
'lpi
ql+ {ql+^e
q3 -15e 1 p2
q- jf qi]^+
1
(38)
-
Equation (37) contains only the even powers odd powers. For the integration we have:
of
p,
and
(38)
only the
112
Coefficient of p.
PERIODIC ORBITS.
The
coefficient of
is
defined
ft-pTsinr.
Coefficient of
(39)
(37), is defined
2
.
The
coefficient of
2
,
from
2
by
P:.+P2
f?:+^-^=(f/3 +c -^)-f^cos2r.
2
The
is
p2
=(i 3 +c
/
-0j) +
a2
COST +
j0
cos2r.
The constant
determined by the periodicity condition on q3 where it 2 and c^, which is deteris found that it must have the value c2 = 20, /3 mined by the periodicity condition on p 4 is found to be zero. If we anticipate these determinations, we have
c2 is
,
p2 =(0
3
.
-/3 )+i/J
2
3
,
cos2r.
(40)
Coefficient of
The
coefficient of p
from
(38), is defined
by
a2 /3sin2r.
it is
integrating,
we
find
g,=/3, sin r
703
/3
sin2T.
From
therefore
&=fa
But
it
will
a,
= 0, and
consequently that
(41)
q^O.
Coefficient of
4
.
It follows
from
(37)
that the
coefficient
of
is
defined
by
2 2 2 p;+P = 3p +3(/ l?3 -6p29 -ff/l +(40 -3c )p
,
2
2
<
+f0
+c
(42)
113
member of this equation we will examine the which we know must be zero from the periodicity condition. It is noticed in the first place that terms in cost can arise only through terms involving p2 and g3 and secondly that all such terms
of cost,
,
No other arbitrary enters the coefficient; therefore we carry a2 as a factor. must take 03 = 0. It can be shown by induction that the arbitrary constant a (the coefficient of cost), which arises in the integration of p, is
(
determined by the periodicity condition on pi+2 and further that its value is zero. The proof is omitted for the sake of brevity. Upon substituting the value a2 = in p2 and q3 and expanding the right
,
member
of (42),
we
find
Pr+P4=[c
os2T+g^cos4T.
is
P 1 =[c 4
+^ + ^^
cos4t.
If
we
is
P,=
[-3^-^^
-^
+ [-^^ +
^ ]cos2t-^^cos4t.
1
5
Coefficient of p\
p.
is
defined
by
= [3c +120?+H0
1
2 /3
3 ]i3sinT-0? 8 sin3T.
/
From
we have
l
Ci
=-id\-
e\fi\
On
we
?6 =-f0i/3
smT+fO
sin3T.
(43)
This
is sufficient
to
make
evident the general character of the series. r and the g-equation conThe r-equation contains only even powers of p
odd
in
The
series
and odd.
114
PERIODIC ORBITS.
On
(a)
we have
+
(6)
(-^ +^/3
2 s
4 2
cos2t
-^ cos4t]
2
+
(44)
= =
[^sinT]M+[0]M
2
+[-f^^sinr+^^sin3r] M +
5 2
(c)
-*;o
[l-^M -(f0
(ii
(d)
=1+
[20 -/?] m
+[-40
-^
2
2
/3
]m
In this solution the constants* a, j8, v0) and t are arbitrary. As is shown by equation (44c) the nodes regress, the measure of regression being 27
r[0
2
+(f0 +i0
/3
y+
is
The generating
plane.
r
circle in
the equatorial
A circle having any assigned inclination might have been used, e. g.,
= Vl
s
2
sin2 T,
= ssinT,
.
= tan -1 [Vl
tanr],
(45)
where s is the sine of the inclination. The solution thus obtained would have been identical with (44) If we should expand (45) as power series in s and put s = /3/z, we should find that the terms thus obtained are identical with the terms independent of 0, in the solution which has been worked out. It might therefore be of assistance in the physical interpretation of the
constants to put
/3/x
=s
Meridian Plane.
When the
of
constant
c is zero
the motion
is
in a
meridian plane.
The equations
motion
(21) are
,
p-p(<P
y+j-
+ |cos2^+7Cos4^ -| 2 4 4 _
pi
0*
+
(46)
-sin2<p
-sin4<?
p<p"+2 P '(<py
-;-i-
el
We
have proved in 65 the existence of periodic solutions of these equations as power series in m 2 which, for m2 = 0, reduce to the circle p = 1, <p = r. Let us therefore put
,
p=
l+p
+p
>
= r+V
J2
+ ?4M +
4
'
'
'
*Thc constant o
is
0];
115
Upon
[?>2-3p2 -2^-f^+f^cos2T
+ j^cos4r]M
+ [pr-3A-2^-2p
=
+ (-3sin2r-sin4r)*>
[^'
2
0f]M
(47)
+ \<pl+2p[+<pl p
+2p' 2
2
<p' 2
+(-2 sin2r+
sin4r)
p2
+ (cos2t-cos4t)
The initial we have:
<p 2
]m
+
On
proceeding to the integration,
Coefficients of
/x
The
2
by
(a)
p2
cos4r,
(48)
2
(b)
<P 2
+2p' =
2
-}0 sin2r+i0
sin4r.
On
(c)
integrating
(b)
once,
we have
2
2
^=-2p + j0
we
substitute this value of
<p 2
cos2t-^0 cos4t + c
If
becomes
2
(d)
P :+p 2 =(2c
+ f0 )-0
cos2r-
cos4r.
The
p2 =
Since p 2 (0)
(c)
(2^+f )+c
=
sin r
+c
COS
r+\ti[
COS 2r
+ ^0
COS 4r.
c2
= C.
On
the initial conditions, <p2 must be zero when r = 0; therefore c4 = 0. 2 All of the constants of It must also be periodic; therefore 2c t = c which will be determined by the integration are now determined except 3
From
periodicity condition on p t
116
PERIODIC ORBITS.
The
p t and
<pA
same
as for
p 2 and
<p 2
except in the right members. The process of integration is j ust the same. In the right members only even multiples of t occur except in terms carrying the undetermined constant c3 as a factor. In the equation corresponding to (48d)
But the integral from this there will be a term in cost carrying c3 as a factor. = term will be non-periodic unless c3 0. Upon putting c3 = 0, the integration proceeds just as before and the constants are determined in the same manner.
The same
powers.
therefore,
steps are repeated in the coefficients of m\ and so on for all higher Therefore no odd multiples of r occur in the solution. We have,
p=
+ [-i + |cos2r+^cos4r]^ M +
2 2
^ = r + [-^sin2r-|5
sin4r]^M
t,
(49)
Since the series involve only even multiples of with respect to both the r-axis and the g-axis.
existence
This completes the formal construction of the solutions of which the was proved in 63, 64, and 65.
II.
MANY
REVOLUTIONS.
69.
The
The
orbits
considered have had the common property of involving only the period 2x. Since this period is independent of the oblateness of the spheroid, the derivation of these orbits has been relatively simple. We shall proceed now to
investigate a class of orbits which involves, beside the period 2w, another period 2t/\, where X is a function of the oblateness of the spheroid, the
We
inclination of the orbit to the equator, and the mean distance of the particle. will start out from the solution which involved an arbitrary inclination
Into this solution four arbitrary constants were introduced, viz., Two more inclination, mean distance, longitude of node, and the epoch. arbitraries are necessary for a complete solution, viz., constants correspondIn what follows ing to the eccentricity and to the longitude of perihelion.
(67).
we
shall introduce the constant corresponding to eccentricity. have found for the differential equations a certain solution, given
We
in (44),
which we
r
may
write
?
= <p(j8,
m; t),
= lK/3,
m; t),
=c
2
,
which
t
is symmetric with respect to the equatorial plane. That is to say, at the particle is in the equatorial plane and its motion is perpendicular to the radius vector. Its initial distance is (p (0).
initial
= *(0)+a,
g(0)
=0,
r'(0)=0,
q\0)=f'(0)+y,
117
c?
of areas so that
= c2 +e.
Can
we determine
these three constants a, y, and e, as functions of /3 and p, in such a manner that the series for r and q shall be periodic? The solutions can
r=<p(fi,n;T)+p,
q=f(P,
m; t)+<t,
<p
=c2 +e,
substitute
and
\p.
If
now we
these expressions in the differential equations (6), all the terms independent of p, <x, and e will drop out, and there will remain the following differential
(a)
+ (^^-|^)cos2t + | 8
j
cos4t] m
3
+
3
jp+jf-
3/3
suitJm
sin3r]p
2
j*
2
=
{3
+ (^^
+
3
jp +{[-
12/3 sin
r]p
+ [(-9O0 /3+f
+{f+[(f ^-f
/3
)sinr-^/3 sin3r]p
2
+
2
/3
+f
2
/3
cos2r]p
]<r
+
(50)
+
[{
-}p+
2
4
e,
(6)
(r"+jl
+ [-|/3
+|^COs2r] M +[-4e
2
3
2
/3
+7e ^cos2T]p
3
+
-}<r
+|[-3/3sinr]p+[(-30
+
-}p
3
4-
jp
+ [(3^-f )+f
2
/3
/3
cos2r]p
+
-}p<7
sin
these equations the coefficients of all the terms containing odd powers of a involve only sines of odd multiples of t and odd powers of even powers of p and cosines of even p; all other coefficients involve only
In the
first of
118
PERIODIC ORBITS.
multiples of t. In the second equation the coefficient of every odd power of o- involves only even powers of p. and cosines of even multiples of r; all
other coefficients involve only odd powers of p and sines of odd multiples of t. These properties play an important role throughout the entire discussion.
70.
The Equations
of Variation.
and
a,
we have
<
(a)
p"+[l
+ [(-4^+|^)-|/3
2 4
/
cos2r] M
4
+[(20^+6^^+| 8 )
J
+---J<7
= O,
r] M
(51)
(6)
</'+jl
+ [-f +f
2
/3
/3
cos2r]p
+
3
-}<r+
3
j[-
3/3 sin
+ [(-30 ^+|^)sinr-|/3
The
solutions of these equations
sin3r] M
= 0.
-}p
a= 2
1=1
c,e >V,(T),
<Pj(t) and ^(r) are periodic functions of r four values of the \ t (real or imaginary) are associated in pairs, equal in value but of opposite sign (33); and since the solution (44) contains two arbitrary constants, i. e., the origin of time, t
The
and the mean distance, a, it is known a priori that one pair of the X, has the value (33). If we suppose that X3 = X 4 = 0, the two corresponding solutions have the form
p
=C
z <p 3
(r)
+C
[^ 4 (t)
+t<p3 (t)],
<t
=C
values of <p3 (r) and \//3 (t) can be obtained at once by differentiating the solutions for r and q [equations (44)] with respect to r; and the values of [v4 (t)+t <p 3 (r)] and [^ 4 (t)+t^3 (t)] are obtained by differentiating ar and aq with respect to a. Thus two of the solutions of the fundamental set
The
We
two solutions
in
which the
X, are
not zero.
x
e*'
Xr),
(7
XT
e''
iKr)
(t-v^T).
119
^>"+2tV+[l-X +a
2
^+2^'+[l-X
where
c^
+fe2
+a M +&
2 2
+ M +
4
.]^
.]^
= = ]*>
,]
(52)
(),]
/J
sin t,
2
3
a3
a2
=(-30 ,3+|^)sinr2
2 2
|/3 sin3r,
=(-40
a1
&2
64
=((20^+6e
2 2
/3
|/3
cos4r,
multiples of
r.
With
respect to equations (52), it is known that <p and ^ are periodic with the period 2x and that X vanishes with n, since the problem then reduces
two-body problem, in which the characteristic exponents are Since <p, \p, and X are analytic in n, we may put
to the
all zero.
<p=2<p
1=0
i J lx
,
n ip=V\p 1=0
J
,
!
,
X=2X,m'1=0
(52) as follows :
The
expressions for
Coefficients of
<Po
<p }
\}/j,
are found to be
+ = 0,
+*. = 0,
<p
= of
0)
*o
*,- 7i
(53)
Coefficients o//x.
The
terms in
1
ju
are
(54)
?"
+ <?!= -2i\<p' -a
rp
</'f+</'i
= -2tX ^o-i'o
Since the periodicity conditions demand that the coefficients of cost and sin r in the right members shall be zero, we must take X, = 0, after which we
get,
on making use of
(53),
tf+^- f
Upon
integrating,
of
(55)
of cos2r + f fi
sin2r.
we have
sinr
sin r
*-*"
f = <y
,
COSr
cos r
+ af + 7f
/?
7f +
f/3
18
7f COS2r-}^Tr
of cos2 r
sin2r,
(56)
sin 2 r.
+f
/3
of + 1
/3
of
120
Coefficients of tf.
PERIODIC ORBITS.
The
coefficients of
/z
are defined
by
flitt
J
* 2+'j=-2tXs
,
?o-2^o-ai^i>
(56),
#+&=
2iM'J-M'o
(57)
or,
ft
cos2r
2
a2o>
0>
(58)
^+ft-|/8a-|/J-<
cos2r +
f/3ar>
sin2r
+ [^7
0>
+2iX2 Ti
sinr
+2i\y? cost.
In order to satisfy the periodicity conditions we must have
2x\a
4^
2
0>
40 af = O,
2
>
2iX2 7i0, +
/3
72
0)
= 0,
j
ai
-2t'X2 af
= 0,
+2lX2 7 2
by taking
0>
= 0.
(59)
The
last
two
yr-Yf-0.
On
solving the
first
two,
we
find
lt
\=2f
Equations (59) can also be
X2 .
'
a-ia
by
0)
= 0.
satisfied
a'
= a = 7 f = 0,
0)
0>
t1
arbitrary.
This leads to the development of the solution in which the characteristic exponent is zero, which will be discussed, beginning with equations (80), by
using the integral relations. It was known at the outset that the two values of X are equal but of We will choose the one with the positive sign. The solution opposite sign.
The condition a can be derived from it. ta 20) = still leaves us with an arbitrary constant. Since the equations are linear, this constant will enter the solution linearly, and may therefore be taken equal to unity, inasmuch as the solution after development is multiplied We will take then a[0) = l, which therefore by an arbitrary constant.
for the negative X
makes a 2 =
0)
i.
Consequently
<
= cosr
sinr,
&>
= 0.
0)
(60)
On
oj
and a2
,
0)
,
we
get
ft=|/372
+ a[
2>
COSr
(61)
l)
2)
j i-
1'
121
n\
The terms
n are defined by
+ [2iX
+4^(a< +tai
1)
,)
2)
)]sin
r+f/S-yfsm
2
2t-| /3<y
2
>
cos2r,
(G2)
+ [2tX27 + |/3 7
)
']sinr+[-|af/3+^^t/3+|^]co S 2r
1)
+ [|af /3+^^-^]sin2r-f^-Y[
From
the periodicity conditions
3
cos3r-f/3
1)
sin3r.
we must have
J
-2X
+40?(ar
>
-iO=O,
1>
f/3
7 i '-2iX27 r = 0,
1
1,
2iX l7 i
satisfied only
i
if
1,
+ f ^f-O.
= 72 = 0.
1)
The
last
"
The
i
first
two can be
only
if
X3
= (a"*
a ") = 0.
2
The
condition a"*
a2
"
again leaves us an arbitrary constant; it gives us ^i = c(cosr i sinr), but this is the same as <p multiplied by Cfi. That is, the solution is repeating itself one degree higher in n, and this, of course, should be expected since the
equations are linear, and <pa multiplied by any power of n must satisfy them. We may then choose the arbitrary multiplier equal to zero, which is the same " On integrating (62) with these values, we find as choosing a"' = a 2 = 0.
*- +f
7? + a? cosr+af
sin
r+\
& yf cos2t-|/3 T
sin 2r,
*.-+[f a'^-ft^^-lt^j+^cosr+^sinr
(63)
in m>
can be shown by induction at this point that <p and X are even series and that ^ is an odd series in ii. Furthermore, <p contains only odd multiples of t, and ^ only even multiples. Consequently
It
7r = 7f =
and
all
C = af = 7 r = 7f =
0,
Coefficient of n*.
The term
)
ju
is
defined
by
?; + ? 4 =
+ [-2XJ-4fl(a
-ia)-16fl-10tf/3 ]cosT
+ \2i X +40
4
(i a<
2
2)
2,
+ [6^0
4
+3/3
al
^
4
[(64)
-^/3 cos5r+^t/3 10 Id
sin5r.
122
PERIODIC ORBITS.
conditions that
<p t
The
shall
be periodic are
On
we
2
find
2
X4
=-80t-f0
/3
a?-ia? = \fi\
2 In the last equation we can choose af = 5/4 /3 and af' = 0. This choice will make the coefficient of sinr in <p2 equal to zero, and since the same thing occurs for each <pj it is evident that the corresponding choice will
by making the
powers of
4
/z-
We
2 ^ = a 4 cosr + [-f0
/3
-^^]cos3r+[f^ ^+t^]sin3r
+ I^/3 cos5r-^i i 8
4
j
sin5r,
+ [-^-^>in2r,
(65)
V>2
=
J/3
cost-
l^cosST+l^sinSr,
X=20
+[-80:-f0
4
,
]M
+
4
The coefficient, a of cost in <p is determined by the periodicity condition of <p That this process of determining the values of the X, and the constants of integration arising at each step is general can be shown as follows: Let us suppose that we have computed all terms of <p up to and including <p
6. t
<pj.
We
have then
^^ai^cosT + a^'sin.T
It follows
+ known
a
1
terms.
from equations
a^ and
"
enter
\}/
1+l
only as shown
explicitly in
J+1
it is
m - f 0a"> +
0$" - }/Ja>n2r.
123
it
(52) that
2
J
<p" j+2
<p J+2
[(40
/3 )
/3
t]^
|
+[-2\ +2 +4:6 M-ia<) +A J+2] cost + [2i\ j+2 + 4tf(i a< + a<") +B J+2] sinr
J)
(66)
where
A j+2
+3/3 a^cos3r + 3/3 afsin3r, and J5 J+2 are the known terms in the
2
2
coefficients of cost
and
sinr respectively.
From
we must have
)+B J+2 = 0.
(67)
X ;+2 = i
[A j+2 + iB J+2],
out,
u)
i
we can
(68)
then
. .
_ A J+2 iB j+2
802
(b9 '
In order to show that X ;+2 and a'" are real, it will be sufficient to show A i+2 is real and that BJ+2 is a pure imaginary. This is readily proved by induction, for, up to j = A inclusive, we have j-\ am i~i j+i
that
'
<Pj=
S
K
,
coSKT+i 2 nK sinnT,
fK
,
= i 2 fK C0SKT + 2 ^j
gfK
sinKT,
where
so on.
wK
and
is
grK
are
all real.
equations
it
From the form of the differential same forms hold for j = 5, then 6, and
is
That
A J+2
real while
B 1+2
purely imaginary.
A j+2 and B J+2 do not contain any Furthermore, terms in /3 independent of d\ and consequently the 8\ which appears in the denominator of a[ n will divide out. This is proved as follows If d\ be put equal to zero in the differential equations, then equations (52) become
it is
to be noticed that
,
the equations of variation of a circular orbit in the ordinary two-body problem, the plane of the circle being inclined to the plane of reference by an angle whose sine is 0n = s. Equations (6) can then be written
^_
Q-s*)(l-e ) _
is
R
(1 s )(l
2
__
e ).
2
=Q
(7Q)
_ (1-e ) Vl -s
sin ( g-7T)
q
o)
(1
-e )ssin(fl- ft)
l+ecos(0-0o )
,_
K
n
J
l+ec6s(0-0
where
124
If
PERIODIC ORBITS.
the equations of variation
now we form
r
by varying
e
r, q,
and
e,
that
is
by putting
= r +p,
dR
de
= q<>+<r,
we
find
= e +e,
where
qt
and
dR
dr
,dR
dq
dQ
dr
(72) are given
,dQ
dq
dQ
de
no
by
125
The
whero
solution which
l)
+i+
2
i2)
],
(75)
(1>
-/3 )cos3r
4
,
+^8
1
cos5t] m
w r
--Bbt+|>8in3T]MH[(J^i8+^)8in3r-^8in5r>+
'
r = [-^sin2r> + [(-^^-^3)sin2r]
+ (-^^ -f 6
2
3
(76)
/3
)cos2r+
-]
By
Xr
putting
e'
=cosXt+{
form
1
^^[^"cosXr-^sinXrl+it^'cosXr+^'sinXr], ^'^[^"cosXr-^'sinXrl+tt^cosXr+^sinXr].
(7?)
j
We have thus one solution of the differential equations. A second solution can be derived from it by merely changing the sign of i, or
P^^t^'cosXT-^sinXrl-tt^'cosXT+^sinXr],
1
^^[^cosXr-^'sinXrJ-if^cosXT+^sinXr].
By
finally
a = A[a
+ a]+B[a w -*},
and
As above developed, there is a certain arbitrariness in these solutions, owing to the manner in which the constants of integration were determined. They may be reduced to a normal form by multiplying each solution by the proper series in p, with constant coefficients. By this process we can make
2
(
"(0)+p
(2
>(0)
= 1,
a\0) -a
<2)
(0)
-#*.
(79)
Since
and [a +a] are sine series, they vanish for t = 0. The third and fourth solutions of the equations of variation are
(1)
(2)
[p
*-Tt
**
H>
Pi
~D
~teT'
ai
-D
~dT'
(80)
the r and q being defined by equations (44). In performing the differentiation in this last solution it will be remembered that r and 0, are functions of a. The third solution also can be normalized by giving the arbitrary
= 0,
r,(0)
= C/V
126
PERIODIC ORBITS.
As already
is
*- j >[t S? + *J
where ^ 4 and
\f/ t
'*- D
+A>
2ir.
(81)
are periodic functions of r with the period = previous solutions, this can be normalized so that at t
As
in the
The
and ^4 are also easily found by substituting (81) in the equations of variation and solving for these variables (which must be periodic)
functions
<p 4
.
Upon
we
fundamental
p=+yl|cos(l-X)r+
[-^
(
cos(l-X)r+ f /3 COs(l+X)r
2
+ |/3 )cos(l-X)r
4 4
+(! a<-i0 ) cos 1+x T + (-!^ -!/5 ) cos 3 x r + X)t+ i^cos(5-X)t] m + i!/?cos(3 j + sin(l -X) r + [(-I^-|^) sin(l sin(l + X)r X)r2
#[
/3
-^ sin(3-X)r]
2
(82)
sin2r] M
/3
/3
/3
/3
,i
a= +Al [
j8
sinXr
|/3sin(2
- X)
r] M
+ [f
2
/3sinXr
3
-^0 /3sin(2-X)r] M +
+ #[[f +
/3cosXr
+|
/3cos(2-X)r]M+[-|-0
/3cosXr
3
+
3
/3cos(2-X)r] M
(83)
<7J[/3cost]m+[0]m
M+
+
2
}
3
+Z>{{[{^sinr]
+ r{[-f/3
/3
'}}
127
Special
Theorems
in
for the
Non-Homogeneous Equations.
The
general theorems, proved Chapter I, solutions of non-homogeneous linear differential equations with periodic coefficients, presuppose merely the conditions that the coefficients are periodic with the period 2w. Additional facts with regard to the solutions can be established
additional facts are specified with regard to the coefficients of the differential equations. The equations of variation, (51), may be written
when
dpi_ =
dr
ft Hi
'
-JT
dt
dpi_-x
da _
1
'
2 ltZ~*?ti dr
'
d (T, J" d T
= ^ Al + ^0-1,
(84)
where the notation with respect to the d's has the following significance: Even subscripts denote functions even in t, and odd subscripts denote functions odd in r; one dash indicates that only odd multiples of r are involved, and two dashes indicate that only even multiples of r are involved. The solution of equations (82) and (83) may be characterized in the same manner, and are then
Pl
(t)+t^(t)],
ft
=4(T)+BA(r)+C| (T)+Z)^(r)+TA(T)],
(85)
(r)
(r)
(r)],
where the notation is the same as for the d's with the exception that in the first two solutions every integral multiple of t is increased by * At, e. g., cos (34-X)r. On these terms the dashes refer only to the integral part. now we have the non-homogeneous differential equations Suppose
dpi
_ =
pi}
dp2 ~
_ =
7T
2
Pi+03<^i+6
W,
(86)
dr
dr
where g(r) and/(r) are periodic with the period 2w. Since the characteristic exponents are 0, 0, s^V^TX, by 31, the general solution has the form
Pi
(Pi) 4-
(Px) 4"
!(t) 4-
at a 3
4" 6 [| r a 3
2
+ta
P2
=(p
2)
(87)
(r l
=(<r
~\
0-2
= (,) 4-17, =
(a,) 4- w 4 0) 4-
ar
5 3 4- b
[|
t 5 3 4- r 6 4 ]
128
PERIODIC ORBITS.
t
(p,) and (<r ) are the complementary functions, and the |, and j?, are the particular integrals of which the w, are the periodic parts, and where a and b are constants which depend upon the differential equations. Let us suppose that g(r) is an even function of r and that/(T) is an odd
where the
function of
r,
and
let
*(0)-*(0)-0.
On
changing t into
J
t in equations (86),
7
-
we
get
^;Pi(-t)= -p
^<j2
(-t),
^p (-t)=
2
p1
(-T)+0
tn(-T)-0(T),
(88)
(t 1
(-t)=-o- (-t),
2
^o- (-t)
2
(-t)-0
(-t)+/(t).
From
we
obtain,
by eliminating
differential equations
^;[p
W-p (-t))=
1
[p 2
(t)+p2 (-t)],
[<Tl(T)+<r l (-T)],
(89)
fo[<Tl(T)+<Ti(
[<r2 (r)-(T2
T)]=
W
[
(T)-<T2 (-T)],
(-r)]
9;[p
(r)-p (-p)]
1
+ i:[p
(r)+p (-T)].
1
These equations are the same as the original homogeneous their general solutions have the same form, viz.,
set (84).
Hence
p i (T)-p 1
(-T)=Aa
(T)
+ Ba
(T)
+ Ca
(T)-\-D[a i (r)
+ T7
3
(T)],
Pi (T)+p2
(-r)=Aj
(r)
(90)
<T2
(T)-<T2 (-T)=AT 2
(T)
3 (T)].
Upon
we
find
from the
first
of these
= AZ (0)+Da
2
(0) }
= 4l
(0)+Z)5 4 (0).
129
But it is or the determinant a2 (0)5 4 (0)-a4 (0)5" 2 (0)=0. = 0. Therefore readily verified that this determinant is not zero. By virtue of the hypotheses made on the initial values, it follows from the
A = D = 0,
A=D
= Bft(0)+CA(0),
and hence
0=*By (0)+CyM,
t
B = C = 0;
consequently
P2 (T)+P2 (-r)=0,
(91)
<r2
Pi(t)-Pi(-t)=0,
<t1
(t)+<t1 (-t)=0,
(r)-<r2 (-r)=0.
r,
Theorem
of
r,
If g(r) is an even function of r andf{r) is an odd function and if p2 (0) =<r (0) =0, then p^r) and o-2 (t) are even functions of r, and
I.
1
P2 (t)
and
t.
In the same way it can be shown that if g(r) is odd and /(t) is even and if p,(0) = <r2 (0) = 0, then p and <r2 are odd, and p2 and a are even. Let us suppose now that g(r) contains only even multiples of r, and that /(r) contains only odd multiples of r. The general form of the solution
x 1
will
be the same as
(87),
and
ij,
and
r? 2
tions
;t!,(t)-&(t) a,T
(92)
nM = %(r),
Let us denote ^(j+t) by into r+7r in (92), we have
(t)
and ^(t+tt) by
i&(t).
Then by changing
&t) = &t),
|-ii(r)-^(r),
(T)
= +e
a[(T)-o3
(93)
f
it
uiW = -
From
follows that
(94)
130
PERIODIC ORBITS.
solutions of these equations,
The
(84), are
^-^ = 4a
(r)+J5^(T)
2
+ C^(T)+Z)[r (T)+T^(T)],
4
^+^ = ^7 (r)+^(r)+C^(r)+Z>[^(T)+r^(r)],
1
r, a
3 (t)].
On forming
we
get
& fLftM
l8
i(H-*r)
2
-^ = "2(r)-
(r+7r)-(a7r+|67r )^-67r(r5;+I),
(96)
1i+?[ = W3(T)+w
l2+'72
(T+7r) (T+7r)
= ''4('")+w
2
4
(a7r+|&7r )53
67r(r53
+5
4 ).
A=B = 0,
CO^T)
oj2 (t)
C=-(a7r+!&7r
),
D=-bT,
(t
+ 7t)=0,
O> 3
(t)+C03 (t
+ 7|-)=0,
o) 2
= 0, (t+7t)
w4 (t)+co4 (t+tt)=0.
oj 4 (t)
Therefore a^ (t) and a>2 (r) contain only even multiples of t, while w3 (t) and contain only odd multiples of r, and by carrying this result into equation (87),
i
we have
2
= iM+aTa +&[-T
3
a3 +T o4
],
(97)
These
results
may
II.
be expressed
in
If g(r) contains only even multiples of r andf(r) contains and 2 contain only even multiples of r, and only odd multiples of t, then and contain odd r; 2 771 only multiples of r.
a
Theorem
If in addition to the above hypotheses we suppose that g{r) is an even function of r and/(r) is an odd function of r, then & and % are even functions and 2 and % are odd functions; therefore 6 = 0. But if g(r) is an odd function and/(r) is an even function of t, then t and ri2 are odd functions and
,
and
t^
and
in this case
0.
131
Theorem III. If g(r) contains only odd multiples of r and /(r) contains even only multiples of t, then & and 2 contain only odd multiples of r, and ^ and Furthermore & Tfe contain only even multiples of t. 2 % and % are periodic
, , ,
2ir.
If g(j) is of the form 2 m_,cos(jX)r and also if /(r) has the form 2ri sin( 7'X)r, then, since lX are the characteristic exponents of the 30 and 31, homogeneous equations, the form of the solution is, by
/
i
a)
C2)
T^(r),
(2)
fc-Zg? cos
X)
T+2gsin(ic
X) X)
t+o VA(t) +a
(I
7-ft(r),
% = 2 rf cos (k
r ?2
X) r
+2 r? sin(/c
2)
r+a W Ty
(1)
(98)
2
(r)
-fa^ry^T)
C2>
r^(r);
an even function and /(r) is an odd function, and i?2 are even functions, and 2 and rjj are odd functions. Therefore all the coefficients in (98) which have the superfix (2) are zero. But if g(r) were an odd function of r and/(r) an even function, then all the coefficients in (98) which have the superfix (1) would be zero. Therefore
but, since o(r)
is
t
Theorem IV. // g{r) has the form 2m cos(,/X)r, and if /(r) has V IX are the characteristic exponents of form 2 n sin(j\)r, where
J
.,
the the
form
= 2??
cos
|2
= 2p sin(6X)T+^r/3
6
_
2
rji
=2 p
*
sin
(cX)
t+At^W,
=
(r),
i?2
(99)
From
similar reasoning
we have
the
Theorem V. // g(r) has the form 2mjSw(iX)r and if /(r) has V 1 X are tfie characteristic exponents of form "LUjCos (j\)r, where
homogeneous equations, then
the particular solution
the
|1
(r),
&=2 r
It
is
cos
(6X)t+.B 7-/3,(7-),
a, b, c, d,
=2 r
i
sin
(dX)
in
t+t
=
5 2 (t).
(100)
understood that
Theorems IV and V.
132
PERIODIC ORBITS.
It will be convenient 72. Integration of the Differential Equations. hereafter to use as notation for the fundamental set of solutions, (82) and (83), of the equations of variation
p
(r
(101)
cos
[(2n+l)X]
t,
7iW
o,(t)
7M
.(t)
74 (t)
o4 (t)
73 (t)
133
a, 8,
and
e,
we may
2
write
2
P=Piooa
a
<r 100
+Pioia+Poii5
o- 010
o'ooi
r2oo
'no
o'o2o5
o'101
ae
c7-
011
+ Poo2e + 5e + +
cr002
',
The
differential equations for the p m and <r are obtained by substituting these expressions in (102) and equating the coefficients of similar powers of
the parameters.
Coefficients of a.
PlOO
The
by
^3 PlOO
02 PlOO
03 OlOO
= 0>
O"lOO
04
"lOO
= 0-
(103)
The
same
i
as the equations of
variation,
+ Co (t) +D[a (r) +t o (t)], *m = A 7l (r) +By (r) + Cyi (r) +D[y (r) + r 74 (r)].
+jBo
1
pm = Aa2 (T)
(t)
initial
conditions
we must have,
"l00
at r
= 0,
= -M
we
Pl00
= ">
= ")
""km^U-
From
these conditions
find that
= l, 4t;(0)+ D[73(0)+%(0)] = 0,
4o,(0)+Z)a4 (0)
j
Ba'M + Ca'M^,
ByM+CyM=0.
is
j
J
The
_>ift
7i(0)+74(0) -r
_4< ft 100
n_ U
_7i(9_) .
_ ft m
.
l
100
R=n B C=n o u.
r>
(105)
A
Hence the
form
3
= 4i
o,(t)
+vOo
(r)
+ro
(r)],
(106)
Coefficients of
Po'lO
5.
The terms
must
satisfy
02 PoiO
03 O"010
= 0,
(103),
^4
0"oiO
^3 PolO
= 0.
(107)
initial
conditions
we
must have,
at t = 0,
Poio
= 0,
Poio
= 0,
2)
1
Coio
= 0,
O" io
=1
The
<ri
<
j
J
-^
where
a
(i)
-^010
B^i 0^(0) A
A
, '
"^010
(2) 4(2) A
Q8(0)
|
134
Coefficients of
POO1
.
PERIODIC ORBITS.
The
terms are
01 POO1
03
O"OO1
= 0OO1
>
^001
04 ^001
+^
Po01
= 0.
(109)
a periodic function of r with the period 2ir. Furthermore, it involves only cosines of even multiples of t. Consequently, by Theorem II of 71, the solution has the form
right
The
member,
dwl
is
Pm = Aai (T)+Ba
(T)
(T)
+ Ty
where a is a constant depending on 8Ml ab (t) is a cosine series involving only even multiples of t; and yb (t) involves only sines of odd multiples of t. From the initial conditions it follows that p xl am p' wl and a' m all vanish at t=0. On determining the constants of integration so as to satisfy these
; ,
is
Am = Ai
where
al
(T)4-^[a
[
(T)
+ ra,(T)] + a,(T),
+TT4 ( T)]+ 76 ( T ),
eoo^AZy^+AZ
73 ( T )
(0))-a
a4 (0)7^(0)-a8 (0)(73(Q)+74(0))
^ = a+^[a
,
_
e
a.(0)7((0)-ai(0)7i(0)
A
(t)
=o
(t)
- aa
it
(t) ,
y6 (r)
=y
(t)
- ay
(r)
It will
an important
vanish.
and
is
By
hypothesis, a5 (t)
73 that the value of a6 (r) for t = plays necessary for us to verify that it does not is the periodic part of the solution for pm in
o-ooi
= lA (t) + ary
(t) ,
where
ip
and
2
\p
We
find
2
cos2r] M
or,
(r)
and 7 4 (t),
2
^"+e
M+
we have put
a = aa +ai ni +
135
Let us put
now
2
P = Po+PaM
^ = ^iMHM
'
'*
i
series in n,
<p
and ^ must be
We find
Po+Po=l,
<p
=l+C
COST,
sin2r.
have
periodic we must put a = 3/2, and then, after integrating, 2 ^"C, sin t 1/2 /3c sin2r. From the coefficient of /x we obtain
is
we
Pa
+ Pa =
7c
/3
cost
+ other terms.
Therefore
Since p2
<p
is
periodic
we must take
in
c
2
equal to zero. a=
=a
(t)
= 1 +power series =o
/x
+power
2
series in
/x
Consequently
a, (t)
6
(t)
a a4 (r)
= 1 + power
series in
p.
(112)
for t
= 0.
of the second degree in a are defined
200
The terms
-^200
>
by
1 3)
(1
+^100 [0200
a
<*2
0ilO
<h Ti
t
^020
7l]
\
+AZA% [2Bm a (Ta +aJ +6m 0,(774+7.) + 7i (t a +a f+2JUYi(r Y+7.>] +^^[02oo(Ta3 + a +0no (Ta + a (774+73) +C (t74+7
t
3
4)
4)
4)
3)
],
&200
+^^^[ i[2^
3
00
+ 7i a (Ta3 + a )+^
a
7i
0oao
10
+ a )l+2^
4 2 3)
20
7i(T74+73)]
"
+^^[^00 (Ta +a +0
2
4)
1 lo
+0Ko(t74+7
1-
By
the
initial
conditions p200
and
Since the equations (113) are linear, the solutions have the form
p 200
= Ao,(t) +5o
(t)
<r200
+^^P (r)+^^^foiP2W+^fooV3(r), - A 7i (t) +5 72 (t) + CT4 (r) +D [y.W +T7 +^Vi(r) +A&AZ Ur) +A%Ut).
)
(114)
4 (r)]
136
PERIODIC ORBITS.
the initial conditions,
Upon imposing
we
find
B = C = 0,
.
, q
...
,<
,>
A A
^100
D_
g.(0h ((0)
-^ (0)0,(0)
"^-lOO
>
On
we have
X3 (t), y3 (r),
(\ll2l
J
where
S.M^.M+^OWOW.M+^O)]
ft(r)-,(r)+
a,M
^(0)^0)
~fty
3 ,
>+ ^(0)
'
T,W
and similar expressions for xi1 yi ,xi1 and ?/ the values of which we shall find we do not need. The properties of and y are known with the exception of and ypu which we will now investigate. The functions <p and ^ are <Pi those portions of the solution of the differential equations which depend upon the coefficients of A[^. These coefficients are homogeneous of the second degree in o^{t) and *i {r). In Rm and Si00 the expressions 6m UO and 6m contain only cosines of even multiples of t; 2OO 8m and Bm contain only sines of odd multiples
,
oo
form
r.
= 2 a,
cos[(2n+l)X]
(t)
7,
= 2
b sin
[2nX]
concerned,
Rm
R =2
i00
cos2nr + 2 a
cos
[2n2X]
2)
t,
5,00=
S
n=0
6i
137
rise only to periodic terms 31 those By parts of Rm and Sm which are independent of X.give rise to terms in the solution which have the form
By
in the solution.
P=VM+c[Ta,{T),
where
ff
=P
(r)+c;T7 4 (r),
2?r.
p,(t) and p 2 (r) are periodic with the period functions x 1 (r) and y x (t) have the form
Consequently the
(r)=P (r)+c
1
ra3 (T),
(116)
Pt (r)
a.8.
2o\
terms are
Coefficients of
The
3
Pno
+9
Pno
+
2OO
<Tno
= Rm
llo
cr
no
O'no
#3
Puc
m 8mi
(11 7)
where
/t llo
= 2 Aj OO 4 olo [0
J
a2
a2
7 +0O2o7i]
1
Tl (ra^+a,)
+a
(r7 4
+7
3) f
+20020
(t74
+7
.
3 )] 2
+2^; J,^
o
o [02oO
(ra,
a4)(T7 4
+ 73)+0o2o(r74 + 73)
6 iJt
];
and
aS 110 is
obtained from
functions
.
R no
by
replacing d
differ
by
The
Rno
and Sm
stants A, Jk
The
initial
200 and aS200 only in the conconditions impose the same conditional equations.
from
Consequently the solutions differ only in the constants express them at once without computation in the form
Pno
-{2
iik
so that
we can
= 2 A lw A 0l0 x^t) + [A lM A 0l0 A 100 A 010 ]x (t) +2 A 100 A 010 x am = 2AZ AZVtM+lAZ AZ+AZ AZWr)+2AZAZy
x,(t)
(t) ,
(T),
where the
of r as in (115).
it is
Coefficients of 8
By symmetry
2
i> 1
seen that
|
J
o2/2
2/3(r).
Coefficients of
2
.
first
powers of a and
2
,
are
2 homogeneous of the first degree in the A's, the coefficients of a a5, and 8 The coefficients of the are homogeneous of the second degree in the A's.
first
power
of
are not
homogeneous
But if the functions a 6 and 76 the second power are not homogeneous. were zero the coefficient of the first power of e would be homogeneous, and therefore the second also. By symmetry, therefore, we can at once write down the terms involving the ^4's to the second degree. To these must be added terms in the first degree in the A's, and one term independent of
the A's.
13S
PERIODIC ORBITS.
The
terms are
Co02
P(X)2
+ 6 pm + 6 m = Root
t
3
(T
O* a00l
+ ^PcKH = "to
>
(120)
where
"'002
"200P001
"noPooi^ooi
"lloPoOl^OOl
= "2O0P0OI O(|02
The terms involved
in
T" "(BO^OOl
in the following table, where the coefficients of the constants given in the first line are the products of the functions in their respective columns and the functions of the same line
P^
are
shown
of
is Oj
UO
and
Origin of
term
139
These terms
#3 0101
= -^101
""lOl
^4
"l01
#3 PlOl
= ^101
>
(1
24)
where
fiaOl
= ^200 [2 PlOO PoOl] + ^110 LPlOO ^OOl + Pool ^lOo] = "200 12 PlOO Pool] "T" ^110 [PlOO ^101 Pool
"o01
"f" ^020
[2 0"ioo
[2 O"ioo
O"ooi]
^101 PlOO
"T"
O"ooi]
following table for R m constructed like that on page 138, shows the character of the terms entering into these expressions:
The
Origin of
term
140
PERIODIC ORBITS.
73. Existence of Periodic
chosen the
initial
Orbits having the Period 2kt. We have conditions so that at t=0 the particle is crossing the p-axis
orthogonally.
It is
if
at
again crosses the p-axis perpendicularly, the orbit will the motion in it will be periodic. The conditions that the particle shall cross the p-axis perpendicularly at t = T are that at this epoch p' = a = 0.
have the period 2o\ Therefore we shall Since p is an even series in t, and a is an odd series, all the choose T=kit. termsin p and a are sines, and consequently vanish at purely periodic = The terms which do not vanish must carry t as a factor. The k7t. t conditions for periodicity give us the two equations
The equations
of variation
p' (kt)
110
101
002
(127)
011
+&i01 ae+&(xa e
where a tjk and bi]t are the
coefficients
in
72.
Their
= -"l0O **l
-"100
/
^010
= -"010 ^!
A -"010 V >
h U 100
n V >
/|
h "010
4<2)
>
77
~
T**m
t
(2)2
*'S
>
""100 *J
"200
M no _94) "-"100
J,
J)
Xr_Lr4(i>
i
L-"ioo
c,)
>
no
4<i) 7T -4- T>4 _oJ( 4-/d A a) ]77 4-9 z A ^-^loo-^oioi/i rl/Moo-^oio ^ioo^oioJ
>4
(2)
(2 >
i2)
J/2
-<*ioo
-^oio
a) 77
i/3
>
_
"020
7)
>
"020
- 4
Cl)2
-"010
(128)
>
"101
ioi
4U) AW~ \\ AW 4(2) Am 4(1)17 9 4(2) 4(2) X 4(1) ITj. 4(2) XT _9 ^-"100-"001 X 1T^L-"100-"001^-"100-"001J^2 ^-"l(K)-"00' -"100^4 -"lOO^S = 2-A 100 4 (W1 ^ 4-[4 1M 4 001 +A, 00 ^ M Jy -(-2^j 00 ^ 001 ^3-}-^l 100 4"-4 100
1 I
~
3
>
t/ 4
t/6
m A a) ~^ 4-\A m A a) 4- A (2) A m ^4-9 n 011 = ^ 9A A <2) /4 <2) 7~4- 4 (1 7"4-/l (2) X ^" u -"010 "001 -^l l"010"001 -e -"010 -"001 ^T ""010 X 4 "010""OOlJ X 2 -"010 5
>
I
I I
h u on
=9 A A 4A 4- A 6 -^oio-^ooiyi^L-"oio-^ooi^-"oio-^r>oiJj/2T^(1)
(1)
~7i
Am
<2 >
i2)
A m l77 4-9 A Am
""001
/d
77
X2
-"001
where u = kit-t-1 UT
,
= KTry
Xi^-r1
(XT
Vi
= Vi
atr = /c7T.
(129)
141
series in o
first
as a
2
We
power
"
and
5.
obtain
=
where the
10
a+
e
01
5+e
2l)
a2 +eiia5+e02
(130)
coefficients
e,n
Qioo
Otooi
(w
11=
142
PERIODIC ORBITS.
On
changing the
e
~i
x,
into
y~,
and u into
v,
we
get
^"
Hence
ioJ
^p1 L
_
\v
aw
10
>
o01
4> l0
J _ ^100
(
^oou
J
j)"i rEi_^.n
lw v J
V* ~\
I '
(2'2
yj
100
["6
2/_6~]
j,
I
J
Lw
But T-
- -1
and T-
- ^~\
vanish since
x^^u,
as
is
Vi^c^v,
xt = c4
u,
yt =ct
v,
and
(123).
is
On
referring to (122),
c,
it is
also seen
that
equal to
A A
-"
\Xj __
i7~I
/4
I
Hence
C 20
_C
,
20
4 \A _ _ ^ioq l/Moo
(2)
-fl-ooi
-^100 -^ooiJ
(2)3
A a) ]f^
-fl-ooi
v A
(136)
we
A
<
find similarly
(2)
010llMOO-' 001
100^ x 00lJ
(1)
(2)
A U>
(2)
(2)
(2)3
-<iooi
=[02-02j
J
-
(2)
(1)
(2)
_ -^010
A (2)
^OOl.
(1)
(137)
(2)3
[a _ y*]
Lw
iloni
vJ
substituting in (133) the values obtained for the coefficients, that the second degree terms in a and 5 are factorable, giving
On
we find
n u
i^- [ x *
1
^1273
.~|
T^ /ljoo^iooi 4 Am M
a>
A 100 Jl A 00l Ji
^~f~
ja
(138)
I-
in
There are therefore two solutions for 5 as power series in a. On substituting the two solutions of (138) for 5 in (130), we find the two corresponding values of e. We thus obtain the two solutions
A (0
A
(2)
moo
-"-po i
loo
(2)
oi
(1)
~a (0)-aa
6
7l(0)-qTa(0) a +
4
%(0)
a6 (0)
a
a+
(0)
A
(2)
(1)
(2>
-"10 A )
-^-010
_ _
-<*100 -"-010
(1)
(2)
+
'
a a5
(0)-aa 4 (0)
o,(0)
,4(2)
5=-floi
1(2)
-"Moo _ A (2) a
_ "^OlO
JT2) -"001
1(2)
S_l_
I
"001
143
is
in (111),
and
was shown in (112) that a 6 (0) is distinct from zero, Thus one solution for e begins in (79) that a, (0) is equal to unity. with the first power of a, while the other certainly does not begin before
dyjdr
for r
= 0.
and 7a(0)
the value
It
first
power
of a.
Construction of the Solutions with the Period 2 kit. We have the existence of series for p, a, and e proceeding in powers of the proved initial value of p, which we will now denote by e*. The series for p and a are
periodic in r with the period 2kt,
e sufficiently small,
is
and
<r
periodic.
The
series for p
even in
t,
and the
is
odd
in t.
e
i
These
,
series
(T
e
e
+p +a
+ +
(?+
e
3
(141)
e=
el
e+
e2
e3
We
way
that p and
<r
shall
and integrate the coeffiand determine the constants in such a be periodic and shall satisfy the initial conditions
or(0)=0,
p(0)=e,
p'(0)=0,
af(0)-r,
a constant which will be determined in the process. substituting the series (141) in the differential equations (102), find for the coefficients of the first power of e
where
v is
On
we
Pi
<ri+0t
^ +0
P!
= 0.
(142)
r,
the condition of orthogonality p must be even in r and a odd in the solution complying with these conditions is
By
and
pl =
cr
J
J
>
where a5 (t) contains only cosines of even multiples of t, and yb (r) contains In order that this solution shall be only sines of odd multiples of t. that periodic it is necessary and sufficient
Z>
(1>
=-,.
,
Upon
D m the
i
3
solution (143)
2
becomes
1
= A w a,[{T)+t a {r)-aa {T)}=A w a {T)+t a {r), = A a) y [(t) +e lT6 =ay (r)]= A w Tl (r) +, *<t).
i
It remains to
impose the
initial
a)
condition that pt =
a2
at t = 0, which gives
(145)
l=A
where
Oj
6
ia
of a 2
and
<x
for t
= 0.
144
Coefficients of
PERIODIC ORBITS.
&
The
2
Pi
<j\
+^ +6 + d P =
0J P2
t (T2 3 2
^001
0H>1
Pi
0*00
Pi+^UO
Pi+0110
es
Pi
Pi
""l
0200
O"i
+ +
^OJO
"l
0O2O 0"l
= ^2 = ^2
(146)
Every term
in this
either
A a>
]
or
Cj
as a factor.
Arranged
R = +4
2
(,,J
[
6^
a 2 +dm
a, 7l
+Bm y\
0020
+A + +
S
2
m
e,
o]
]
02oo <*6
^iio
a6 76
7e
0ioi <* 6
*2[0OOlL
=+A (m
+vl
(1>
e1
2oo
llo
6)
+20o2O
-y,
76
0200 tte
^llO
a6 76
0020
7o
The coefficient of A lm in both R2 must examine the character of R2 and S2 Its expansion thereand *S2 is homogeneous of the second degree in a, and y fore involves terms carrying 2\r and terms independent of Xt. By 30, the The terms independent of Xt are solution for the terms in 2Xr is periodic. cosines of even multiples of t in R2 and sines of odd multiples of t in S2 These terms have the same character as those in the coefficients of el and e2 and will be considered under the discussion of those terms. The coefficients of A w e in both R2 and S are homogeneous of the all terms of which carry the first multiple of At. first degree in a 2 and 7, By Theorem IV, 71, the expression for p2 will carry the term tcl^t) and for <r2 the term T72 (t) Non-periodic terms of this character do not arise elsewhere in the solution. Hence, in order to avoid them, we must take
.
either
Aw
or
4 = 0.
e1
If
we choose A w = 0,
= l/a^,
A the existence proof. But if we are in agreement with the second solution (140) of the existence proof. We will commence by developing the first solution, in which
then, by (145), x is determined thus agreeing with the first solution (139) of we choose e^O, so that by (145) m = l/<^,
A> = 0,
% -I.
FIRST SOLUTION.
Since
Xt, vanish.
A a) = 0,
2
R
no
Xt, or
any multiple
of
R =e
2
2
t
[02OO
S = e\ [0joo
+ al +
<h
a6 7 a 7
no
+ 0^ 7e + + 7e]
O2O
0ioi
ae]
2 0ioi
>
C1
J
47")
145
We
have
also
Pl
= ^l),
Ctc
*-*&*..
OLs
Cl
= iCl<)
(148)
contains only cosines of even multiples of Since p2 is an even t, and S2 contains only sines of odd multiples of r. function of t and <r2 is an odd function of r, the solution is
2
It follows
p2
= yl
(2)
a2 (r)+Z) '[Ta3(r)
Tl (r)+i)
(2,
C2
^=A
(2,
[a 6 (r)+ara 8
(r)],| (
[T 5 (T)+aTT 1 (r)].j
In this solution the terms are grouped according to their origin. The first two terms are the complementary function. The third arises from 2 the terms carrying e as a factor. The fourth arises from the terms having e2 2 as a factor, a2 is a constant depending upon the coefficients of e in the
differential equations,
and ^(t) and y (t) are the power of e, 2 (r) and f2 (t) r with the period 2-k, and so constituted that 2 (r) even multiples of r, and f2 (r) contains only sines In order that p2 and o-2 shall be periodic we must
5
same functions
as in
first
t]
rj
of r.
Dm =
which makes
p2
<
a2
ae 2
r2
a4
(t),
(2)
(150)
Ta(r).
In order that we may satisfy the initial conditions, we must have p2 = t = 0, which determines ^ by the equation
at
_
2
ai ai -y2
-A
a)
a,
at
It is
obvious that
3
which so
far
is
arbitrary,
coefficient of e
it
terms
must be
<2)
A w = 0.
)72
we have
Ct
^^a (r)+
6
(T)-a2 a1 (T),
(151)
a*
ri
-M^ T
a6
of
t.
,Cr)+r.<r)-flb'*(T) l
r,
and
<r 2
contains only
odd multiples
remains to show that this process of integration can be carried on indefinitely. On assuming that up to and including p,_, and <7,_i every that the p, contain only cosines p and <jj is periodic with the period 2x, and
It only
}
of
even multiples of
and the
07
r,
except that
146
p,_,
PERIODIC ORBITS.
contains the term ^"""^(t) and a,-i contains the term A^^y^r), it will be shown that the same conditions will obtain for the next succeeding For p, and <r, we have, from the differential equations (102), step.
.(152)
.
crf+04 <r,+0,p
From
the properties of the differential equations it is readily seen that $j contains only known terms all of which are cosines of even multiples of t, and that Vt contains only known terms all of which are sines of odd The coefficients of A u ~" are homogeneous of the first degree multiples of r. in Oj and 7,, and consequently each term involves a first multiple of Xr.
give rise to non-periodic terms of the form ra^r) and ty2 (t) in the solution. They carry A"~" as a factor, and since terms of this type arise
They
nowhere
else,
putting
A u ~ u = 0. The
periodic with the period 2 77-, and where by Theorem II, 71, rj^r) contains only cosines of even multiples of t, and f (r) contains only sines of odd multiples of t.
f,(r)
are
In order that
that
p,
and
<r
shall
be periodic
t
it is
D w =-a
which makes
ft
)
-ae
i ,
= ^" a2 W+'7iW-ta4W+e a (r), = A"' 7l (r) + Ur) ~ ai73 (r) +etJi (r)
<
From
j
we must have
p,
=
;i
at t
= 0, which
determines
by
_a
Thus the constants
properties
<
a4 -7?<
-A " a
(
a.
The p and a have the are uniquely determined. for those having smaller subscripts, and the process of
t t
indefinitely.
Every
Am
is
zero.
Since no
terms involving the Xt enter, the solution has the period 2t, and the orbits represented belong to the class of generating orbits with which we In other words, we set out with a generating orbit for which the started. initial distance was, let us say, r and we have found another generating orbit for which the initial distance is r +e (e arbitrary). There is nothing in r is for a constant of an function this, /3. arbitrary surprising
,
147
Let us suppose we had started with a definite value of 0, for example which gives a definite generating orbit with a definite initial distance r O Let ^s seek now the generating orbit for which the initial distance is r +e. If e is sufficiently small we can evidently give an increment e to /3 which
,
.
will increase r
by
the
amount
=/(&),
e.
r.
We have r +e=m+e).
the second equation in powers of
2,
e,
Expanding the
right
member of
e
we have
-*L a.L<.
2
which
gives,
by
form
'
Then, by substituting
/3
e = c e+c e + = +c e+c e +
2
2
/3
and
arranging the solutions as a power series in e, we obtain the orbit in which the initial distance is ra -\-e. As these are the same conditions that were
imposed when we sought new orbits through the equations of variation, it was to have been expected that one of the class of generating orbits would
satisfy them.
SECOND SOLUTION.
We
tion, in
return
which
and therefore
now to equation (146), and continue with the second solu= and A m = l/a From (82) it is seen that a, = o^ (0) = 1, a) = A 1. Hence in the second solution
et
2
.
Pi
=a
a)
(r),
et
<Ti
= 7i(t).
of (146)
(155)
On
and
2
and S2
0020
become
/
= [0200 a = [0200 $2
fl
+ +
#110
0110
+ <hJl +
G2 Yl
Tl] 7?]
20OO1
>
,m
j
0020
All of the terms in these expressions except Cj^oo, are of the second Therefore they involve only terms carrying 2Xt and degree in a2 and 7! In the solutions terms independent of Xr, and dm is independent of Xr.
.
As for the terms independent the terms depending upon 2Xt are periodic. of X, R2 contains only cosines of even multiples of r, and S2 contains only These terms give rise to non-periodic terms in sines of odd multiples of r.
the solution, which has the form
p2
=4
=A
(2,
o2 (r)+Z)
<2,
[ra 3 (r)
e2
[a 5 (t)
+ ara
(r)] ,
t2)
<r 2
7l
to +Z>
(2)
[tY4 (r)
+7
to]
+&
2
(157)
(X, r)
4
+ [f (T)+a T7 (r)] +
2
e2
[7 6 (r)+aT7 4 (T)],
terms involving X; % and f2 <p2 {\, r) and ^ 2 (X, r) are the periodic are the periodic terms with the period 2w; a 2 is the constant belonging to the non-periodic part; and the coefficients of e2 are the solutions depending
where
148
PERIODIC ORBITS.
coefficient of
2
on the
necessary that
<r,
to
2 2 4 2 8
p2
(r s
(T),
J
(T).
To
since
we must have
p 2 (0)=0.
Hence,
= 1, A Oj(0)
defined
by
r? 2
(0)
+a a
2
(0)
-e a
2
(0),
is
a constant which
;
is
the coefficient of
Coefficients of e
P3
3
.
+ O2P3 + 03^3 = #3
0101
<h
+ Q&z +
'lP2]
#3 Ps
= ^s
<T X (Tt
>
1 59)
where
R} =
^OOl
3
2Pi
^3
+ +2 0^0 + 202OO PlP2 + + P^i + ""lPi+0030 +2 PlP + IV2P1 + ^lPa] + 2 + Pi +0120 Pi +0030
0ilo[ Cr2Pi
0210
0120
"l >
+ +
3OO
Pj
2OO
0110
0020
O ! (Tj
0300
Pi
0210
0"l
0"l
"l
and S3
we bear
1.
in
mind that
The
involve only cosines of even multiples of r, except those odd powers of a (i. e., where j is odd), and these involve only sines of odd jnultiples. The opposite is the case in
6ilt in
3
which are
the 8tjk of
of t.
coefficients of
If j is
j is even, the Bilk involve only sines of odd multiples odd, the diJt involve only cosines of even multiples.
If
2.
of X involve only cosines of even multiples of t and p2 and only sines of odd multiples of t in
,
cr2
It is seen, then, that in those terms of only cosines of even multiples of t enter
;
In the process
two types of non-periodic terms arise. First, those coming from the terms which involve the first multiple of Xr, and It is secondly, those coming from the terms which are independent of X.
important, therefore, to separate the various terms into three classes, (a) terms independent of X, (6) terms involving first multiple of Xt only, (c) terms involving multiples of Xt higher than the first.
ORBITS ABOUT
AN OBLATE SPHEROID.
form
t),
j
149
We
pl+62 Pa+03
where
flooi
Wi
(X, t)
+/, (X,
r)
^3+^4 0-3+^3 Pz =
e2 0i(X, t)
+Q
(X, r)
+& (0 +(7
1
(X,
t),
(X, r)
(X,
=0
lol
S'i
= t)
+7 a ]+20 +7 a +20
6
1 6]
O2o
7 76
1
O2O
7 76
1
The
terms are homogeneous of the first degree in Oj and y and consequently involve only terms which carry the first multiple of Xr; they are considered separately from other terms of the same character, because they carry the undetermined constant e2 as a factor. The solution for these terms has the form
/i
t ,
and
&
=F
(\,
T)+b ra
1
(r),
cr-G^fX,
r)+b ry
l
{r),
where
and
are periodic
multiple of Xt;
6j is
and involve only terms carrying the first a constant depending upon / and g and is distinct
x t
,
from
zero.
2
have the same properties as /t and g They are considered separately, since they do not carry c2 in their coefficients. Their solutions may be written
r)
The / (X,
and g2 (X,
t)
=F
(X,
r)+62 Ta
(r),
where
and
3
X, and f3 carries only cosines of while carries sines of odd multiples of r. The t, g3 only solution for these terms has the form
The /
(t)
even multiples of
^ = G3 (j)+b3 Ty,ij),
where
of
F and G
3
are periodic.
r)
The /4 (/cX,
and gt (/cX,
first.
t)
involve only terms which carry multiples The solution for these terms is periodic and
t
=F
is
(r)
(kX, t),
= (?
(kX, t).
The complete
p3
solution
i3)
therefore
i
3
<r 3
=A a \(r)+D [Ta + a (T)]+e [a + aTa (r)]+e [F (\r)-\-b Ta + [F*(\ T)+b Ta + [F (T)+b Ta (r)]+F U\ r), = A w 7i +D [t7 +73(T)]+e [7sM +ar7 (r)]+e (X, r) +6 rX (r)+b (K\ r)+b + + [G (X, y [G ry (T)]+G
3
(T)
(r)]
(r)]
l3)
(t)
(t)
[(?1
4 (r)]
(r)]
>
r).
(r),
p3 and a3 shall be periodic, it is of Ta3 (r) and t7 4 (t), and the coefficient of ra^r)
are periodic. In order that necessary and sufficient that the coefficient
(t),
(?4 (t)
and
zero;
whence
D =-b
i3)
-ae3
**"-
150
PERIODIC ORBITS.
Consequently the value of e2 is determined. In order to satisfy the initial = 0, which determines A a> by the conditions, we must have p3 =0 at t
equation
^a = 6
>
a4 (0)-63a 6 (0)
^
1
(0).
is
Thus
Pi
all
e3
=A
(3)
a3 = 4
63
T6
| *\(X, r)+F (X, r)+F (r)+FM, (X, r)+G (\r) + G (r)+GM (r)-|(?
2
i
3
>
.
r),
r).
be determined in satisfying the periodicity con4 It is obvious that this process of integraThe p3 and a3 have the same properties tion can be continued indefinitely. It is evident from the properties of that had been found for p and p2
3
The constant
will
so
the differential equations that these properties persist for p4 and <r4 and on indefinitely. The coefficient for *,_, in so far as it carries the first
,
,
Therefore the arbitrary constant always the same as for e2 be determined as to avoid can so ,_! always non-periodic terms of the type m The constant D of integration can always be deterto^t) and ty2 (t). mined so as to destroy non-periodic terms of the type 703(7) and tja (t). The constant A w can always be determined so as to satisfy the initial conditions. The analysis of the types of terms entering is the same as for
multiple of Xt,
is
.
the subscript
3.
have, therefore, a periodic solution with the period 2 kit which does not belong to the class of generating orbits from which we set out, for the
particle
We
its
orbit re-enters.
"
c^
and integrating, the solution contains five arbitrary (except for the restriction that X shall be rational) constants corresponding to the mean distance, the eccentricity, the inclination, the longitude of the node, and the epoch. One
more, a constant corresponding to the longitude of the perihelion, is necessary for a general solution of the differential equations. The periodic orbits
developed here are special in that they are the equatorial plane of the oblate spheroid.
all
Chapter V.
OSCILLATING SATELLITES ABOUT THE STRAIGHTLINE EQUILIBRIUM POINTS.
FIRST METHOD.*
Statement of Problem. Lagrange has shownf that if any two finite spherical bodies revolve about their common center of mass in circles, then there are three points in the line of these masses such that, if infinitesimal bodies be placed at them and projected so as to be instantaneously fixed relatively to the revolving system, they will always remain fixed relaThere are also collinear solutions in which tively to the revolving system. the ratios the mutual of distances of the three masses remain constant, only in but this chapter we shall consider only the case in which the distances In Chapter VII the more general case will be themselves are constant. treated. The three positions which the infinitesimal body may occupy are separated by the finite bodies; i. e., starting from minus infinity, the order is an equilibrium point, a finite body, an equilibrium point, the second finite body, and the third equilibrium point. It is not necessary that one of the three masses shall be infinitesimal, but we shall limit ourselves at present In Chapter VIII it will be shown that the problem can be to this case. n masses. There are also solutions in which the bodies lie to generalized at the vertices of an equilateral triangle, and the oscillations about these points will be treated by Dr. Buck in Chapter IX. If the sun, earth, and moon were so placed as to satisfy the conditions for a straight-line solution, and if the earth were between the sun and moon, then, as Laplace first pointed out, the moon would always be full, and either the sun or the moon would always be above the horizon of every observer. But these conditions would not be preserved unless the moon were in a If the position were one of complete stability position of stable equilibrium. from it, then it would perpetually disturbed were and the moon slightly oscillate about the point of equilibrium; if the position were one of complete instability, a slight disturbance of the moon would cause it to depart widely from the point of equilibrium. In the intermediate case of incomplete stability and also incomplete instability, the moon would either oscillate about the point of equilibrium, at least for some time, or it would speedily
75.
Read before the American Mathematical
Society,
June
American
Mathematical Society, vol. VII (1900), p. 12. The second method is givan in Chap. VI. tLagrange's Collected Works, vol. VI, pp. 229-324; Tisserand's Mecanique Celeste, Moulton's Introduction to Celestial Mechanics, Chap. 7.
I,
Chap.
151
8;
152
PERIODIC ORBITS.
depart from it, according to the character of the disturbance. If it is given such an initial displacement that it revolves in the vicinity of the point of equilibrium in an orbit closed relatively to the moving system, it is called an oscillating satellite; for, as seen from the earth, it oscillates in the neighbor-
hood of the equilibrium point in an apparently closed orbit. We shall consider here the motion of infinitesimal satellites oscillating in the vicinity of each
of the three collinear points of equilibrium. The literature of oscillating satellites
quite extensive, but in most of the papers the differential equations have been limited to their linear terms. In the discussion of the stability of a solution, it maybe justifiable
is
terms when the differential equations are but with these restrictions, which are inadmissible in power series; it is not possible to determine whether or not a treatment aiming at rigor, Poincare made a few remarks* upon this subject, periodic solutions exist. to the equations of Hill, which lack the parallactic relating his methods Burrau discovered several orbits in a special case from successive terms. Perchot and Mascart trial computations by mechanical quadratures.! treated the special case in which the finite masses are equal. J Sir George Darwin found examples of these orbits about two of the points of equilibrium in his celebrated memoir on Periodic Orbits.|| His methods, like those Under the assumption that the orbits of Burrau, were purely numerical. exist, Plummer gave a convenient literal development of expressions for the coordinates.! His method is simple, but apparently it is not easily extensAll of the writers mentioned ible to most of the more complicated cases. in the of motion of the finite masses. have treated the problem only plane It would be practically impossible to discover three-dimensional orbits by numerical processes, but there would be no difficulty in applying Plummer's method to infinitesimal satellites oscillating in three dimensions when the
to neglect all except the linear
infinite
finite
Differential Equations of Motion. Let us take the origin at the center of gravity of the system and refer the motion of the infinitesimal
76.
The
We will choose the and ij-axes in the plane a set of axes, rj, f of motion of the finite bodies, and suppose that they rotate in the direction of motion of the system, with the same angular velocity. The initial position of the axes will be determined so that the finite bodies continually lie on The distance between the finite bodies will be taken as the the -axis. unit of length, the sum of the masses as the unit of mass, and the unit of time will be chosen so that the Gaussian constant is unity. Let the masses
body
to
,
.
*Les Mithodes Nouvelles de la Mtcanique Cilesle, vol. I (1892), p. 159. \Astronomische NachrichUn, Nos. 3230, 3251 (1894). Xliulletin Astronomigue, vol. XII (1895), p. 329. Apparently their work is vitiated by an error in establishing the existence of the solutions, and their construction fails where they stopped.
\\Acta Malhematica, vol. XXI (1897), p. 99. \Monlhly Notices, Iibyal Astronomical Society,
vol.
LXIII
(1903), p. 436,
and
vol.
I.XIV
(1903), p. 98.
OSCILLATING SATELLITES
of the finite bodies
FIRST METHOD.
153
The units have been chosen so that p and p. the angular velocity of revolution is unity. The differential equations of motion of the infinitesimal body are then*
be
1
3?
2^-
W H^
body
is
dt
'
d\ "*" 9 df_dtf
,
d?_dU
'
d$
de
dt
dv
de
'
ar
(1)
>
ri^F&F+T+P,
Necessary and
tesimal
^=
v^-^+^+r,
a#
a)?
,= -p,
&=1-m-J
infini-
au
as
p
ar
(2)
may
be.
third of these equations are satisfied by rj = |* = 0, whatThe first equation has three solutions:! (a) one between
+
1
oo
and the
finite
.
mass
p, (b)
p,
and
(c)
one between
p and
oo
re;
ci; /a
Fig. 2.
r^+(3- M )r*+(
2
5 4
3-2^-/^-2/^-^ = 0,
2 2
2
(3)
= p -(7+m)p +(19+6/x)p -(24+13m)p +(12+14 m )p-7m 0, (c) r and where r is the distance from p to the equilibrium where p = 2
2
,
point.
(a)
The
v^
(4)
(6)
'2
(f)'-i(f)'-Kf)
,.""
(c)
2-V M
12
23X7'
"
Mechanics, p. 185. tSee Introduction to Celestial Mechanics, Art. 121, and especially Charlier's Die vol. II, pp. 102-111, for a detailed discussion.
*Moulton's Introduction
to Celestial
Mechamk
des Himmels,
154
PERIODIC ORBITS.
not be necessary Now give the to distinguish among them except in numerical computation. infinitesimal body a small displacement from one of these points, and a small velocity with respect to the finite masses such that
of
will
Suppose the coordinates of (a), (6), or (c) are = depending upon which point is in question. It
= +*',
= 0+
y',
= 0+
i
z',
^l-no-^ ^
dt
4t'
^_ n ^ dz'
dt
, ,
,2
(5)
dt'
The
dV _ 2 dy' _ dU _
(6)
^ = V($o+x'+m)
where Pi
77.
,
+?/'
+2",
r2
= vU-l-r-z'+M)
2
+2/'
+z",
and
x',
?/'
and
z'
2
.
P P
2
, .
It follows
from
the form of
Pl} P
and
common
are considerregion of convergence of the expansions of \/rx and 1/V2 ing only real values of x', y', and z', and consequently the conditions for
We
the convergence of the expansions of 1/r, and l/r2 as power series in and z' are respectively
x', y',
_,
t
2x'
/2
a;
+t/
/2
+2 ^
/2
2
,
|
2x
(7)
^^o-l+M^fe-l + M)
The
surfaces which
g"+y*+g* <
+
,
bound the regions of convergence of the expansions are obtained by replacing these inequalities by equalities. For the convergence of the expansion of 1/r, the equations of the bounding surfaces are
,
x'
+i/'
+ 2 +2a
'2
+M)x'+(| +M) = 0,
2
The first is the equation of the point occupied by the finite body 1 mThe second is the equation of a sphere whose center is at 1 m and whose radius is V2( +m) The convergence of the expansion of 1/r, holds
2
-
this sphere.
OSCILLATING SATELLITES
FIRST METHOD.
155
The equations
x' z'
2
) a;
= 0, = 0.
(9)
J
These are respectively the equations of the point occupied by the mass n and of a sphere whose center is at n and whose radius is V2( 1+m)
2
-
The convergence
of the expansion of l/r2 holds for all points within this the center. sphere except The distances from 1 n and n to the point (a) are respectively +/x and 1+m- The radii of the spheres which have been defined in (8) and
(9) are
V2
Since V~2
+m)
1> v^ (
1+m), the
sphere around
center.
n as a center is entirely outside of the one around n as a and 3 converge in the case of the 2 Consequently, the series x
P P
,
transformation to the point (a) for all points within the sphere whose center is at m and whose radius is V2 ( +m)> except the point itself. 2 The distances from 1 -^ and n to the point (b) are +n and ->/( l+/i) The radii of the two spheres are V2 times these distances, and hence they both include the point (6) in their interiors. In this case the two spheres
ju.
.
intersect unless m
is
small,
1 1
when one
n and
//
The
Since
distances from
(1
V2
m)
> V2
will be entirely within the other. to the point (c) are n and 1 p. ju), the sphere around n as a center
includes in
(c)
in its
,
the one around 1-ju as a center. interior, and everywhere within it, except at
its interior
The
1
latter includes
n,
the series
Plf
and
converge.
Parameters
and
transformations
x'
= xe',
and
8
y'
= yt',
z'
= ze'
(eVO),
(10)
where
e'
Then equations
become
{\
g -2(1+5) f =
T T ^+2{\ + 8)f
+ 8YP {x,
1
y\
2
*
e
)
f
=(l + 8T[X
+X +
z
2
+X
nn (e')
+F +
e'
^?
(1
+ 8YZ
=(l + 8yzP
(x,
y\
156
PERIODIC ORBITS.
We
shall
now
e,
e'
(see
13)
by
replacing
it
everywhere by
where
may
in its
neighborhood. When e ^ e', the differential equations belong to a purely mathematical problem; but when e = e' they belong to the physical problem. Since the value of e' has not been specified except that it is distinct from zero, the generalization may appear trivial, but the same method can be used where the parameter corresponding to e' does not have this arbitrary We have character, and where the device is of the highest importance. consider the to differential therefore equations
** _
2(1
+ 5) g=(l + 5)
[X
+X +
2
+X
-'+
1
],
^+
**
2(1 +
5)^=
(l
+ 8y[Y + Y
i
e+ 6+
.
+F - +
B
.
.
],
(12)
a+tyz =
a+5m+z
Equations
+z
e-
(1)
(i)'+$)"+(f)-**-.
the constant of integration. This integral was first given by Jacobi in Corn-pies Rendus de VAcademie des Sciences de Paris, vol. Ill, p. 59.
is
where
For equations
where
is
now a power
series in
e.
80.
(12)
The Symmetry Theorem. Let us consider the solution and suppose that at t = we have
x-c,,
of equations
dy d~ T
_ Cl ~
'
dz Tt
_ C" ~
_ Z _ dx _ n ~
~Tt~
'
that
t = 0.
is,
body
The
*=/i(t),
y=f
(r),
,
-,
-/t(r),
dx f 57 =/
i '
(T) '
^ =/
y=-y',
dy dr
dy
ff
*
(r) '
dz ft ( s =h{T) dr
Now
x = +x',
z=-z',
'
t=-t',
dz' ^ dr'
.
d& = dr
_^
"
'
dr'
j-d]/ ='^
=
dr'
dz dr
'
OSCILLATING SATELLITES
FIRST METHOD.
157
The equations
consequently,
in the
if
variables are precisely the same as in the old; the values of the dependent variables at r' = are
new
Ci
'
dr'
"
d T '~
'
~Z
~W~
0>
the solution
is
z'=/.(r'),
</-/,(/),
z'-/,(A
%-fM,
/,(T)
%-r,v),
%=s\w).
f[{T)
= = =
-f[(T')
=
=
-f[(-T),
2 (r)
+f
2 {r')
+f {-r),
2
t {T)
+f'*(r')
- +/J (.-*>
Therefore, if the infinitesimal body is projected perpendicularly from the x-axis, ihen x, dy/dr, and dz/dr are even functions of r, and dx/dr, y, and z are odd functions of t; that is, the orbit is geometrically symmetrical
it
is
81. Outline of Steps for Proving the Existence of Periodic Solutions of Equations (12). In (12) we put 6 = e = and find the general solutions special values of the constants of inteThen we change the initial values gration there are periodic solutions. of the dependent variables by small amounts and take d^O, e^O. The of the resulting equations.
For
equations are integrated as power series in 8 and e and in the increments to the initial values of the dependent variables. By 11, these parameters can be taken so small in numerical value that the solutions will converge
any preassigned range, and in particular for the periods of the periodic solutions obtained when e = 0. After having formed the solutions as power series in the parameters, the conditions are imposed that the solutions shall be periodic with the
for all t in
as the generating solutions have for e = 0. These conditions are that the orbit shall re-enter at the end of the period; or, in the case of the symmetrical orbits, that they shall cross the x-axis perpen-
same period
in r(not in
t)
These periodicity conditions are relations dicularly at the half period. imposed upon the initial values of the dependent variables and upon 5.
It
is
satisfied
by
expressing
e,
initial
series
values of the dependent variables as power converge for the modulus of e sufficiently small.
series in
158
82.
PERIODIC ORBITS.
General Solutions of Equations (12) for 8 = e = 0. On referring to equations (1) and the succeeding transformations, we find that equations
(12), for 5
= = 0, become
explicitly*
K+3$-r,-
(X-A)y,
(14)
'*-Z
where
--At.
AThe
general solution
is
<
+ r (o,3-
[(
^ +/x)2]j
+ ^_ 1+)li)2]
[(
(10)
first
is
independent of the
2
two, and
its
= c cosVAT+c sinVAT,
l
(16)
where
ct
and
first
c2 are
The
the constants of integration. two equations of (14) are linear and homogeneous, and they
coefficients.
have constant
To
x = Ke
y = Le*
T
,
(17)
where
The
[\
-a+2A)]K-2\L = 0,
K = L = 0,
may have
(18)
than
we must impose
the condition
As
We
X2 -(1+2A),
-2X
+2x,
shall
y-(i-A)
= \ +(2-A)\*+(l-A)(l+2A) = 0.
i
(19)
now
Its
discriminant
is
D = (2-AY-4(l-A)(l+2A) = (9A-8) A.
We
shall
(20)
(6),
show that
juf,
negative for each of the points (a), and therefore that is positive. From (15)
I is
1-A = 1--jmt -p = (1
At the points
(a), (6),
/*)[_!
^wJ+/i|_l
^bjsJ-
and
dx
(c)
we have
dn dx
dn dx
to Celestial
Mechanics, and Charlier's Mechanik des Himmels, vol. II, pp. 117-137. of Royal Astronomical Society, vol. LXII (1901).
OSCILLATING SATELLITES
It is
FIRST METHOD.
that
159
U in
1),
(1)
h
For the point
(a)
f =(!-,)(,-
^(r,-fr
is
<*>
,
rf
dr? _dr?_ dx dx
Hence, equations (22) and (23) give for this point
dU drf~
dU
dr?
(l- M )[<
4]
-M[r-r- i]'
/
Therefore, since the first two factors are positive while the third (21) becomes, for the equilibrium point (a),
is
negative,
l-^ = (l- M
)(l-
i)(l- j-)<0
(24)
Similarly, since only the second factor in the expression for negative, for the point (6) we find
r <0)
\A
is
_ _
1
(0)
dr
(0)
dx
(l- M )(ri
0>
dx
dr?
6U ~ dU m
dr\
(25)
i)=M(rr--L)'
(c)
l-^ =
(l-M)(l- ri)(l+7k)<0.
we have
r?=-l+r?,
=
dr? _ dr? dx dx
1kn),
dU
dlf
dU
(0)
'
dr'.
(26)
(l-n)(r?-^)
Then
1
-n(r?-
l-A= M
(l-^)(l-^!)
A
1
is
Since
negative because the third factor alone is negative. A is negative in every case for 0^m^0.5, it follows that two
of the roots of (19) are real and equal numerically but opposite in sign, and that the other two are conjugate pure imaginaries. Let the real roots
be
For each of these roots there <rV l. the imaginary particular solution (17), and the general solution is
+p and
is
x=
e e
aV:riT
+K
e-
aV
lT
+K
<?
+K
+L
jf
e~ pT , e~ pT (27)
,
y=L,
where, from (18),
T
aVzriT
+ L *****+ L
2
<f
W+1 + 2A] v
2a
3
-Kir
(28)
L=
3
+l^-^K =
+mK -=g*L
t
The constants
160
83. Periodic Solutions (14)
is
PERIODIC ORBITS.
when 5 = e = 0.
x = y = 0,
=c
cosv/ylT+c2 sinv/ZT,
,
The constants c, and c2 and the period of this solution being 2-k/Va. the t on which t depends, are not independent. We shall suppose t is = = taken so that ct 0, c2 c/VA. Then one of the periodic generating
solutions which
we have
to consider
is
x = y = 0,
z=-^=smVAr.
(14)
),
(29)
These will be called Orbits of Class A. Another periodic solution of the differential equations
x=
is
aV - 1T
+K
e-"
vzriT
,
y=
nV^T(K
we
aV
^ -K
T
e-'
rVZ:ir
= 0.
suppose, then
We
shall
zero.
suppose t is chosen so that the imaginary Let a/2 represent their real part. Then we
x = acos<TT,
y=
nasmar,
= 0,
(30)
the period being 2ir/a. These will be called Orbits of Class B. A third periodic solution will exist if <r and are commensurable.
vT
We
shall first
The
condition for commensurability is <r/VA=p/q, where p and q are positive Since aV l satisfies (19), we have from this relation integers.
A>-(2-A)A
The
(l-A)(l+2A)=0.
is
found to be
A_
4-i^i
and
0.5.
In order to establish the possibility of the. commensurability of a and VA, it is sufficient to show that p and q can be assigned such positive integral
values that the
from
(15) for
some n between
that the solutions of (3) are continuous functions of ii, and consequently A, as defined by (15), is a continuous function of mTherefore, if there are positive integral values of p and q such that
It is to
be observed
as defined by (15), there (31) lies in the range of values of r are infinitely many values of fi for which <r and \ A are commensurable.
the
defined
by
OSCILLATING SATELLITES
FIRST METHOD.
161
taking the upper sign in (31), we find that in order that A may be 2 and positive we must have /V<9/8. For values of p and q satisOn taking the lower fying these inequalities, A lies between +00 and 8/5. 2 2 in we must have in that A may be real and order < sign (31), 0<p /q 9/8 The values of A for these limits are unity and 8/5. Equation positive. takes the indeterminate form 0-^0 for p = q, but it is easily found (31) that the corresponding value of A is unity. For 0<p /<f <1, the value of
On
real
Kp
A, defined by (31) with the lower sign, is less than unity. On taking both the upper and lower signs, it follows that A takes values in every finite interval from 1 to 00 as p-/q 2 goes over all rational fractions from to 9/8. A < for each one of It was proved in the preceding article that 1 the three solution points (a), (6), and (c). Therefore for these points and 0<!/x<0.5, we have A>1. Consequently, there are infinitely many values of fi between and 0.5, such that a and VA are commensurable. When the commensurability relation is satisfied we have the periodic solution
x= y
K
Cj
s/
l
~lT
=nV=\{K = z cosvTt + c
l
+K,eaV=lT -K e
2
aVz:iT
e-
v=lT
),
sin-s/Jr.
We
can choose
so that cx = 0, and let a represent twice the real part of 1 and b twice the imaginary part of Then this solution i 2
t
K
z
7= sin VAt.
(32)
The period of this solution is P = 2irp/<x = 2Trq/VA. In this period x and y make p complete oscillations, and z makes q complete oscillations. These
will
Normal Form for the Differential Equations. We are about to prove that when e^O the initial values of x, y, z, and their derivatives
can be so determined, depending on
periods of (29) and (30) exist for all In this discussion to these solutions for e=0.
differential equations in a
first
that periodic solutions having the values of e sufficiently small, and reduce
e,
it is
it is
terms of the solutions as power series in 5, the initial values of the dependent variables. The linear terms of (12) are found by (14) to be
p -2(1 +
2
8)
= f- -(l + 8y(l+2A)x 0,
J
0+2(l+8)|-(l+a) (l-A)y-O,
||-r-
(1-H)
^=
0.
162
PERIODIC ORBITS.
general solution of the
first
The
two
of these equations
4
is
x = K,
^ v=ra+* r +X, e- ffv/rTa+i)r +#, ep(1+S)r +X Hl+S)T }+rn[K y = nV=i [K, e'^+^-K, e-"^
we
see that the transformation
- pc,+a)T
,
e"
(1+S)T
-# e~^^%
4
Therefore
X =(Wj +
+ (w + ^4),
3
l
= ^ ar
y
<T(l
+ dW=l(u -u )+ P (l + 6)(u -u ),
i
(33)
1
=n\^T(u -u )+m(u -u ),
2
3
+u + mp(l + 8)(u +u ),
i)
3
-^-<r{l + d)i Ul
dik,
dr
(r,ur-np)V=l
= V=l),
hj_tMu U-l-djMi,
-m(l + 8)[]e
2 ( TO <r
- np) V=T
wp)
i_
|
(1
+ 3) He
> '
(mp + wcr)
(34)
^-p(l + 5) W3
dt**,
= _ -
-n(l + 5)[]e
2 (mo-
^tPllt>Jtf
2
/...%.,
C[-4a;
z+y z+z ]e +
2
5
= |B[-2^+?/ +2 ]+2(7[2a;
2
2
2 -3a;2/ -3a;z
]e+
(35)
= 3Blxt/|+fCl-4^+2/ +
3
2/
?+
i
B
In
+ +
1
'l
M +
(0)4
_A
_(0)4
l-M
(0)6"
1 1
A*
'
~T
(0)5
'2
'2
the upper, middle, or lower signs are to be used according as solutions This transin the vicinity of the point (a), (6), or (c) are being treated. formation is always valid, since we find from (28) that
2(m(r-np)
It
is
= -(1+24)
^p!1 ^0,
op
(36)
not advantageous to transform the z-equation, nor ] and ] It follows from (1) that whenever the infinitesimal body is displaced from the xy-pl&ne, it is always subject to a component of acceleration toward Therefore it can not revolve in a closed orbit entirely on one this plane. of the Hence we may determine t so that 2 = when r = 0. side xy-plaxie. That is, without loss of generality we can take the initial value of z as zero.
OSCILLATING SATELLITES
Let the
t
FIRST METHOD.
163
initial
conditions be
t
u =a
where the
a,
+a
c
(i
= l, ...,4),
= 0,
H=c+ T
(XT
can be given such values that we shall have, for e = 0, either (29), (30), or (32). The a, and y are to be determined in terms of e so that the solutions shall remain periodic for e ^ 0.
and
all
the parameters
power
since
series in
it
a l}
we can more conveniently integrate them as 8, and the e; parameter 5 can be introduced in connection with t, always occurs in the combination (l 6)r; and the parameters at and y can be introduced when the constants of integration are
,
a4
7,
determined.
are independent of
<o)
are defined
by the equations
>
ff
(l
+ 5)vA3TM
=0)
d_ _ p(l + 5K =
0)
lOr +'(i+*)v=Tttf-o,
^
= 0.
+p(i+5)<=o,
(37)
^+r\l + 5)^0
The
<> =
(a,
2
C = (a
>
^=*
,
<=
(a,
4
+a
3)
e+" <1+5,r
ff<1+iS '
+=*
t
2)
uf = (a +a
4)
e~' +* r
(38)
These expressions can at once be expanded as power series in 5. The coefficients of higher powers of can be found by the usual process, but we shall not need them in proving the existence of periodic solutions.
85. Existence of Periodic Orbits of Class A.
For
=
,
the coordinates
in these orbits are given in (29). Therefore, since the determinant of the transformation (33), viz., A = 4(mp is distinct from l np) V w<r) (mo= = = = The general solutions 0. a2 a3 a i zero, it follows that in this case a
form
(aj
l
. . .
U = P,
!
a4 a4
a4
7, 5, e; r)
(t
= l,
4),
z=P ,(a
z'
7, 5, e; t),
7,
.
(39)
= Pt(*i
>
5, e;
t),
where the
a4
y,
5,
and
e,
and where
z'
r.
Since equations (34) do not involve r explicitly, sufficient conditions that the solutions (39) shall be periodic in t with the period 2t/VA are
(a lf
a4
7,
5,
"^5 )~^(ii
>
a <> 7,
5,
e;0)=0.
(40)
164
PERIODIC ORBITS.
last
These conditions are not all necessary, for it can be shown that the one a consequence of the first five. If we make the transformation
is
-+.,
the integral (13)
t
i
z=^^-smVAr+t,
may
be written
z'
= (c+y)
cos
VAr+f,
t ,
0,
C+y,
8, e)
=0.
find
(41)
0,
and we
from
!=2(C + 7 )^0.
It follows that (41)
e,
vlt
9lt
f,
for f' as
vt
a power
series in
a,,
.
y,
,
8,
= = 0.
i;
That
is, if
u l}
w4 and
= 2t/Va,
z'
of the first five equations. It follows from (38) that the explicit
forms of the
first
=o
+<Ji i
x
+ S)V=i 2L VA
[e
-l]+Q
1}
=a
+p+) -^
3
[e
^-lJ+eQ,,
= a,
[e
^-lJ+eQ,,
0=
4
=o
[e
,
^-l] + eQ
(42)
4 ,
vj(ffit)
sin2x(l + 5)+6(? 5
( ,
where the Q are power series in the a 7, 5, and e. The coefficients of a3 and a 4 are always distinct from zero, and the parts of the coefficients of Oj and Oj which are independent of 8 vanish only if <j/Va is an integer. We shall suppose at present that this ratio is not an integer, and that it is incommensurable. Part of the discussion becomes quite different when it is commensurable, and this case will be taken up when we discuss, in 96,
the question of the existence of orbits of Class C. 2 involve terms in z alone, It follows from (34) that since [ ] and { and since z does not vanish identically for a y 8 = 0, the Q, carry terms
\ x
in
alone.
The determinant
is
,
,
a 4 of the
. .
.
first
and
Oj
,
distinct
.
a4 in
a4 the product of the coefficients of a x Therefore these equations can be solved for
, ,
= eR
(y,8,t),
8,
(43)
where the
are
power
series in y,
and c. we have
e
When
VJ(l + 8 )
sin27r(l
+ 5) =
P(y,
8,
>.
(44)
OSCILLATING SATELLITES
FIRST METHOD.
165
The
solution of this equation gives us the periodic orbits in question. have the two arbitrary parameters 7 and S, and we shall show first that
We
we can not
give
for
7 as a
power series in e, vanishing with e. Suppose that 5 is neither zero nor an integer. Then equation (44) is not Now suppose 8 is satisfied by 7 = e = 0, and the solution can not be made. zero or an integer. Then the left member of (44) vanishes, and the equation 2 It is, in fact, divisible by e It is seen from (34) that x is divisible by e. and y do not enter in the last equation except in terms involving e as a factor, and since the a, defined in (43) enter only through x and y, the 2 It is part of eP coming from the first four equations is divisible by e
. .
member
which
is
member
y
e
2
is
.
multiplied
by
2
.
We
shall
is
now prove
been divided out there is left a term which which is distinct from zero.
independent of 7 and
and
Terms
the
a,
in
in the right
member
both through
defined in (43),
(34).
The terms
directly in the integration of the last equation of obtained in the former way involve as a factor and
and
depend upon a and p; the terms entering in the latter way carry C as a 2 factor. Hence, if the coefficient of e is to be identically zero, the parts
involving B and C as factors separately must be zero. that the part involving C as a factor is distinct from zero.
We
shall verify
in (44), so far as it
is
independent of
and
7, is
(45)
-8TOaF sin3V7(1+
The solution of this equation satisfying the conditions z = 0,
2
^ )r
z[
= 0,
at 7
= 0,
is
z-=
64ffa+- sinVT(1+i)r
"
wMfW' TCOsVJa+>)T
(46)
'
*(!9-*<>
,
[-8ATO+---]'
(47)
2 Hence, after division by e there is a term independent of both 7 and e, and the solution for 7 as a power series in e, vanishing with e, does not exist. That is, periodic solutions of the type in question do not exist.
166
PERIODIC ORBITS.
us give 7 an arbitrary value and attempt to solve equations a 4 and 5 as power series in c, vanishing with e. Since (42) for di , c is arbitrary, we may put y equal to zero without loss of generality. Or,
let
. .
.
Now
more conveniently for the construction of the periodic solution, we may But for give 7 such a value that z' = c/VA for r = 0, whatever e may be. in we shall is in included c. The first four simplicity writing suppose that y can be solved for a a in 8 terms of and 4 equations t, and the results substituted in the last. The result differs from that above only in the terms multiplied by e, and, as before, we find that the lowest term in e alone has e2 as a factor. There is a linear term in 8 alone whose coefficient is 2nc/VA. a 4 have been eliminated by means Therefore, after a,
t
,
. . .
of the first four equations, the last equation can be solved for 8 as power 2 series in e, the term of lowest degree being e When this result is substituted
.
in the solutions of the first four equations, we as power series in e, vanishing with e. That
of (42)
have a
is,
when 7 =
=e
t
p(e),
*ep,()
series in
e.
(i
= l,...,4),
(48)
When
(e; t)
=z
t
+eq(e;
t),
u = eq
t
(i
= l,
4),
(49)
where q and the q are power series in e and are periodic in t with the These series converge for |el sufficiently small. The circle period 2ir/V~A. of convergence is determined by the singularities which are present in the differential equations (34), which are introduced in forming (39), and which
are introduced in the solution of (42). Since the right members of (49) and are for all e converge periodic sufficiently small, the coefficient of each
|
|
power of
86.
separately is periodic.
Properties of Solutions of Class A. It will now be shown that the orbits under consideration are re-entrant after one revolution, and that they cross the x-axis perpendicularly.
Some
Let us find the orbits whose periods are 2i>t/Va, v being an integer. form equations analogous to (42) simply by replacing 2ir by 2vir. The determinant of the linear terms in a, a4 is distinct from zero unless is an We exclude this case here and treat it when we conva/VA integer. sider orbits of Class C. Therefore the first four equations can be solved for a 4 as power series in 7, 5, and e. On substituting the results in a, the last equation, we find, as before, that the solution can not be made for 7, taking 8 arbitrary, but that the solution for 5 as a power series in e is o 4 and 8 as power series in e, unique. That is, the solution for a,
We
vanishing with
is
unique.
Hence
there
is
We
have shown
also
OSCILLATING SATELLITES
FIRST METHOD.
167
that for a given value of e there is one orbit of Class A having the period 2t/Va. Since an orbit of period 2ir/V~A has also the period 2vk/Va, the latter are included in the former. It follows, therefore, from the uniqueness of both orbits for a given e, and from the fact that, for e = 0, they re-enter after the period 2-k/\Ta, that all orbits of this class re-enter after
a single revolution. Let us now suppose that, at t = 0, we have dx/dr = that the orbit crosses the x-axis perpendicularly at r = 0.
y=z = 0;
that
is,
It follows
from
Mi(0)-,(0)=0,
whence
ox =
Oj
,
w (0)-w (0)=0;
3 4
(50)
a3
=a
to (39) will
suppose in this way will be included in The tions, under which it was always permissible to put y equal to zero. orbits obtained with these initial conditions will be symmetrical with respect to the x-axis. Therefore necessary and sufficient conditions for periodicity
are that the infinitesimal
half period.
We may
7=
now be power series in a lt a3 y, 8, on the start, for orbits that may be found those found from more general initial condi,
body
VI
=y=z=o.
imply that, at t = t/VA,
2
It follows
from
Uiu = 0,
2
u3
1/ 4
= 0,
= 0.
(51)
= ai =a
3
[e
P(l+)-=:
^-e ^-e
-pa+S)-
VA
]+eQ[,
[e
^]+eQ'
3
(52)
,
0=
where Q[
,
V2(l+8)
sin27r(l+5)+e$,
series in a x
,
Q' 3
a,
8,
and
, ,
e.
It is easy to
a3 and 5 as see that these equations are solvable uniquely for a x series in e, vanishing with e. Therefore, for a given value of e there
power
is
one,
Since for a of these symmetrical periodic orbits of this class. orbit for unrestricted initial congiven value of e there is but one periodic cross the x-axis perpendicularly at ditions, it follows that all orbits of Class
168
PERIODIC ORBITS.
87. Direct Construction of the Solutions for Class A.
In the practical
A
3
it is
most convenient
are
The
members
i
= (l+2A)x,
|B[-2x 2
t
+2/
+2
],
Y = (l-A)y,
l
Y =SBxy,
Z =3Bxz,
t
3, F =f C[-4x>y+y*+yz>],
3 3
(53)
Z,=-Az,
the A, B, and
Z^Ci-Wz+tfz+z ],..,
The
x, y, z,
being constants which are defined in (15) and (35). and S can be expanded uniquely as series of the form
t
2=2x,(tV,
where the x
t
y=7iy
(r),
(j)e
z= - T sin\/jT + 2
z,(t)
,(r)e',
5=2 5
e\
(54)
are each periodic with the period 2ir/\r A. On substituting these expressions in (12) and making use of (53), we obtain a series of sets of equations for the determination of the xt y z S
(r),
and
The
coefficients of
in (54)
must
<^-2<^-(l + 2A)x
=lB[-2xl+yl+zl},
|p+2^-(l-
A)y
z
.
-3Bx y
t
(55)
^+A
But x = y = 0,
satisfies
,
Zl
= 3Bx
c/yJ~A sin
vTt.
r 2ir/\ A and
Therefore the solution of (55) which and z shall be periodic with the period
= z[ (0) = 0,
>
is
-3c
Xl
2
,
"^
3(l+34)c
2
>
2
2
(56)
zi
VT(l-74 + 184
4>1,
e
2
Since in
all
cases
The
coefficients of
the coefficients are always finite. in (54) must satisfy the differential equations
2
^f-2^-(l+24)*
= 0,
2
|&+2^-(l_4)fc-0,
1
(57)
^+Az
Upon
= -2A6 z,+3Bx
and x
2
x
,
zB
+^Czl
1?
**+A,-[ +AZ*~
9Cc (l-34 + 144 )c smVjT + L" SA^] 84^(T+^4KT^74+"184" ) 9B 2 ( l+34)c3 3Cc l Q r-r 8in 6VAt +f |_84 (l-74 + 184 )~ 84J
o VJc8 rio% 2
27
-|
'
OSCILLATING SATELLITES
FIRST METHOD.
169
The
solution of this equation will not be periodic unless we impose the condition that the coefficient of sinV At shall vanish. This condition is
satisfied
by
reject this
= 0, but this leads to the trivial solution x^y=z=0. If we solution, we may use the condition for the determination of 8
c
2
.
After
satisfied, the periodic solution of (57), having the period = 0, is 2-K/yfA and fulfilling the conditions that z2 = z' 2 =0 at t
it
has been
x>.
-n
</*
,_
",
3
o,
)c
,2
z,
= 3c r3B (l+3A) - C
2 i 64 A t[ {-7A
(58)
+ lSA
][
3sinVJ'- S D3VI
'
*}
In this manner the construction of the periodic solution can be continued We shall prove this statement by induction, and as far as may be desired. at the same time we shall derive certain general properties of the solution which are satisfied by the terms already computed. xn -i y z n -\ 82 5_i y n -i 2 ouppose x
,
. . .
',
have been determined and that they have the following properties:
1.
2.
3.
4.
5.
6.
The x2J and y2j are identically zero, j an integer. The z2J+1 are identically zero, j an integer. The function x2J+1 is a sum of cosines of even multiples of VTt, and the highest multiple is 2j+2. The function y2j+l is a sum of sines of even multiples of V~At, and the highest multiple is 2j+2. The function z2j is a sum of sines of odd multiples of vTt, and the highest multiple is 2j+l. The 52j+l are zero.
, , ,
It will
now be shown that these properties hold for xn y zn and The terms xn yn zn and b n satisfy the differential equations
, , ,
d?x n
-?
~ 2 dy, ~ ^+^A)xn = Pn {x d7
n n
Vj
ht y' u
8,),
^+2^-(l-A)y =Q
^- +Az
where
y'j8 n -j
n
(xJ
(59)
= - 2A
8n
+R
(Xj,
y
.
zn 5,),
Q,
It is seen
and from
and
z' } in-,.
8 S (j=l, are polynomials in x, n-1). in the and involve and that Q products y' x\ only (12) } If n is even, these terms are zero by properties 1 and 6, for
,
. .
n-j must
be odd.
But
if
is
170
PERIODIC ORBITS.
shall
We
now prove
that
P=0
<;
if
is
even.
The
general term of
P
(
is
T
where
X,
, .
*=<[
.
.
<;
X,
,
. . .
*%
;
<;<
,
<? *
;
%
>
'
\,
Mi
P*.
Pi
2
,
p2
9i
and
&
l
are
all
Since the
dj
+ 5)
&
01+9,^2.
(61)
:
Mi+
first
+/**' is
of the
equation of (12)
+/Zk
is
,
.
(b)
(c)
n"+
X,
,
. .
\K
\[
>
1.
(d)
(e)
(f)
Xf
x
by property
6.
2.
by property
mA+
+m*X k +MiXi+ +m*'X^+m!' Xf+ + + +M* Mi+ +P2& M1+ +/v + M?+
+/v,x,+p,g
+ n"-l = n
+
because the term of degree m^' +m+Mi+ -\-Pk?+Pi in x, y, and z has e as a factor to a degree one less than this sum, the terms z, introduce e to the degree n \, and the sum
' '
^+
"
of the exponents of
must equal
n.
or an
There are two sub-cases, according as Mi+ +m k is an even integer is odd integer. When /x + an even +m integer, the following
t
(7)
(5)
( e)
an even number of odd /*i nK X is an even +Mk x Mi X,+ integer by (c) and (a). K >K- is an even integer by (a) and (c). +n' M1X1+ +/i"X is an even integer by (d). m[X'+ P1Q1+P2Q2 is an even integer by (e).
There
is
It follows
(a)
,
. . . ,
+
is
+m
odd.
(e)
left
is
member
of (f )
identically zero
even, and in general is not identically zero +Mk is an odd integer. Then Suppose now that Mi+ M* (a') There is an odd number of odd Mi +m K is an odd integer by (c) and (a'). Mi Xi-p(/3')
> ,
is
odd.
(5), and (e) respectively. and hence again odd, every T is identically zero if n is even, and in general not identically zero if n is odd. It follows that P is identically zero if n is even, and in general is not zero if n is odd. The treatment of the general term of Q can be made in a similar way. The only differences are that in (a) and (7) the sums are individually odd instead of even. But since (f) involves their sum, the result is that Q n is identically zero if n is even, and in general is not zero if n is odd.
The
properties (7'),
left
(5'),
and
(e')
are the
same
as (7),
Therefore the
member
of
(f) is
OSCILLATING SATELLITES
FIRST METHOD.
171
general term in Rn has the form of (60), where the subscripts and exponents satisfy the relations:
The
(a")
(b")
/v is an even integer because the right member of the third equation of (12) is a function of y 2 ' +M" is an odd integer because the right member of juf
fi'
.
this
equation involves only odd powers of z. and (f") are the same as (c), (d), (e), and
(f)
respectively.
+M* is an even integer. Then (a"), (0"), (7"), (6"), as Therefore in (e") (a), (18), (7), (5), and (e) respectively. this case the left member of (f") is an even integer. It is shown similarly that the same result is true when ju t +/x K is an odd integer. Therefore,
Suppose Mi+
are the
and
same
is identically zero if
n
is
is odd,
and in general
is even.
The
Case
discussion
now
naturally divides into two cases, viz., where odd. shall treat them separately.
is
We
We
shall
multiples of
term
(60).
VAt, The z
1* 1
is 1.
sum
of sines of
odd
w+
Xi
are
all
is
VAt;
therefore
the product x
of property 3
x *'
1
sum
of cosines of
even multiples.
Because
an argument,
M
1
which occurs
K
1
is
(X 1
+ 1)+
+H (K+l)=Hi\+
K
+M X*+M +
,
+Mk.
(62)
Similarly,
(a")
it
y"'' is
sum
of cosines of
Mi
(>W+1)+
+/vOC+l)=Mi^+
and
+Mk'^'+Mi+
,
+/va
(63)
From
of
properties 5
of
M""
is
sum
of sines
odd multiples
VAt, the
*f(xr+i)+
+*&te+i)-Mfxf+
+mI"C+/x;'+
+/v<. (64)
taking the product of these three sets of terms, we find that of sines of odd multiples of VAt, the highest multiple being
On
is
sum
N = n \+ + M!+
1
+mAc+mX+ +M+Mi+
of the
'
+Mk'Ak'+m!X'+
+Mk'X'"
+/V+m('+
as
(f),
+/V'-
By
(f"),
which
is
same form
we have
q
l
N = n+l-(p
For those terms
in
+p
q2 ).
which q^q^
0,
we
N = n+1.
172
PERIODIC ORBITS.
*3s_23&_(i+2;i)srw-0 cLt ut
w+ i2
{1
- A)y =0
>
(67)
>
+C|1 ain(n+l)v'Zr, +Cr;+1 sin(2i+l)VZr+ where the C +1 are known constants which depend upon the coefficients of The solution of these equations which the terms with lower subscripts. = at r = 0, is satisfies the periodicity conditions and z n = z' n
j
^ +Az =
n
n)
xn = yn = 0,
zn
&n=
<n>
2^ Ac
'
=y
sin
Vlr+y sin3VAT+
4/
tf
+7S+1 sin(n+l)VJT,
T2J+1
[(2j
+ l)*-lM
7,
+D 4
>
2 /'
(68)
Z/4i(i+l)v'^
sum of From
cosines of
Case II.
We
shall
even multiples of
1
VJr,
properties
and 3
it
follows that
a/*'
of
(62).
From From
properties
1, 4,
and
(a) it
sum
by
VAt,
(b) it
properties 2,
5,
and
'
is
also a
sum
A,
a^,,
Therefore
(62),
is
sum
(f)
from
(63),
(64),
and
Qn
is
sum
of sines of
even multiples of
R = 0.
n B
jVIt+
n
+A% cos(n+l)VAT,
l
^+2^-(l-A)y 7
d?z n
= B?8in2VAT+
+BUsin(n+l)VAT,
(69)
+Bfjsm2jVAT+
di
+Az=
2v/Zc5 B sinv/ZT.
OSCILLATING SATELLITES
FIRST METHOD.
173
The
the conditions z
and
C+
b)
c4
cos2VZt +
(4"j 1
2/
=/?<"> sin 2
\/Jr+
a
+(3
sin2 JV2t+
+/t! lS in(n+l)VJT,
z
= 8 = 0,
2(8i
4 2
=
2
AT 1+2A
2
(70)
'
+ 2i -lM
4
Bm
- +4 j v^
2(8./
+2f-lM -(8/-lM +
2
-(8i -l)A
2
+l
(>->
( /-l,
)
)
- [2 (2j + 1) A + 1] flff ft
n+
Since
^1
>1
any
integral
j.
obvious that in practice it is not necessary to refer to the differThe most convenient method to follow is ential equations at each step. to substitute as many terms of (54) in the right members of (12) as will be required in carrying the computation to the desired order in e, and to arrange the results as power series in e of the form
It
is
P +P e +
2
ie
+P +
n e"
,
and similar series for the other equations. From the P, Q n and Rn the Af, Bf, and Cf can be computed sequentially with respect to n without
explicit reference to the left
members of the differential equations. The are given by (68) and (70). The whole process
may
be desired.
It will
, ,
be observed
i+1 computations have been carried, x } y } and z, carry c 1 We shall prove that this is a as a factor and that 8j carries c as a factor.
general property.
Suppose
for
it
is
j = n.
5j
The
2/,'5 n _,.
It
and consider the question by equations (59). In Pn follows from the assumed properties of the y,
.
.
and
in
Now as a factor. carry c" follows from the assumed properties of x Jt y } carries c as a factor to the power
Q
iv= Ml (x
1
z}
and
b}
+i)+
+ Mx (x
,
+i)+m;(x;+i)+
+/v(x;+i)
.
+M
It follows
,
;
(x;
+i)+
+^(K"+i)+p
q1 +pi qi
from
(f)
is
general.
174
PERIODIC ORBITS.
In order to obtain the coordinates in the physical problem, we must Then and c replace by e' and multiply x, y, and z by e' [equations (10)] occur in every term in x', y', and z' to the same degree, and are equivalent
'
.
That is, without loss of generality we may put c equal to a single parameter. to unity, and the value of e' will determine the dimensions of the orbit. Or, if we put e' equal to unity, c will determine the dimensions of the orbit. It
from these results that the coordinates and 8 are expansible as power series in c, and the solutions could have been derived in this way without the introduction of e and e', but the discussion would have been less simple. The explicit expressions for the periodic solution, so far as they have been worked out in (29), (56), and (58), are*
follows
s' = 0'+r X
U+
~ 3J?
1.4,4 (1+24)
*Q+ 3 ^
4A(l-7^ + 18^
cosZVAT e + -cos2x/ZtV+--)
W+[ vi<X-7A +
=
{
:
18A<)
2
k2VJr]+
(71)
[= sinVZT-Je'+Ot'
2
~ c 3 sin VJt " sin VJr + 644p Ll-74 ] [ y+ [r^S+i^t + 9 = n/ T 3B'(1-3A + UA rl * 16A L(l+24)(l-74 + 18^ J
5
2
,s
sums of cosines and sines respectively of even of and since z' is a sum of sines of odd multiples of VAt, it VAt, multiples follows that x' and y' are periodic with half the period of z' Since the relations
Since x' and y' are
.
z'(t)=x'(-t),
y'(r)=-y'(-r),
z'(r)
= -z'(-r)
was
are satisfied, the orbits are symmetrical with respect to the x-axis, as shown in the existence proof.
Then
=y'(T+r),
z'(r)
=x'(T+r),
y'{r)
= -z'(T+r).
x'y'-plsaie.
Simi-
x'(r)=x'(T-r),
y'(r)
= -y'(T-r),
z'(r)=z'(T -r),
the orbits are symmetrical with respect to the xV-plane. It follows from the form of (71) that a change of the sign of equivalent to changing t by a half period.
e'
is
period of the solutions in r is 2ir/VA, but last equation of (10) that in the time t the period is
The
it
P_
27r(l
VI
+ S) =
"
2ir
VJY
x', y',
_
z'
3fi (l-3vi
rl
)
\
J
no '^
,
*The
and
OSCILLATING SATELLITES
FIRST METHOD.
175
It is found from (71) that the equation of the projection of the orbit on the x'y'-plane is, up to terms of the fourth degree in e',
16A 2
r
2
3fle"
(l+3A)
This
a;
l_
^4A(l+2A)J
ellipse
T ^ Ay
.
_
at x'
9flV
'
"(1-7A+18A
is
7 U6)
whose center
y'
0,
'
3(l+3A)e'
The equation
approximately
of
2
2
3t'
)'
2 2
4A(1-7A + 18A
VX(1-7A + 18A
,-,. *
)"
U
g/V-plane
is
y
This
is
-6Bz'
(1-74 + 17.
VavV
1
t'
'
(75)
the equation of a figure-of-eight curve with its center at the origin, touching the y'-axis at no other point, and having two other intersections with the z'-axis.
The equation
approximately
of
the
x'z'-plane
is
4A
The
orbit
is
Ll+2A^1-7A + 18A
is
/J
is
UD;
at
9B(1-AW<
'"(1+2 A) (1-7 A + 18 A )'
2
n
to the orbit,
which
<t n /A belongs
the infinite branches having been introduced in eliminating r. It is at the vertex of this parabola that the orbit has the double intersection with the x'-axis.
It
is
In
all
cases
seen from (35) that B is positive for the point (a). intersect the x'-axis between (a) and the finite mass /x.
Hence these
orbits
It follows from (35) and the second equation of (4) that is negative for the point (6), at least if n is small. Hence these orbits intersect the x'-axis between the point (6) and the finite mass /* Similarly, those orbits
near
(c)
between
(c)
and the
finite
body
ix.
vertices of these parabolas are the ends of the ellipses whose axes are given in (74). It is seen from the expressions for the coordinates of the
The
centers of the ellipses and the ends of the parabolas, that the distance of the parabolas to the centers of the ellipses is
+3(l+3A)e'
It follows
2
*
4AU-7A + 18A
from the signs of
the points of equilibrium near which they
176
PERIODIC ORBITS.
the properties which have been derived it is possible to infer the In a general way they have the shape geometric character of these orbits. of one of the two handles being situated on one side of the handles ice-tongs,
From
and the other symmetrically on the other side of this plane. In the case of the is where they cross the -axis. The points (a) and (b) they open toward the finite mass m, and in the case of /*. (c), toward the finite mass 1 In Fig. 3 an orbit of each class is shown. No attempt has been made to represent them to scale for any particular case, but the figures show their general positions and the directions of motion in them. The curves are drawn on elliptic cylinders to make the figures as clear as possible.
of the
ij-plane,
(C)
zM
(b)
JL
(air * s
Fig.
3.
89. Application of Jacobi's Integral to the Orbits of Class A. The original differential equations admit the integral (13), which holds for all orbits, and therefore in particular for the periodic orbits which we are disIt has already been seen that it plays an important role in the cussing. of the existence of the periodic solutions when we start from general proof initial conditions, and we shall now show that it is almost equally important
We
simple problem, a
The
explicit
new and valuable use to which integrals form of the integral (13) is
2
may
be put.
F= (S) 2+
+(l) -( 1 + 5
+B(-2x +Sy
2 i
+3z*)xe+
= constant.
(77)
OSCILLATING SATELLITES
If
FIRST METHOD.
177
we
and
substitute the solutions (49), or rather the equivalent series for x, y, z, their derivatives, and the series for 5 in (77), and then arrange as a
series in
e,
power
we have
.
F=F
(x
,<jfy
. . .
+Fl (x
. . . ,
x lf
,
,)*+
.
~\~^n\X
x,
o2
6)e
is
-f-
= constant.
(78)
Since this
is
an identity
in
e,
each coefficient
a constant.
It is seen
We
from
shall
now
of the general
term
of (78).
(77) that
Fn
will contain
^,&W dr
rfr
fyfy=i, dr dr
%>%=>, dr dr
and
(60).
of x, y, and z that the terms of the first three types are zero unless n is even. Now consider the term of type (60). All the properties of its exponents and subscripts are the same as when it belonged to P except that in the relation (f ) the 1 in the left member must be replaced by 2. Hence
It follows
we
term
is
is
,
even.
involves only even powers of the y, and z, it is a of even multiples of Vat. It follows from the relations
Since
sum
(a),
of cosines
.
(f)
(the last
is
n+2.
Hence
we may
(79)
Since
is
identically a constant,
we have
D'
ln)
=
.
constant,
. .
ia) D' 2
=
(
=2)^=0.
a ;\
.
. .
(80)
The yf
,
a?; 0?, ....,?; Suppose all the a t p Jf 7j up to <- p;~ yf-" have been computed and are known to be accurate. Equations (80) can then be used in two ways, as we shall show. First, they test the accuracy of *~ v the computation of the a^~ ft"~ \ y \ for these quantities must have And secondly, we can be satisfied. such values that the equations shall compute the e^" ", ftj~ from equations of the type of (69), and then find the y and 5 directly from (80) without referring to the differential equations
quantities
. .
D'\
2)
2)
yf.
l)
obvious and we need to consider further only the second. We are working under the hypothesis that n is even. Therefore the af and the /3 ;B) are identically zero. It is seen from (77) that the only terms which can introduce the y" and 5 are
The
first
use
is
Tr It
+ 2Az z + 2As
z l-
81 )
178
PERIODIC ORBITS.
the form of z given in (68),
From
we have
S^ =C v/J
2Az
zn
7 )+[7 >+3t
'" '"
)
^ COs2VZt+
,
'
'
'
=c
n)
Vj{Ti
+ [-7r +Tr]cos2Vl-r+
)
(82)
+[-Y?;-i+7r;+1 ]cos2iviY
- 7 S-xCos(n+2)
2^5 n Zo = c 5-c
2 2
vTt}+
...
5 n cos2\^i"T.
D' W D'
6U 5n
(/-a,
...,),
where
o<l>
Hj
>
Z)i"| 2
(0)
are
'
known
(n_2)
IJ
a,
a,"
";
li
'
>
The
, . . .
last
n)
,
gives 5 uniquely.
The
7b+1
=_
Jy n+2
cVJn
_
7u-t
7
_1 T
m
(n)
(j-l)cVA
"
73
^ 7s
D 2cVA
V"2 4 ^ VT
'
'v
(b)
'V' -I- D
(n)
(84)
by
All the constants are uniquely determined except yf*, which is z' This condition gives n shall be zero at t=0.
denned
SOf+lhfi,.
< )
(85)
Thus we see that in orbits of Class A we can suppress the z-equation, " if we wish, and compute the 7 from the integral; or, we may use the a check on as the computations. integral, step by step,
OSCILLATING SATELLITES
90.
FIRST METHOD.
179
periodic orbits of
it is
this class
of Class A.
No
It is clear that
practically
suppose impossible to discover them by numerical experiment. We the ratio of the finite masses is ten to one, or 1 ^ = 10/11, /* = 1/11. Then, in the computation of the coefficients of the series for the solutions in the
shall
and
(c),
180
91. Construction of
PERIODIC ORBITS.
a Prescribed Orbit of Class A. Suppose the masses Then a periodic orbit of Class A for the of the finite bodies are given. infinitesimal body may be defined (1) by the place it crosses the rr'-axis, (2) by the y' or z'-component of velocity with which it crosses, (3) by the greatest value of the y' or z'-coordinate, (4) by the constant of the Jacobian
It is understood, of course, that these integral, and (5) by the period. various quantities are arbitrary only within such limits that the series for the coordinates converge.
the Jacobian constant C and the period have definite values The increments to these values and all the other depending only upon /z.
For
e'
defining quantities enumerated above can be developed as power series in These series are odd or even, depending upon which e', vanishing with e'. other quantity is taken as defining the orbit. If s represents any of these
quantities,
we can
write
s
= s e'+s
1
.
2 e'
+S e"+
3
where the
If s
of
e'
are constants which depend upon ju alone. is assigned numerically the inversion of this series gives e'. This value substituted in (71) gives the desired orbit. Thus, the methods which
coefficients s^Sjj, s 3
, .
.
have been developed not only prove the existence of the periodic orbits and give convenient processes for constructing them and testing the accuracy of the computations, but they furnish a ready means of finding any particular orbit that may be desired.
92. Existence of Orbits of Class B.
For
orbits are given by (30). Therefore a t = a 2 v* 0, a, = a 4 = in (38). Sufficient conditions that the solutions (39) shall be periodic with the period 27r/<r are
c=0
'2t\
,,
rt
....
(87)
z(^f)If
*(0)-Q,
z'(^)-z'(0)=0.
we
let
w =(a
1
+a )e +ff ^=7'+v
1
1 ,
u,
= a,+v
=0+t,
may
/
be written
F[(a
+ a,)e" = Tr T
5,
1)
(<h
+ **)
e~f
vCrrT
+ft
a,,
a 3 +v3
a4 +iu,
5,
f,
7 + f,
e]
-F[a, + aj, a
+ o,,
a4
0, y,
e]
= 0.
(88)
OSCILLATING SATELLITES
This equation
is satisfied,
FIRST METHOD.
vt
181
find
at t
2ir/a,
by
= f= f=
0;
and we
from
F and
=
(19)
and
(28)
we have
n=
Therefore
v2 v3
a2 +l
+ 2A
2a
o*+l-A
2
nV
,
2A
quently, for r
5, e,
,
= 2t/(t,
,
vt
f',
Conse1), which is always positive. can be solved for in terms of a (88) i\ uniquely y, = = = = and this solution vanishes for v2 v 3 v 4 f f = 0.
f ,
= a +n {A
2
'
at r = 27r/<r, Hence, if we impose the condition that v2 = w3 = 4 = f = f = = = the equation v at r 2x/<r will be satisfied. Therefore the first equation is redundant, and it will be suppressed. It will be shown that the orbits of Class B lie in the xy-plane. It follows from the form of the last equation of (34) and the initial values of z and z' that P5 and Pe contain 7 as a factor. Therefore the last two equations of (88) contain 7 as a factor. The explicit form of the next to the last one is
' i>
l
P.Qf)
When VA/a
(89)
is not an integer the only solution of this equation, vanishing with the parameters in terms of which the solution is made, is 7 = 0. The case where VA/a is commensurable will be considered in connection with the orbits of Class C. Therefore z = 0, and the orbits are plane curves.
Necessary and
solutions of Class
= 2,
3, 4.
The
explicit
forms
0=(a
(a 1
a4
8,
e),
0= 0=
a3
+2T ^ a+5)
[e
_2,r
-lj+e^K,
-l]+eQ
4
,a 4
a4
5,
e),
a4
" <1+5>
(a,,
5,
e).
have three equations to satisfy and five arbitrary parameters, The parameter a enters only in the combination besides e, at our disposal. ai + a,, and since a is as yet subject only to the condition that it shall not We may determine t which enters in the vanish, we may let it absorb a x
x
t
.
We
182
definition of
fulfilled
r,
PERIODIC ORBITS.
so that at t
0, a condition which is which the coordinates have continuous condition becomes
we
shall
have
x'
in
derivatives.
By
(33) this
oV
which we
in
e,
ai +p(a3
.
ai )=0,
may
We now
and
5 is
a4
and
as
power
series
,
vanishing with
The determinant
T
of the linear
2x
terms in a3
a4
27ra
V^l[Y
-l][e-
-l],
(91)
Therefore equations (90) have a unique solution for a3 a 4 and o as power series in t, vanishing with e. When these results are substituted in the first four equations of (39), the latter become power series in e which are periodic in t with the period 2ir/a. It will now be shown that all orbits of this class are symmetrical with We choose t so that we have y = x' at t = 0. respect to the x-axis.
which
,
is
not zero.
Therefore
u,
it
that
Oj
follows from equations (33) and from the initial values of the = o4 0, a 3 Necessary and sufficient conditions that these
.
dx -
d,T
r =y = 0,
at
!-
IT
fl-
it follows
from
at
t = 7r/o\
(33) that these equations are equivalent to The explicit expressions for the latter become
,,
ut = ut
= a,
r *<!+> v^i
[e
[e
s 5,
t),
=o
The determinant
a3
(92)
3
-e
'
]+eQ
of the
(<h,
5,
c).
of
the coefficients
in
and
5 is
2Ta
Vi
[e
a
\
>
which is not zero. Therefore equations (92) can be solved uniquely for a and 5 as power series in e, vanishing with e. Since, for a given value of e, there is but one unrestricted orbit of this class, and since there is also one which is symmetrical with respect to the x-axis, it follows that all orbits
3
of this class are symmetrical with respect to the x-axis. The orbits of this class all re-enter after one revolution, for
if
we impose
is it
the conditions that they re-enter after v revolutions, we find the solution unique. Since it includes those which re-enter after one revolution, follows that all orbits of this class re-enter after precisely one revolution.
OSCILLATING SATELLITES
FIRST METHOD.
It has
183
been
shown that
u = 2 wu
t
(t
= l,
,4),
5= 2
Sje
(93)
where the
the
5j
and where
We
loss of generality
we
.
can, put, at r
= 0,
(94)
u = ui = a
It follows
=^,
we have
u = uA
3
from
(33)
and
(93) that
also
x= 2
i=0
*,',
y= 2 y
i=0
l
t
(95)
Upon
coefficients
substituting (93) in (34), arranging as power series in e, and equating of corresponding powers of e, we obtain a series of sets of
differential equations
of
are defined
by
UT ar
(IT
a,T
The
and the
solution of these equations which satisfies the periodicity conditions initial conditions is
aV
uT=^e
From
T
,
u?=e-^
we
get
T
,
uf = Q,
<-0.
(96)
=a
cos (xt,
e
na
sin or.
(97)
The terms
are defined
by
__ aV-lu,
>
+ aV-ld lUl
r
8
-r-
4(wi<r
_ np)
^f
>
2 (mp+n<T)
d<
^r u
du? _
dr
3
_ _ aV
(o,
_ 3mB[-2xl+yl] _
,
3Bx y
(98)
>
m _ 3nB[-2xl + yl]
4(mo--wp)
"
3Bx y
2(mp+n<r)
^C x tP
dr
<
184
PERIODIC ORBITS.
The
conditions that the solutions of these equations shall be periodic with the period 2ir/a are that the right members shall be periodic with this period, that the right member of the first equation shall not contain the ^ term eaV lT and that the right member of the second equation shall not
,
.
The first condition is satisfied, and on referring contain the term e -<r%/=rTr to (97) we see that the second and third conditions can be satisfied only
by
5,
Therefore
we put
5,
equal to zero.
Upon
du
dr
become
3
in full
_
,
to
1
- (n + 2) co s 2 or]
np)
wBa
sin 2 <tt
8(m<r
/
V^l
,
4(mp-\-n<r)
3 nBa sin 2 or
2
du^
dr
U*a) =
a,
3
4(mp+n<r)
(99)
du
dr
_
,
(n
3nJ?a sin2<rr
4(rap+n<r)
3 nBa sin 2 o r
2
8(m<r
2 2
(,)
np)
du
dr
np)
4(mp+n<r)
The
solution of the
first
w =c
(.)
(.)
cog 2
_ ^AZ7 &<
s i n 2 (rr,
100)
where
of
t,
the arbitrary constant of integration. Upon substituting this expression in the first of (99) and equating coefficients of like functions
cj" is
we
get
+ 3mBa (n -2)
2
'
8(ni(rnp)<T
2
11
m=
mBa\n +2)
8(ni(r
nBa?
' (
np)<r
2
.
2(mp+n<r)(r
(101)
b 12
m=
_ mBa {n +2)
2
nBa
4(ra<r
np)a
4 (rap+no-)<r
from the form of (99) that the solution of the second equation can be obtained from that of the first by changing the sign of V l.
It follows
Therefore
a$-a$,
The
ag-ag,
6--iff.
(102)
u^-cfV+aff+aS? cos2rT+6^
sin2<7r,
OSCILLATING SATELLITES
FIRST METHOD.
5
185
where, because of the periodicity condition, c" in the differential equations that
a
a)
= 0.
We
find
by substitution
= +<= +
r*^42
(n
2
-2)
8(manp)p
(n
2
u32
+2) P
+2
,
ZnBtfa
(mp+n<r) (4 o
2
+p )
2
'
(103)
7>>_^<u-_L U 32 T
"42
/_.
4(mo--np)(4oat t
. .
.
(n 2)o_ ITTTl 2
l
3nBa
2 2
+p ) ^4 (rap + no-)(4o- +p )
2 2V
"T
Since x = a and
?/
=
,
e,
we have x
(0)
(0)
= 0.
find
Mi
rw
(104)
Then the
x,
expressions for
Xj
and
become, by (33)
cos
err
+2 [ag + ag]
cos 2rr,
(105)
=+2n[< +<+<
)
+a< 2 sin<rT+2[n6g+m&g]
]
sin2<rr,
6\
= 0.
In order to see how the construction goes in general, the process must be continued one step further. The differential equations which define the -next terms are
OB
dul
i{
2
,
raC[2xg-3x
(rao-
j/ g]
3(7,4x
4 (rap
j/
-j/^
np)
V
x
1
+ no-)
^_ +<rv^Twf
=-
rV^=T<V
3m^[-2x
2 (rao-
np)
ra(7[2x^-3x
(rao-
t/g]
np)
} ?
(106)
P^a
2(ma
t ...
~_ ^ np)
2(mp+na)
3
~t...
_
du ? j-.i
nC{2x
-3x
np
yl ]
3C{*x y
-y
3
\
>
ma
?/
i(mp+na)
1
3nB[-2x
2 (rao-
3g]z y +g,y
2 (rap
np)
+ no-)
2
]
nC[2xg-3x y
3C|4x y
4 (rap
-^[
manp
+ no-)
186
PERIODIC ORBITS.
In order that the solutions of these equations shall be periodic with the period 2x/<r, the right members must be periodic with this period, the right member of the first equation must not contain the term e ay/ iT and the ay/ ^riT right member of the second equation must not contain the term e~
The
condition is satisfied as the equations stand. The expression 0) x x lt y ylt x ylt and Xj?/ en ters through w{ The sum of its coefficients must be put equal to zero; this condition determines o 2 by the
first
e <rv^ir
equation
82
m_
+mb%)]
_ 3B[-2n(a{+a
2
)+n(a{ I
+<4
2
u
)
+ (n6gf+w6g)]
2(mp+na)<r
(107)
8 {ma
np) a
32
(rap
+ no)
<r
This disposes of all the arbitraries appearing in the equations, and the third condition remains to be satisfied. Upon comparing the first and second
equations of (106), we see that the signs of the [ in the second are opposite the signs of the corresponding terms in the first, and that corresponding have the same sign in the two equations. It is observed that the
] {
\
are
of
sums of cosines of multiples of or, while the are sums of sines the same arguments. Since 8 2 enters in the second equation with the
\ \
\ ,
sign opposite to that in the first, it follows, as a consequence of the properties of [ ] and \ that the condition on the second equation is satisfied
62
first.
.
We now
y
l
,
.
x v ^; proceed to the general term. Suppose x have been and that they have been found to ?/_! computed,
,
2. 3.
4.
5.
The Xj are sums of cosines of multiples of gt. The y are sums of sines of multiples of or. The highest multiple of <jt in x and y, isj + l. The is an even function of y. The is an odd function of y.
} }
[
] \ \
The equations
dr
are
2{m<T-np)V-l
"
2(rap+n<r)
*
dr
+<rv
^r ur = -*v=\ <>s- 2
2(m<r-np)
V-l
2(rap+n<r)
;108)
dr
2(ma
n[J + 2(ni(T
np)
2 (rap
+ n<r)
d< +
dr
w= u*
. . .
np)
+2
U
(rap
[
+ no-)
]
where the
5,
and
OSCILLATING SATELLITES
FIRST METHOD.
[ ]
187
of cosines of
1,
{
J
2,
is
4,
and 5 that
is
sum
multiples of
sum
The
*-<
The exponents and
(a)
<- <;
(109)
Mi+
q
is
+/v
or
1,
is
of 4.
(b)
since
(c)
mA+
+hkK+h'iK+
+Mi+
+M* + Mi+
is
ocfc
+HK>K>+pq
+/V
1=".
sum
by
\
1.
of
Those factors w
for
which ii s are odd are sums of sines of multiples of or. even number of such factors is a sum of cosines of
The product
ar.
of
an
It follows that
is
or.
is
N = n (\+l)+
l
+n
1 (\ K +l)+n'
(\[+l)+
+n'AK'+l).
It follows
from
(c)
that
N = v + l-pq = v + l
Therefore
It
[ ]
when
= 0.
(110)
is
sum
in
being v+1.
can be shown
a similar
way that
is
sum
of sines of multiples
in the
form
du?_ ay/ZZiu? =
aa v
e'
oV=\
-i
"+l
[v+i
J,, 00
e-
aV
^T8
v
-i
mv
du*dr
-v+\
{
[v+l
(111)
- P u ?= +n\X Al L(=o
+pt*fand
<"
cosi<TT
du?
where the
-n[s
are
all
^cosiVtt]
Af
known
real constants.
188
PERIODIC ORBITS.
In order that the solution of equations (111) shall be periodic, we must aV=lT in the first equation, and impose the conditions that the coefficient of e
It is easily seen that the of e -,rV=TT in the second equation, shall be zero. two conditions are identical, and they uniquely determine 5 by the equation VI a
( ">
ab
_ A R<">
form
(112)
The
e ffV=TT
+ c? e~ aV^ + 2
T
v+l
<
cos jar -
V-
_ v+1
1
^=2
2 &# sinjVr,
(113)
<> =
<+ <
2
.7=1
< = <+ 2
where e? and
5
<>
2 fc^sinjVr, + J=l
CJ"
Upon
io
i^20
a 30
"t"
a 40
~~
I
"^0
>
a%=+a%=-[mA?-B?\
(i=2,
. .
u(>
__ ~
ijW
__ u~
(jAj -\-Bj
a(f-l)
OSCILLATING SATELLITES
FIRST METHOD.
189
The
and
arbitraries
t/
at r
= 0.
Upon
x
i/,
-2[a>2*j---2
Zl
[<+<]
cos<rr
+2
's
[<+<>]
tan jar.
cosjVr,
(116)
=2n 2 [<+a]
sincrr+2 "s
[n6J?+m&$
These expressions have the properties 1 and 2. Since x x y y also have these properties, the induction is complete and x depends only upon cosines of multiples of or, and y upon sines of the same argument. The
,
,
geneous of degree v+1 in a, and that 8 p+l is homogeneous of degree v+1 in a. Consequently, it follows from (10) that the actual coordinates, x' and y' carry in each term of their expansions e' and a to the same degree. That is, e' and a are equivalent to a single parameter, and we may put one of them, say a, equal to unity without any loss of generality. Then the
,
become
cosar
/2
+
y'
>
+a32 )cos2<n.]6
)
.
;
= -n
[
sin ar]
e'
+2 +n
[
(117)
2 )
Since
is
is
these orbits
a positive constant, it follows that in all cases the motion in For small values of e' the orbits are approxiretrograde.
in
mately
x'
elliptical
i/'-axes.
and
The
it
F(x,
y,
j^>
is
^)=F +F
in
e+F
=constant.
is
an identity
e,
each Fj separately
a constant.
It
sum
,
2. being j = These coefficients involve v-\-2) separately is zero. 1, cosjVt(j constitutes a check on all their and and the a\" vanishing b^ linearly, the a\i\ bl^ f rom the beginning of the computation to the step under
. . .
of cosines of multiples of ar, the highest multiple is an identity in r, the coefficient of each
consideration.
190
95.
PERIODIC ORBITS.
Numerical Example
of Orbits of Class B.
In Darwin's memoir,
cited at the beginning of this chapter, there are a few examples of orbits of In all his this class in the vicinity of the equilibrium points (a) and (6).
computations Darwin took one finite mass ten times that of the other. To be able to compare the results of this analysis with his orbits, we shall apply the formulas for the same ratio of the masses. This was the ratio used in the computation of 90, and the first part of the table given there can be used here.
Upon making
(104), (107),
and
use of the preceding computations and (36), (101), (103), (117), we get the following table of results:
Coefficient.
OSCILLATING SATELLITES
96.
FIRST METHOD.
191
the Existence of Orbits of Class C. For e = the equations of these orbits are given in (32). It follows from these equations and (33) that in this case
On
ai
If Wi
= ^ a + o77=r
>
a2 =
7T
a~ o
t 2V-1
/
'
a3 = a 4 = 0.
(119)
the
initial
conditions are
=a
+a
Ui
w2 = a2 + a2,
^3
= 0,,
w4 = a 4
0,
z'
= c+y,
(120)
may
,
a*, 7,
5,
e;
t)
t), t),
(i
= l, ...
,4),
a4
7,
8,
5,
e; e;
(121)
0.4,
7,
where the
Pi} P
and
are
power
series in e^
ait
X, d,
and
e.
The
VA
>
5* =
a4
a4
S,
,
a
7, 7,
5,
are
0= (a + o )[e+< l+*^^
1 1
0=(ai +a )[ei
e), e),
5,
=a
=a
+P(1
2JTJJ
3 [e
^-^^(a!,
-pa +6)
2tj>
1
,
a4 a4
.
7,
y,
. .
e),
4 [e
0=
^--Tj-^--vsin2T?(l
l]+Q(a + 8)+Q
(122)
,
8,
e),
(o lJ
a4
y,
8,
e),
0=
last
8,
e).
two equations
by c+y.
7.
is
divided out.
termined constant
distinct
We
from zero, it follows from the integral (13) that the last equation There remain five equations of (122) is redundant and can be suppressed. a4 and 5 as power series in e, vanishing whose solutions for a
x
,
. . . ,
with
e,
will
now be
considered.
appear in the course of the work that we shall need all of the terms of the first degree in e, and all of those of the second degree which are not periodic. We integrate equations (34) as power series in e, introducing a 4 by means of the initial 5 in the combination (1 + 8) r, and ^ conditions (120). It will be found that only the first power of 8 is needed, and then only in terms independent of e; elsewhere it will be omitted. Likewise only the first powers of a3 and a 4 will be needed, and therefore the higher powers will be omitted. Since a and a 2 enter only in the combinations e^ + a! and a2 + Oj, we may omit them for brevity until the end, using simply a and a2 and then restore them where they are needed.
It will
,
.
.
192
PERIODIC ORBITS.
The terms
tlf
of degree zero in
1 ,
(123)
VA(l + 8)
sinVT(l + 6)r,
are defined
where
=V
1.
The terms
d
by the equations
\,
^-aiuT=-mEi[
d
}-F\ }-F\
\,
Wd
pU '=-nE[ )+F\
^+aiu?=+mEi[
where
\,
]+F{
\,
(124)
^+A
E_
2
[ ]
Zi
=+SBx&,
+3
'
p=
2<rir
A{m<r
np)
2
2(mp + no-)'
=-[2(2-n )a
a2
+2 (2 - mm) a, a
+(2+n )a^4
2ff,r
+^-cos2v^4"r
T
+2 (2+ mm) a, a
+2(2-mm')a
3 2
e +"-'
(
+ 2(2+wm')a
=
{ }
a3 e
- <7l+p)T
a4 e
- ff '- p>T
],
4
\nia\ e
ialT
-nia\ e2
"r
+ (m+m) ai a e^'+^+tm-mK a
(
(ff
'- p,r
-(ni-m)a
aiT
a3 e'-'ri+p)T -(ni+m)a2 at e
- ,rt - ( )T
'
\.
The solutions of (124) are the respective complementary functions, alT #<V T Kfc- pr plus terms of the same character as their K?e K?e, , , ,
right members.
equation, the coefficients of these terms are respectively the coefficients of the right members written in the order given in (125), omitting the term cos2vAJr, divided by
first
ai,
+<rt,
2
3<n,
/
+P,
p,
2ai-\-p,
2ai
p.
gives rise to
.
+
,
mEa&
2WA-?)A
C0S2
V Ar +
/
mEi&
(M^)Vl
2ai
n , sin2VAT
The corresponding
respectively
<rt,
+3(rt,
ai,
2ai+p,
,
p,
+p,
p,
(+m
t'c
/2.4)cos2\/ AT are
)A
C0S 2v^T
,-r
mEic
77-3
{AA
)VA
TT
sln 2
V A r.
,r
OSCILLATING SATELLITES
FIRST METHOD.
193
In the solution of the third and fourth equations it is unnecessary to compute the terms which carry a and a 4 as factors. Omitting these terms,
3
the divisors for the third equation are respectively p, and the terms coming from ( nEc2 /2A) cos2 VAt are
2<ri
p,
2aip,
nEpc
2 2
2(A+ P )A
+
'
2<ri+p,
nEpc
2<ri+p,
2(4A+p
)A
2 cos wa "
n.r-r.
VJt v "'
is
nEc
2
==
(4A+p )VA
sin2V At.
The
z,=
+ +
xTat+L
sin
\TXt
+
2(<t
SB did + 2<tVa)VA
'3
(,("+
V7)tT
zBckci
,(-<t+VX)(t
Bead
(p+VAI)T
2(<r-2crVT) V~A
SBcaii
2
2(/+2?pVT)
(-p+VAt)T
2
VT
(126)
,(p-VAt)T
2(p
.
-2ipVA)VA
zBa 2ci
(_<r_V^4)[T
2(a
_|_
2
-2a\rA) V~A
3 Bead -2ipV~J)V~A
2(<t
+ 2(tVA)VA
3 Bead
2(p
(_p_\
At)T
2(p
+2ipVA)\/A
=j^e
of integration K[ K[ l) l) conditions that u[ at t shall vanish u[ = L< and the conditions =0. by ,(()) Lf z((0)
, , ,
The constants
x>
are determined
0,
It is necessary to compute all non-periodic terms of u u u 2 and z which a are of the second degree in e and which are independent of a, and 5. The right members of the differential equations involve
,
.
[}
|
!
w=
<2,
lB[-2x
+y
t/
!-
+z
zl
}+2C[2xl-3x
yQ -3x
2
zl],
= 3B-;xo2/l +a{
(127)
:
+ic\-44ih+ti+v&.
The
quantities xA
x^u^ + iV + uy+u'",
-O+m-i
).
In order to get all the non-periodic parts of the solutions at this step, the terms of the differential equations which are non-periodic, that is, which carry p in the exponential, must be retained; in the first and second
T respectively must be retained; and in equations the terms in e" and e~ the z-equation the terms in cos VA t and sin VJt must be retained, for these periodic terms give rise to terms in the solution which are multiplied
atT
by
r,
periodic.
194
PERIODIC ORBITS.
The
t
2ir7) *-
2 trH
rj
= T are
- u (T) - u = z(T) -
(0)
+
z (0)
K (D-wf
)
(0)]
[z,
(i
=i
,4),
,
= [z
(D - z
[z2
(0)] +
(T) - z, (0)] e
(128)
+
By means
u?\T)
(!T)-z2 (0)]e'+
find explicitly
+2xp
" -u(0) = (a +a )[e +J a+8,J - 1], w(l+i>< 1], u?{T) uf (0) = (a +a )[e1 1
u?(T)-v?>(0) = a3 [e
l-l]
2
uf{T) - M
(0)
=a
4 [<f
"*-l]
(129)
2,(T)- 2.(0)=
vz( 1 + 8)
i
1
sin2T(l
+ g);
l
u a)
D -w
H)
(0)
2mEi{2
a3
+
,
2mEi{ 2
mm)
F(ni
T
m)] a^
~\
r e + 2t pt
/>
2<n-p
f
[
jL
il *J
2mJgt'(2
+
,
2mEi{ 2
mni)
_l
,
~i
J'
mW/'Pi
w,
^;
w <o/ays <u;-
[-2m-gt(2-mm)-^(m + w)]aia
2<n'
,
j
[
r +2*pt
_ j"l
+ /Mm')
2ai
F(ni
p
m)]^^
-2
+{
,
+ 2mEi{2
mni )
*pt
_ j"|
u? (T )
u'
(0)
2nE(2-ri )a a t
l
+
M a>
+ tf)-nFi\a\
2<rt+p
( 4
2nEc2
s
^ + p )pjL
l
2 lr.+ X e ^7_il
J'
^ _w
a)/
)
[ng(2+n^-wft]^
1 )
2nE(2-n*)a a1
2nE&
6flca4
(4^ + p')pJ
\ r 6 -**v
I r
, "i
'-I'
GBca3
r+2'p^
,i
-*>
. -i
OSCILLATING SATELLITES
FIRST METHOD.
195
2nE(2-n
p
)a a2
1
[nE (2
+ n - nFi]a\
2
2ai-\-p
2nE&
-n )^,
2
2ai
+p
p
~
2ai-p
-ll X \e*"* l J
__
| J
(4A + p )p\\p
2
~2ai + p)
+2m^([-^ (2 + n2) p
[
- wF
2
2crt
2w g(2-n2 )a a
1
!!
[ttff(2
+n
1
-nA']a
n \J
(2
2ai
1
+p
~ n*) q
p
] [at
_
.,,
2n.Ec
2 2
fai
~(4A + p )pJlp"
en
r **>$ e
2(
rl-pjL
2m *nE \ ^ lE{2
I
[-tt.E(2
+ n )-nFt]a _
2
2?lEc
2(7i-p
~1JA + pTp\\p
i
a,
2ai
+ p)
*^ x IV L
2
,
ll J
2
2nE(2-n
p
)a l a i
[n.E(2
+n
2 )
-nFtja
il
2<rt
+p
"'J
2n.Ec
1 fat
a,
"(4^ + P
L
1|y^4 e
2
2nff (2
2
- n )^^
2
2<n
2
+p
2
[-?i^(2+n )-nFt]a _
2<n'-p
2
2n.Ec
fai
aa
2ai
JjA+TTpj {p^~
L
+ p]
x [Y L
,
^_
p
i] -J
2
_ nF A\-nE{2 + n*)-nFi]a\
2<n
p
2
2nE(2-n
,
)a ai
l
[n.E(2+n
2 )
- nFi]al _
r [nff(2
2
nEc
2cn
]
2
+p
1
2
(4^+p )pJ
fa,
lp
,
r^ _ ^J-T ^ 2ai-p\\a2
l
+ n )-nFt]a
2<rt+p
2nE(2-n )a
p
a2
[-n.E(2-f tt )-ni^]a;
2<ri
_ "
196
PERIODIC ORBITS.
where
L = 4 mETi f [2m(2+n )-4nF] + [-2n PE(2 + n + 4n<rF] 3a 4<7 +p 1 _ 4i?(2-w )(mp + tt(r) ] 2 m n ETi I [~4m^(2+n + 2nF]
a
2
2mn[2(r^(2
2
4tr
+p
J
2 ) ,
1
i 2 2
3<r
+ pF] + 2n[pE(2+n )-<rF} } 4 4<r H-p +p j l imECTi (2 w +nFTi [ 4(2-n )(mp + na) "
_
2m[2o-ff(2+w
2
ff
2 )
(130)
<rp
] "J"
~5 B
2nFCTi
B
2
ZtfFCTi 2B
'
SmBETi (4A-a )A
2
nEFTi(mp + na)
apA
mECTi
nFCTi
AB
4AB
There are corresponding equations for u^(T) u(0) and z2 (T) z2 (0). The third and fourth equations of (128) can be solved for a3 and a 4 as power series in e, a, a2 and 8, vanishing with e. The explicit results are
, ,
\
3
2ai
,
p
2
2nEc
2 2
" "
i
2<n+p
(4^+p )pJ
%
(131)
-{
When
fifth
-7?ff(2+?t )+flFt]a
_ 2nE(2-n )a a
l
2ol+p
i
.
p
2 2
,
these expressions for a3 and a 4 are substituted in the equations of (128), we obtain
+2,r',5
1
second, and
= (a, + a = (a
n=
2
)[e
2
'-l]+(a
5,
+ a )[(a + a )(a + a )L + c
1 1
iVf 1
]e
2
+
+
4-a
)(o 2
+ a )L +c
2
I
,
M]
1
,
vT(T+5y
Sln27r(?5
ainoJ^t
65(ai4-ai)(ai+a2 )c
A
A -a
2
1
^{2(4-n )(a
+ a )(a + a
1
)c-f|}e
OSCILLATING SATELLITES
FIRST METHOD.
5,
197
After removing the factor c from the last equation, solving for substituting the result in the first two equations, we have
and
(a i
(133)
where
L= Jji "
2*jri
9B
2
" M=
2wJW 2irpi
_i_
)
j
/ ma + SV^-^ 2 V 4A l^
(134)
,
np
2
2 "*"
ff
! 4 A + p /J 4i
\~|
16-4'
Equations (133) can not be solved for a and vanishing with e, unless
t
a 2 as
power
series in
e,
[a 1 a 2
L+c M] = 0,
2
a2
[a 1 ai
L+c M] = 0.
2
x
(135)
solution of these equations is 0,-0,-0, and with this determination of a and a 2 equations (133) are uniquely solvable for a and a 2 as power series
Y
,
One
with e. In this case the generating solution reduces to the form of that of Class A. But the orbits of Class A heretofore treated were This restriction was not those for which V~A and a are incommensurable. necessary in order to prove that orbits exist which re-enter after one revolution, but it was not certain that there are not others re-entering only after
in e, vanishing
many
revolutions.
The uniqueness
a single revolution.
= a = 0,
2
proves that
assumption was made and this the that c is zero, suppression of the last permitted equation of (122). If we had assumed that a is distinct from zero, we could have suppressed the first equation of (122); and solving in a different order, we should finally have arrived at two equations corresponding to (133) containing (c + y) as a factor. The equations would have been found solvable after imposing the condition c = 0, and we should have arrived at the con-
At the beginning
distinct from
0.
(136)
This equation defines c when a and a 2 have been given arbitrary values. If the orbits are to be real, 0, and o 2 must be conjugate complex quantities. must be Under these circumstances their product is positive, and L and opposite in sign in order that c shall be real. After the condition (136) has been applied, equations (133) become
0=(rt
+ a,)[a a
l
+rt 2 a -f a^lL
1
e [a,
a2 a2
e]+ e]+
0=(a +a
2
2 )[a
a2
+a
+ a a lL+e[a
1
198
PERIODIC ORBITS.
Since the terms of these equations which are independent of t are identical, except for the non-vanishing factors a^a^ and flj+Oj it follows that if
,
one
is
solved for
a,
by e. 6 which must be equal to zero in order that the equation may be solved This term involves the for a, as a power series in e, vanishing with t.
divisible
coefficients of
e
3
3
result substituted in the other, the latter becomes After dividing out e, there is a term independent of Oj and
and the
out.
of a 3
,
in the original solutions (122), since e has been divided enter from lower powers of e through the elimination terms Likewise,
,
a4 and 5. It is not possible to construct these terms without an unreasonable amount of work. But we see from the way in which they originate that they are homogeneous of the fourth degree in a and a 2 Unless one or the other of these constants is absent, their ratio is determined by this constant term set equal to zero. If one is absent, the only solution is the other set equal to zero, which throws us back on Class A, which
2
.
has been already completely treated. Suppose both constants are present and that their ratio is determined. Since they must be conjugate in order that the orbit may be real, the solution for the ratio has the
cti
form
'
a,
clear that only for special values of the coefficients, which might never be possible in the problem, could the solution for the ratio have this form.
It
is
The complexity of
tinct
the problem is such that no further attempt will be made here to determine whether there exist solutions of Class C which are dis-
and Class B.
attempt is made to construct the periodic solution of which are the terms independent of e, no difficulty will be encountered until (32) 2 in e are reached. Then it will be found that equations (135) the terms
If the
is
somewhat
less laborious.
Chapter
VI.
OSCILLATING SATELLITES.
SECOND METHOD.*
treated in this chapter is the same as that considered in the preceding, but the method employed is quite different. In this particular question the preceding method is somewhat
in other problems where the same general style of can be used it is much less so. analysis There is a definite physical situation for which the analysis is to be developed. One of its principal features is that the periodic orbits form a continuous series from those of zero dimensions at the points of equilibrium, and as they vary in dimensions the periods undergo corresponding changes. In the analysis of Chapter V the dimensions were controlled by means of the scale factor t, and the varying periods were properly secured by the introduction of 8 and its subsequent determination in terms of e' As e' approached zero the orbits approached zero dimensions and the period approached the value which corresponds to 5 = 0. In the present treatment no parameters corresponding to e' and 5 are Instead, we introduce a parameter X by means of /* = ju +X, employed. where Mo 1S kept fixed in numerical value while X is a parameter in terms of which the solutions are expressed. Periodic solutions are found for all X whose moduli are sufficiently small, but only those solutions belong to the The dimensions of the physical orbit physical problem for which X = /i /x That is, we depends upon this value of X, and its period depends upon /x find a family of periodic solutions having a constant period depending upon It is because of Ho but only one of them belongs to the physical problem. this fact that it is not necessary to make the period variable and dependent upon the parameter in terms of which the solutions are developed.
. .
97. Outline of
Method.
The problem
of
We shall start from equations (6) the which will not be needed. The right accents omitting Chapter V, in involve the the last of these members equations parameter n explicitly two terms of U, and implicitly through ^ and r2 which depend upon rf deWe shall make the transformation fined in (4).
98.
The
Differential Equations.
M = Mo+^>
(1)
but it which
is
not necessary to do so
in all places,
both
explicit
and
implicit, in
this
parameter occurs
*The problem of oscillating satellites was first treated by the author by the methods of this chapter. However, the two methods were reported on simultaneously in the paper referred to at the beginning of Chapter V.
200
PERIODIC ORBITS.
For simplicity the transformation will be made where it appears explicitly in U, and elsewhere p will be supposed to retain its original given value, which is regarded as a fixed constant. This particular generalization of the parameter p is not the only possible one, and the series obtained differ according to the particular generalization made, but when the conditions
for
convergence are satisfied their sums are identical in t. After the transformation (1), equations (6) of Chapter
V become
)x=:
P.fo y\ z\
i
X),
X), X),
)y = yP
z = zP
(x, (x,
y\ z\ y\ z\
(2)
=
l
&
-4-
-&-
[2x
2
-3xy
.
-Sxz*-]
+
(3) Mo
(0 )5
+2/
<B)i
+2 ]+
_Mo
^.(0)4
, '
A' = ^
?
L_i__L,
.(0)3
I
1J
j.(0)3
l^Mo + R = + + r
first,
r _
'
i I
\lk f
,
'
^.(0)5
the signs in
being the
The regions of convergence of vicinity of (a), (6), or (c) are in question. and are the same those found in 77. as P, 2 precisely
in the
normal form so
far as the
the right
members
y = n V^TCK",
z
{
e-"
),
(4)
= c cosv^4^+c
. .
.
sin\/^<,
c2 are arbitrary
n
,
where where
c,
and
i
,
+ V
<r
a
J
+p
and
and
a,
+ (2-.4 V+(l+2,l
2<r
)(l-,4 )=0;
and where
also
(5)
2p
is
= +(U =
+ U + (V
2)
+ll
),
u,
-u
),
+V-M
r
(m,-m,)
,)
+ wi
oA>
(G)
(m,-h
),
y'=
(, +
(!+)>
OSCILLATING SATELLITES
SECOND METHOD.
201
(2) to
- a.iu, = + m Pl fa
2(w
(To
V\
2
2
>
>
^oPo)
2
V
,
*)_,
1
u'+r iu = -
"v^fe*/
2(m
o-o
,*
WoPo)
*1 _
1
^_ PoMs
4
=_
WqP,
(x,
O"o
y\ z\
X)
2 (m
Wo Po)
(x, + yP 2 Wo
t
y\ z\ X)
(7)
+W
2
(To)
2
w4 + Po u =
(*>
cr
2
>
g2 >
x)
2 (m
po)
z"+A where i = V
z
i
+ zP,{x,
if,
z\ X),
in (3).
Equations
i 2
i
(2)
admit
(2x
,i
-y -z )+A'(2x -y -z
2
,i
)\
(8)
+B
which holds for
x, y,
is
lt
(-2x +3xy*+3xz
j=constant,
which the series converge. a of acceleration toward the xy-plane, always component there can be no closed orbit entirely on one side of this plane. Therefore,, in all cases we can take the origin of time so that z = at t = 0. Suppose
z within the region for
and
Since there
u,= au
= 0,
z'
= y aU = 0.
series in the a,, y,
(9)
We now
power
and
X.
The
= ciie
+Pi(a n
202
100.
PERIODIC ORBITS.
Existence of Periodic Solutions. Since the right members of contain t do not sufficient conditions that (10) shall explicitly, equation (7) be a periodic solution with the period T are = ii,(T)- 11,(0) = a + /riT p (T)-p (0), [e -l]
1
O-tt.CD-u.CO) =
a2
riT
|V<V
+
>>>'
(0),
= u (T)-u
3 4
(0)
= =
*ie
a4
+ p,(T)-p,(0),
(11)
= u (T)-w
T
[e-".
(0)
-l]
+p
(T)-p (T)-p
(0),
0=
(T)-
(0)=^=sinVT T +p
(
(0),
it
0=2'(T)-2'(0)=7[cosVT T-l]+p (T)-p (0). The last two equations of 1 1 ) are satisfied by y = 0. Suppose y j* 0. Then follows from the form of the integral (8) that unless \/A T = (2w + 1)t/2,
6 6
where n
We
integer, the last equation is a consequence of the first five. shall suppose T does not have one of these special values, and we shall
is
an
suppress the last equation since it is a redundant condition. The first five equations are to be solved for a, a 4 and y in terms of X, and
,
.
by a, =
a,,
.
. .
= a = y = 0.
4
X,
terms in
a4
and 7 must be
T
This condition
e
->. T
explicitly
(12)
[/a*
=
_
1]
[-V=i
_
1]
[>
_
j] [
_
j]
sin
vjoT . 0.
VT
'
T2=
^f
.
(*
an
integer).
(13)
For this value of T the determinant Consider first the solution T = T x of the linear terms in a, a 4 of the first four equations of (11) is distinct from zero unless v<jJ2 is an integer. This condition can not be fulfilled a for all v unless <r is an integral multiple of 2 V~A Now since w = <r V 1
, . .
.
VX
satisfies
can not be
fulfilled
unless
is
its
always positive. not an integer, and one of these values of v is necessarily It follows that the first four equations of (11) can be solved for unity. a 4 as power series in y and X. Since u lf w4 and z vanish a,, = a 4 = 7 = 0, these solutions vanish with 7, and since the with a, = 2 2 first four equations of (7) are functions of z they carry 7 as a factor.
is
definition in (3),
v for
is
On
it
substituting the solutions of the first four equations of (11) in the becomes a power series in 7 and X alone, and is divisible by 7.
fifth,
OSCILLATING SATELLITES
SECOND METHOD.
203
In order to prove the possibility of the solution of the fifth equation for 7 in terms of X, and to determine the character of the solution, we must work out the first terms of the series. Terms in X alone can not be intro-
divisible
by
of the first four equations of (10) unless the fifth 2 shall 7 for the former carry y as a factor.
2
,
We
that the fifth equation of (11) has a term in 7X. It is seen from (7) and (3) that the coefficient of 7X in the expression for z is defined by
first
show
>
Li
""!!
A," *M
A'
VA
sin
\r A n t.
z(0)
(14)
The
and
(15)
<
(0)=0is
2,,^-^sinv^+^cosVi;*.
from the non-periodic term of this equation that the of (11) has a term in 7X, and therefore that the solutions exist. character we must find the terms of lowest degree in 7 alone.
It follows
fifth
equation
get their
To
The
d
coefficients of
<
dt dt
"
* l 0)
4-
oI
I*
2
VZ7 1
W0P0)
L
I
(2,
0)
2 ( Wo (To
[
2(w p
I
+WoO'o)
(2. 0)
0)
du?-
m + <' =- 2
<r
]'
-'v^T
W0P0)
Wo <r
<r
2(WoPo+WoO"o)
aU3
J.. a.
o)
dt
n ,,<2,0>_ Po M 3
2(w
+
WoPo)
]
[
(2. 0)
(16)
ffo)
2(WoPo+W
I I
dul^
dt
Zj,o
+
1
yr= +
[-2z?,
(2, 0)
2 ( Wo (To
+
noPo)
2(w
p +WoO'o)
-^o^j.o
~r"3 -Oo^i.o^i,
where
<2 ' 0,
-fJ3
+2/?.
+ 4o],
u"-
i
1
i (2, 0)
f
O
<5
_
2
-Do^i.oi/i,
(17)
We
shall
m
,
2,0*
wj
carrying 7
.
as a factor.
Hence
From equa-
tions (G)
and
(10)
we have
<>
^ 0= 2A ~~2A cos2Vir
^1,0 ^i,o
l>
=
1
2
y
+ 2V^
r (-r.+vTj V=it e
_ e (_<r,-vT) v^I<
_
(+ p,-vt; vri)(
"1
a/
(+pQ+-^i vCi)i
(18)
a ,7
:
2V7;
v^
-P,+Va, ^=i)t
p(
a*y + 2VX V
_p,_VT, VZ1)I
204
PERIODIC ORBITS.
Therefore, integrating (16) so far as the first four equations depend upon terms involving y"- as a factor, and determining the constants of integration
so that w (2,c>
( ,
22
and
z 2i0
are zero at
0)
= 0, we
get
0)
l0,
f
0)
;r+<
0)
0)
e +<w=T
'+i2
'
cos2 >/A
t+b%
d t
wa2VAt t,
sin 2
u? - oS
tt* Mi
22 z
+ a* e-*^ +
0)
11
2 0)
;
cos 2 VA - 6g
0)
v^
e -a
w
0,
2 - 0>
= a^
-N& +3:
e+'
0)
+a< 2
2 0)
cos2Vi;*+6< 2
2 0)
sin2v/i;<,
e- p"
+ar
(7
>
cos2VIJ-6
0)
sin2VI^,
n-^,|
- B| 7 e <-KH-*aav=i<
-
-1
2(<r8+2V
(
i
<r
)jroVToV
+ v/
<t
2 (ff
a,
e<+
- v/:*i)vC:1
)o-o
'
(^7 e
^ v -i'
(19)
V^
e(
a2 7
2((7o
- ff'-
V^o V-l
v ^ ri
'
+ 2 V^
a3
)r,
V^o V-l
<
2(p
-2 vT V-l )p
e<+po-
v^"o v^r
VT V-
2(p
'
Vl
~
2 (p
where
a (2.o)
'
(
= ""
,
'
8 (OT
i
3m 5 7 -n p B
2
,
fl
yl
2
'2.o) 30
_i_
3w
8
(?
<r
cr
fio7 -n p )
2
,4
a i2
(S,
w_ 0)
"T a
/, 8(w
<r
_ -n
3rn
o-o7
'
n <2 ->" 32
R(m 8(w
_
/r o-
fl *
2(w
<
<r
(2,
o) 0)
Bo 7
2 2
"
)ff
2(r?((j
-w
p )(4
7l
+p
)p
(20)
7l
2.
'12
4(m
4
(r
-n
3
/?
3 Wo Bo 7 2 p)(4^ -o )
VA
\r=\
'
W- +
^
0,
n Bo 7 2
p
)
(m
<j n
(4^4,,+p ,) vGi o
2
+ (--^(4,o^V^+(---)(4^P
(4
* - " + Cv*>
The terms
W
.
)Po'
7 are defined by
="B
x2iO z,, o
z 10
(21)
We
need only the non-periodic terms and those whose period is not 27r/vT \ The former come from terms in sin V~A~ t and e*""', and the latter from terms involving e*m v=u which together with &** are introduced by
shall
,
x.
Since
OSCILLATING SATELLITES
SECOND METHOD.
205
we
terms
5
)
35 0%0 z =35
1
,
0,
'
0,
o'
0)
(22)
p + -^^(e VA
'
+ r.
sin
VT^,
Therefore the part of the solution of (21) not having the period T^
is
/ a
rr
^.(2.0)
sm(<x
+VA
)t
"
-,
sin(o-
- \/X)J
(23)
(po+2
off n
0)
ffiT 7\ / V^ V^ V-l)
.
sin(p vo
po
fl
V=T-V^*.
0/
We
Upon
can
now
write the conditions for the existence of periodic solutions. we find for the values of (11)
tTT.vCj
-']- 2K-:)(
-i]- 2 (m
-']-i(
J
V,['" ^2
r-
_^rovv^l
]+
+ +
o-*|V3
4:% +P;)J^
,
"']
-.["
^.-^(W J
,;)
**
_-']
y7T<r
+ [(-l)'-l]^' +
v
(ii l i
^
ft
V=T [(-l)'2 a
V3I
-l]
(24)
~
(4ii
6M-7_ -
**)>. v^T L
Tr (
i} e
iv/^r
il ~X J
+
i
6J?o!r'7
(-''""'O-feysM'^^(4A+pJ)p
L
v
'
206
PERIODIC OKBITS.
a,
, ,
Upon solving the first four equations of (24) for the result in the last, and reducing by (20), we get
\ n U _ vkA'i nil 2A\
QvirBlirrUipo
a, substituting
of^,
,*,
+ Wo^o) y* _ Ai
\
'
A,ilJ
3
4(4ii
- aj)
9vrBly (4A
3
*,'
+ Po)Aj
+
From
9vTBl(mopo-\-n<,<To)<ToPoy
'
9ptCo7
16A|
(5),
we
find
m
z<ToPo
+n
-JL
y^->
(25)
B ).
-*IpI=(1-A )(1+2A
Hence, after dividing by
y,
),
-<rl+pl=-(2-A
we have
-^+i[-^S&$ar - *>+
which can be solved for y
in
<*>
7 = X*P(X
).
(27)
substituting this result in the series for at expressed in terms of y and X from (24), we have
Upon
at
when they
are
a,
= XP,(X*)
(f
= l, ...
,4).
(28)
become power
After (27) and (28) are substituted in (10) the coordinates u and z series in X } vanishing with X*, and they are periodic, since
t
,
the conditions for periodicity have been satisfied. The series converge for The radius of convergence depends on p and pa and |X| sufficiently small.
,
easy to see from the explicit forms of the equations that it remains finite For \ = p p the orbits belong to the physical problem, as Mo approaches p. and p can be taken so near p that the series converge. That is, periodic
it is
x= 2
where the
x,
,
x X2
4
y= sy.xs
= 2
z,X
2
,
(29)
y
.
<)
t ,
and
z,
having the
period 2ir/VA~~
OSCILLATING SATELLITES
SECOND METHOD.
207
The last two equations of (11) are satisfied by 7 = 0. For T = 2ir/VA^ the determinant of the linear terms of the first four equations is distinct
from zero; therefore
X,
a 4 as
, ,
.
power
.
series in
,
vanishing with
X,
a:
= a = 0.
4
But then
ut and
are identically zero. That is, the solutions having the period 2tt/Va^ are in three dimensions and not in two alone. In this respect they agree with the solutions of Class of Chapter V. It will now be shown that for
=M
Mo
r
series in
e'
The
form
solutions of Class
of the
x= 2
<=i
x e'\
(
2 y= <=i
ytf*t
z= 2
<=i
z,e",
(30)
y and z are periodic in r = t/(l + 5) with the period 27T/VT. The appearing in the x,, y 2, of (29) is different from the A of (30); in the former n has been replaced by /x in certain places, while in the latter it remains /x. Now let /x = /x +X in (30) in those places where this transforma-
where x
t ,
t ,
t ,
tion
was made
in the
development
power 27r(l+5)/VT, where mation on n in this expression so that A =A +p (X) and set it equal to the Since 5 starts with a term of the second degree period of (29), viz. 2tt/Va in e' this equation determines e' as a power series in X*. Substituting this } expression in (30), we have these equations expressed as power series in X For sufficiently small X these series converge, the coefficients of each
.
series in X.
The
of (29), and develop the right members as of the solutions (30) when expressed in t is period 5 is a known power series in e'. Make the transfor-
power of
X*
satisfy the correspondingly transformed differential equations, and they are identically equal (in t) to (30) in their original form. Having the same form
as (29), it follows from the uniqueness of these solutions that they are identical with them. That is, equations (29) and (30) are two different
sets of expressions for the coordinates of the orbits of Class A. return to the consideration of equations (11). Let the last
Now
two
be
satisfied
by 7 = 0.
As we have
.
seen, there
is
no solution
1
Therefore a4 ^0 if T = T = 2ir/VA' except a 1= vanishing with X = = = The first equation is redundant because of the we take T T 2 2t/(t The third and fourth existence of the integral (8), and will be suppressed. a 2 and X, equations can be solved for a3 and a4 as power series in a in the are substituted with and When these results a, c^ vanishing second equation, we have
x ,
,
.
(31)
where
ax
and
0^
20S
PERIODIC ORBITS.
have one equation for the determination of the parameters a, and a,,. Consequently we may impose one condition upon them. It will be convenient to take a, and a2 so that
x'
We
= 0at = 0,
t
(32)
a condition which
is
satisfied in
From
o-
every closed orbit for which the right memequations (6) it follows that this condition is
(a
1
v^T
-a )+p
2
(a 3
a 4 )=0.
(33)
We
shall regard this equation as a relation between a, and a 2 which is to be used in connection with (31) for the determination of these parameters. In order to complete the proof of the existence of the solutions it is
It follows necessary to discuss the second and third degree terms of (31). from (7) and (3) that the terms of the second degree are determined by
(2)
dt
-oj,uT=-m
w
a[
]
i-\ r,
(34)
du?
dt
Pa
uf=+n,[
r +
(2)
where
I
(2)
'
2(m
It follows
<r
2(mopo+no<ro)
(35)
of these four equations that their solutions will contain Poisson terms,* whose coefficients involve Xa4 respectively as factors. The coefficients of all the other Xa,
. ,
of the right
members
terms arc of the second degree in a, and those a, and linear in B which are not periodic involve a3 or a 4 at least to the first degree. Consequently, when we solve the third and fourth equations of (11) for a3 and a 4
, .
the results will start with terms of the second degree in a, cu, and X. When these results are substituted in the second equation of (11), it will contain a term in X ctj and terms of the third degree as the lowest in a, and au alone.
, ,
If
we now
eliminate a
of (33), we have an equation whose terms shall verify first that the coefficient
We
*In Celestial Mechanics terms which are of the form of / multiplied by cosine or sine terms are called Poisson terms, from the results in Poisson's theorem on the invariability of the major axes of the planetary orbits.
OSCILLATING SATELLITES
It follows
SECOND METHOD.
terms in
209
from
(34)
and
i
are
(2>
=_
m A'\a te-<
(mo
(To
v='u
n<>po)
V^l
^V^Ti^aj/c-''^'
2(m po+no <To)
/2tt\
\<r /
_ w (2)/qx = _
irA'Xat
2m,
WoPo
"I >
O-oV
(.moffo
po+Wo(7oJ
&) - tew =
Vffo/
T5yffftf? /^ V (1+2^1 K + p
) ) cr
(1+
)]
Xa <
( v
36 )
Therefore the coefficient of Xa2 in (31) is not zero unless A' = 0. But A'?*0 except for the center of libration (6) when the finite masses are equal, and
in this case a different generalization of the
keep
o,
it
distinct
from
zero.
tically satisfied
=o = 0,
4
is is
divisible
in X alone
distinct
X,
from
zero.
in
terms of
vanishing with
The form
nated.
It
is
of the solution
term of the
easy to show It has been shown that the terms arising from the soluzero. identically There are also terms of the third degree tions of (34) involve linearly.
equation after a a3 and a4 are elimithat the coefficient of ai) in this equation is not
o4 arising from the terms of the third order. in o,, w, of the third order are defined by
.
.
The terms
of
dih
dt
r
<
ZT %P* =
2
ot
v'
]__[
2(rao<r
noPo)
2(wop
(37)
+no<ro)
where
[
f = 3B [-2x
)
x2 +y yi \+2C [2xl-3x
l
yl],
as a factor are
=_
"
3(7,
-n
(m
ao
a^ e-v^i' + noPo) V
)
t
3(?
(4
- 3 njj)
po+w
a.cjfe-'.^"
cro)
4(ra
Hence we
some
reductions,
+- '-H'
3
(38
210
It follows
PERIODIC ORBITS.
from equations
(5)
that
-j
Hence the
plus non-vanishing terms multiplied by C a are eliminated from the second equation and suppose a,, It of (11) by means of (33) and the third and fourth equations of (11). follows from the properties of these equations that in the result the term Its independent of X and of lowest degree is of the third degree in a2 coefficient consists of two parts, one of which is (38) and contains C as a
of terms multiplied
by
Now
a3
as a factor.
If the coefficient of
is
3 identically zero, the coefficient of a2 is distinct from zero, because, as we have Even if the coefficient of seen, the part involving C is distinct from zero.
is
not zero,
(a), (6),
it is
coefficient of a\
easy to show that the sum of the two parts of the can not be identically zero for each of the three libration
(c).
points
and
(a)
and
of
(6).
The
quantities
<r
both points, but B is different because /x of the change of sign in its second term [eq. (3)]. Consequently, the sum of for both the points the terms in B and C can not be identically zero in
for
/j,
Hence in this case the second, third, and fourth of libration (a) and (6). of and o 4 as power series (11) (33) are solvable for o, equations
, . . . ,
in X*, vanishing
2ir/<r e
with
X.
are expansible as power series in X*. In a manner similar to that used to prove that (29) are series which represent orbits of Class A, it can be shown that the orbits now under consideration belong to Class B.
As
in
the
Chapter V, the coordinates in the orbits of Class A are most confrom the x, y, and z-equations. Consequently we start obtained veniently from equations (2) and (3). Since the solution is periodic for all |X sufficiently small, each term of the expansion separately is periodic with the
method
of
at
= 0,
The
by
1
x!-2y' -(l+2A
l
)x 1
= 0,
y:+2x' -(l-A
)y l
= 0,
<-M*-0.
and
initial
(39)
The
con-
ditions are
*i
= Vi = 0,
*i
-^T
sinv/ ^o t,
40 )
where
c, is
so far undetermined.
OSCILLATING SATELLITES
SECOND METHOD.
211
The
by
<-2| :-(l + 2i
/ i
a;2
2
=fB
o
JJ
[-2x;+|/;+4
(41)
3 B.x.y,,
Zl
Upon making
initial conditions,
we have
x>
4(l
+ 2,4 )A + 4(1-7A
+1SAI)A
C0S ^ VA
^
(42)
z2
+ ^j-sinVAj,
where
c2 is so far arbitrary.
be necessary to carry the computation two steps further in order to show how the general term is found. The coefficients of X are defined by
It will
3
x:-2yi-(l+2A A t -(lyl+2x'
z" 3
)x3
B
)y3
z3
+A
+\
A &
j-*_i_ 3 t'oCi
(43)
_
8(1
27ff(l-3.A +14A;)ci
sinv/x<
In order that
it
shall
be
sinV^i must be
&
zero, or
_ JL c lT 4-5. 8 VA
This equation, which
solutions
is
^i
'
(U)
2V2A
Ci
'
A'
(45)
Ci_
3\
n~ C
(l
8flj(i-3ii.+i4i ir
+ 2^)(l-7Z 4-lW
212
PERIODIC ORBITS.
solution
Cj
The
= y = z=0,
as can be
shown
The double sign before the easily by an induction to the general term. radical plays the same role as the double sign before X* in the existence.
If it is
used in one place in the final solution it is superfluous in the other. of c, determined from the second of (45), the solution
and
initial
conditions
is
x>
2(1+ 2A )A
2(1
- 7A +18AI)A
C0S ^ VA
^
(46)
where
c 3 is so far
The equation
z t is
<+4*
-A'z
+3B
(x 2 z2
+xiZl +
)
z\z2
(47)
In order that the solution of this equation shall be periodic the coefficient of sin V~A~ 1 in its right member must equal zero. This function of t arises
from every term in the right member of (47), and it follows from (40), (42), and (46) that its coefficient carries c2 linearly and homogeneously. Therefore this condition determines c2 uniquely by the equation c2 = 0, whence
z2
=x =y
3
=0.
t
(48)
After the sign of c, has been chosen all the other c are determined uniquely by the conditions that all the z separately shall be periodic. For, z,_i have been suppose that xlf x^; ylt y_t ; z u
t
. . . , . . .
known except
(
the arbizt _ lt
/VA
c,_ 2 ,
in
.
and
z t is
The
z';+A
zt
=-A'z - +3B
i
(x i zi _ t
+x _ z + ^C
i
1)
z\z _ 2
t
(49)
enters linearly in the coefficient of sin V^A~ t in the equation, which does not involve c _i, and the constant
c,_ 2
is
uniquely
OSCILLATING SATELLITES
It
SECOND METHOD.
213
properties
1.
Therc2i+1
y2j+l
s
= 1,
2,
oo).
2.
3.
The Xj, ylr z involve c homogeneously to the degree,/. The x2j are sums of cosines of even multiples of \^A t,
multiple being 2j.
4.
The y2j
The
are
sums sums
of sines of
even multiples of
y/~A a t,
z2j+1 are
of sines of
odd multiples
Changing the sine of c is equivalent to changing the sign of X which is equivalent to increasing t by ic/VT^. Therefore the two values of c (or X*) belong to the same physical orbit, the origin of time being different by half a period in the two cases.
t
,
7.
The
and the
It is
xz-plane.
not necessary to go into the proofs of these properties, which are the same, so far as the comparison can be made, as those found in 87.
102. Direct
advantageous to use the first four equations of (7), being identically zero. We have proved that the solutions are expansible as power series in X } that the coefficients of each power of X* are periodic with the period 2ir/<r and that x' = at t = for all X.
orbits
it is
,
,
The
by the
differential equations
dUl
dt
<
% - PX
dt
- 0.
(50)
dt
+<7 V
1Ui
'
dt
The
<-a ^ v=I,
l
u uf = a e-"^
2
u =u
it is
<
= 0.
(51)
From
equations
a 1 = a2 =
(6)
and the
x
0)
initial
value of x'
found that
(l)
,
=a
(1)
cos
<r Q t,
j/,
n a a)
sin
<ra t,
(52)
where the
coefficient a
is
so far undetermined.
214
PERIODIC ORBITS.
The
by
I
du%
dt
(2)
(2)
(Woffo
/loPo)
V V
)
(WoPo+Woffo)
l
du?'
eft
ff
ft*. 2
Mj
(mo
<7-
r
(2)
I
l"
(2)
nPo)
(moPo+Woffo)
(53)
l
duf
dt
Po<="
^
p* Ul
nj
(m
<r
(2)
n p
+
'
(mo
Po+ Wo oo)
J
t
duf
dt
(nhao-noPo)
L (niopo+noao)
J
where
[
r=fB
I
[-2x?+^]= =
+ |5
-|fl
(l)
(a
<],
(54)
T>
(1)
(a
The
a;'
are
tt
a (B
e^+a+a
e<r
)
00B2*,*-t&ffsin2r,<,
)
ufM*
where
a (2)
<2)
(8)
2) 2
sin2(r
sin 2 (r,
sin 2 <r
<,
(55)
+<+a< + off +
2)
cos2*r
*+
1
6
6
J,
32
cos 2 <r
*,
is
so far undetermined,
= =
3m
3n
fi
(2-n )(a
a> 2 )
8(moOo
(2,
nopo)<T
fl
B (2-nl)(ay
(2+nl)
a)
"30
8(m.o<ronopo)p<i
<= +
2)
m B
(a
Y
2
B(a
a)
)
8(Wo(To
ttoPoVo
)
2{mopo-\-noao)<ro
)
(56)
// "l2
=-
m Bo(2+^
4(m
3n
8(
fl
ffo <ro
2
)(a"
+
2
a)
(a
y
ffo
2
noPo)o"
(1)
4(m po+^o
3n
2 ( mo po
<r
0o)
U 32
(2+tt )p (a
noPo) (ial
2
?
+ pl)
<1)
+n
+ pi)
2
6<2)
Jgo(2+r )r (a ) 2 4(moffo-noPo)(4<rJ+p )
3
3n /? 4(mopo+n
o-o)(4(7-o+p5)
OSCILLATING SATELLITES
SECOND METHOD.
215
(l) is determined, except as to sign, by the periodicity arbitrary a condition in the next step of the integration; and the double sign is 1 After a a) has been determined, an equivalent to the double sign on X
.
The
uniquely determined by the periodicity condition at each succeeding of the integration. step The coefficients of X ? are defined by
is
aw
* -a iu=+ dt
'
(m a a
<r
^ -n
w
<7
t
r
p
)
V^T
V^i
p
(m
U* +n
p
i
<r )
<+aiu=- (m dt d<_ ^ 0U v = _
2
-n
^
p
)
noLJ"
(ra
(moPo +nQ a w
LT!L_,
)
(57)
,
dt
-n
T
.
(m
p
1
+n +n
<r )
dvl
d^
flW4 ^
noLJ"
(.m
-n
+^ (m
ra
where
l3>
A'z
+fB
[-2o
;i;r 2
2/ 1 2/ 2
]+ C
+ 2^-3x^l,
(58)
equation shall be periodic it is s/:=lt e"' be equal to zero right That part of this coefficient which arises from A''x involves a a) linearly and 3 homogeneously. Those parts which arise from x y^ x\ etc. carry (a ) as a factor and involve it in no other way. Consequently, the condition that the coefficient of e v/=T shall vanish is satisfied by a = 0, or by an equafirst
member
the coefficient of
y
(1)
<r
'
(1)
tion of the
form
P(a
where
U) )
+Q = 0,
It
is
(59)
and
are
known
constants.
a?,
easily
are
and a\ which
100,
in
the
determination of a^ leads to the trivial solution x = y = 0; equation (59) gives the double determination for a mentioned on page 210. In order that the solution of the second equation shall be periodic it
The
is
necessary that in
is
to zero.
which
_<wzrT< be equal right member the coefficient of e (1) It is easy to see that this condition determines a by an equation That is, the same value of a a) makes the identical with (59).
its
first
and the second equations periodic. particular integrals of the third and fourth equations are periodic. In solving the third and fourth equations the constants of integration are
The
216
right next order are
PERIODIC ORBITS.
The
members
T =A'x +\ B [-2xl+y\-4x x +2y y + C [6xlx -6x y y -Sx M \\ =~A'y +lB {x,y +x y +x y + ^C >-8x x y -4x\y +3y y
t 1 3
1
s}
y\]
(60)
i \.
i
l
1 \
first
e a,v=Tt in
its
right
the terms of (60) that a <2) is <2) Moreover, a is the only unknown quantity in this coefficient. Therefore a w is uniquely determined by setting the coefficient of e'"^11 equal to zero. The condition that the solution of the second equation shall be periodic
is
third
At
[ ]
first. The particular integrals of the are periodic. step the right members of the differential equations involve
imposed by the
w = A'x
w
^
-3x _
t
y\}+
2
(61)
-^'^- +f5
2
^x _ -4^^_ +3^^_
(
;+-
In order independent of a that the solution of the first equation shall be periodic it is necessary that the coefficient of e'w=rt in its right member be put equal to zero. This <i_2> coefficient carries a in and and the linearly general non-homogeneously, (<_2) known factor by which a is multiplied is precisely the same as that of a a) in the equation by which the latter was determined. Therefore the arbi<1_2) is uniquely determined by setting the coefficient of e <7,v/=T< trary a equal to zero, for it carries no other unknown. When this condition is satisfied
explicitly written are
.
'~ 2)
the coefficient of e~ v:=u in the second equation is zero, and the entire solution at this step is periodic. Therefore after the sign of o has been chosen
(1)
the process
is
unique, and
it
Chapter
VII.
The
Suppose the
finite
bodies
describe ellipses whose eccentricity is e. Let 1 /x and p (p^0.5) represent their masses, and then determine the linear and time units so that their mean
distance apart and the gravitational constant shall be unity. units their mean angular motion is unity.
With
these
system to a set of rectangular axes with the origin at the center of gravity, and let the direction of the axes be so chosen that the i7-plane is the plane of motion of the finite bodies. Suppose the 7?-axes rotate with the constant angular rate unity around the f-axis in the direction of motion of the finite masses, and suppose 1 p and p are on the -axis
Now
refer the
when they
of
Then
df
,
dt
cFn " 9 d + 2 dt dt
_n
ri
(1
pXq
r\
m)
niy
ri
m)
'
(1)
<k
dt*
a-M)r
hj
'
ri
where
r^Va-S^+CTj-T^+r,
and where
, ,
r2
r) l
and
vt
?j 2
are determined
by the
move
in ellipses.
If
we
let p,
and
/x
referred to fixed
t>
axes having their origin at the center of mass of the system, and p2 and the corresponding coordinates of p, we have
& = -p
7],=
cos(
-0,
& - +acos(i>,-0,
77 2
-pjSinOi-0,
jl
=+p sin(>
2
-0,
[
Pi^+M
Pt
ecost+
(1
cos 2<)+
(2)
=(l-n)fl-eco&t+j(l-cos2t)+
},
vl
= v =+t
i
+ 2esmt+ - e sin2+
2
where the
initial
value of
1
= 0.
218
104.
PERIODIC ORBITS.
The
Elliptical Solution.
elliptical
body we
The equations
of a
of
mass
are,
when
(3)
= V?+v +?2
We
orbit
and
if
the
If the eccentricity of the to the fixed axes is unity, respect determination of the origin of time and
= rcos(#
r
t),
r)
= rsm(v
t),
= 0,
}>
=k
(l-cos2i)+
(4)
= 2esint+ 4e
4
sin
2t+
It will
now be shown
tesimal
body moves
that equations (1) will be satisfied if the infiniso that the ratios of its coordinates to the corresponding
by
rj
=f 5=5
Upon making
M|
^-~t+m-'
(5), it is
*- i+m
'
(o)
found that
.-*,=
t
+ _ l_ My ti{
l
*>-%-+ i_ m( + M)'
i
-t ?s
(i its' i-/i(i+M)
i
i
Mo
" Vo
"2 Vi
=_
Vo
i- M (i+M)
(6)
^ i-m(i+m)
r2
^r- M (r+^/)
219
Then equations
? So
(1)
become
2
"
-f>
(7)
>"
io
= _ [(i-m)M +m]Ci-m(i+M)]
3
'
f
rl'
The
first
two
same form
as (3),
and
their solu-
=r cos (-0,
r.v
ij,
=r
sin(i;-0,
= 0,
I
[a-KW-dfft
e a+Ml'
!1
_ eco8(+
(8)
+ 2esmt+^e 4
and
sin2t+
From
these equations
(2),
we
find
2
& =
and from
(6),
vo
& = ""
&
38
r?i
= _ [1-M(1+M)] M(l+M)
"
,
On equating these two expressions for the ratio &/, = ^j/^ and rationalizing,
we have
(l- x)M
f
(9)
It
is
Therefore, for those values of and are a (8) particular solution of the threesatisfying (9), equations (2) body problem where one mass is infinitesimal. As is well known, there As in are three real solutions, one for each ordering of the three masses.
obtained.
Chapter V, we
shall call
them
(a),
(&),
and
(c)
Equation (9) is Lagrange's quintic in case one mass is infinitesimal and the units are chosen so that the masses of the finite bodies are 1-ju and yu. For example, if in equation (60), page 216, of Introduction to Celestial Mechanics, we put w, = l-ju, m 2 =0, and m3 =n, we get equation (9).
220
105.
PERIODIC ORBITS.
Equations for the Oscillations.
We
shall
For
this
= +*,
in
= v+y,
= 0+2
x, y,
(10)
equations (1), and expand as power series in transformation and expansion, we let
and
z.
After this
M = Mo + X
in those places
(11)
where n appears explicitly. This is not the only way in which the original n can be divided into the new n and n +\, and sometimes The coefficients of the various powers of x, y, and z others are advisable. are expansible as power series in e, the terms independent of e being constants, We find from (6) and (8) that as is seen from (2) and (8).
f>
G,-&-rf[l-**-
|(l-cos20
]>
$,-&-r[i-eeo8*-|(l-eos2*) +
]>
*-%?[
U,-ih-f[
r,
+2esmt+^sm2t
+2esmt+^sin2t
f[l
0)
+]>
+'].
coe*
- Vtti-eO'+Oj.-*)' -
>
r,
=V(o-y
i
j
+(%-rji) = ^ [l-ecosi
2 2
J ,
r'
-_
~M
-M Li-m(i+M)J
r (i-M)M
_
?2
-|
+]> 2
*
.
Consequently, after making use of these expansions and the transformations (10) and (11), equations (1) become
x"-2y'-h+2A+()Aecost-3Ae (l-5cos2t)+
-\6Aesint+^Ae sm2t+
h
l
-\x
-\y = X,
y"+2x'- J6,4esin*+
A(?sm2t+
2
-\x
(13)
-jl-^-3^1ecos<+|^c (3-7cos20+
z"
]y=Y,
=
Z,
-\z
221
where
j yi
1
I
Mo
(0)3
Mo
cms
'
i"
-r
'l
'2
+ - 6 |_f8a
1
sin
psajc
<+
12
yX
eco
+4- 3
+
^+
-}*
+{l[Tr-^]
7r=rrO
eco
^+
}v
F= +
-6|
(oy3-
-l
7
76T3jesini+
(14)
+{4^=?]+
+ {+3[j
Z= +
\
i2
[
?--^] ecos< +
^m
-(6T3J
^]esini+
+3[pr
}*'+
cos
[py
pa]
t+
J
zX
+ {6 [
^ = ^]
esini+
}*+
,
+ +,+
(a),
All terms
up
and
X are written.
222
106.
PERIODIC ORBITS.
the right
(a)
The Symmetry Theorem. It follows from (2), members of (12) have the following properties:
(8),
and
(1)
that
The
and
Y
of
z,
and
involves only
odd powers
(b) In
X the
coefficients of all
sums
all
and the
coefficients of
(c)
sums
and the
coefficients of all
Suppose the
x=
a,
initial
conditions arc
x'
= 0,
y = 0,
if'-ft
= 0,
z'
= y.
(15)
Then
x=f(a,
y = g(a,
0,
0,
0,
0,
y;
t),
x'=f'(a,
0,
0,
(3,
0,
y) y; y;
t),
0, 0, 0, 0,
0, 0,
T T
0i
0,
z'
-/(,
= h'(a,
0, 0, 0, 0,
0, 0,
t), t).
(16)
z=h(a,
0,
0,
0, 0,
transformation
y'=+y',
(a)
,
. .
.
*--*,
,
*'=+*',
form
t=-t.
(17)
with the
x=a
= 0,
y=
0,
*P,
1-0,
of (17),
z'
= y,
x(t)=x(t)=x(-t) = +x(-t),
y(t)=y(t) = y(-t) = -y(-t),
z(t)=z(t)=z(-t)=-z(-t),
'
(18)
.
Therefore, with the initial conditions (15), x, while x', y, and z are odd functions of t. t,
body
apse,
motion
is
when the finite bodies are at an with respect to the z-axis. symmetrical
223
The Terms of
Suppose the
initial
conditions are
=a
y(0)
=a
y'(0)
=a
(0)
= as>
z'(0)
=a
,
(19)
, ,
We
a5
,
a2 a4 integrate equations (13) as power series in a, a, Since there are no terms in the differential equations independent of x, y, and z and their derivatives, there will be no terms in the solutions independent of c^ a6
,
shall
,
now
A.
a6
and
The terms
. ,
a 6 are defined
by the
differ-
ential equations
xi-2y[-[l+2A+6Aecoat+
y'{+2x[-[6Aesmt+
2
~\x l
-[6Aesint+
-^ =
0,
]x
[l-A-3Aecost+
]&-0,
(20)
]%-0,
The coefficients are power series in subject to the initial conditions (19). with the and reduce to constants for e = 0. The first e, periodic period 2w,
two equations are independent of the third, and conversely. For e = equations (20) become simply
x';-2y[-[l+2A}x =0
1
>
y';+2x[-[l-A]y = 0,
1
zl+Az^Q.
(21)
23-25 it follows that the properties of the the character of the roots of the characteristic depend upon equation of (21). This equation for the first two of (21) is
the results obtained in
solutions of (20)
s
i
From
+(2-A)s +(l-A)(l+2A)=0.
(22)
Two
roots of this equation can be equal only if A has one of the values The first two are excluded by the fact that A is necessarily 0, 8/9, or 1 When n is near n in value, as it will always be taken here, is positive.
- 1/2,
greater than unity and, consequently, the last two values are excluded. Equation (22) has two pairs of roots equal in numerical value but opposite in sign, and for A > 1 two of them are pure imaginaries and two <r V Let us represent them by where aa and p are are real. l p real. We now raise the question whether two of the roots of (22) can differ by an imaginary integer. In order that this may be so we must have
,
>
cr
V^T=
2
-j\T^l,
where j
is
an
integer.
we
find
(23)
224
PERIODIC ORBITS.
The
question is whether there are integral values of j giving admissible values for A by (23). To insure the convergence of the final series it will be necessary to take m nearly equal to n, and we shall suppose at once that It was shown in 82 that when Mo = M the value this condition is satisfied. of A exceeds unity for each of the points (a), (6), and (c) for all values of m-
From
of
the expansions given in equations (42), p. 206, of Introduction to Celestial Mechanics, it is seen that for very small values of n the values
are approximately 4, 4, and 1 for the points of libration (a), (6), and (c) respectively. In the numerical example of 90, for which n = \/\\,
was found that A is 2.25, 6,51. and 1.08 for the points of libration (a), In the extreme case of /x = 1/2 we easily find from (6), and (c) respectively. 76 and 82 that the values of A are 1.56, 8, and 1.56 for the formulas of the points of libration (a), (b), and (c) respectively. While n varies from to 1/2 the values of A vary from 4 to 1.56, 4 to 8, and 1 to 1.56 for the points (a), (6), and (c) respectively. From this we see what values of A are possible in the actual problem, and we shall be able to determine whether j
it
takes integral values for any of them. When the negative sign is taken before the radical in (23), the function 2 16 A has its greatest value of zero for j = 4. Consequently we get no
using this sign before the radical. When the positive sign is taken before the radical in (23), j and found to have the following relations:
If
admissible values of
are
then
A =0.92,
1.00,
2.01,
3.73,
5.94,
8.70,
and A is larger than 8.7 for all j larger than 6. The values of A for j between 2 and 6 are admissible for either the point (a) or the point (6), but not for the point (c). Consequently since A, and therefore a, is a continuous function of m and n there exist pairs of values of n and n for which the difference of the imaginary roots of (22) is an imaginary integer, but they are exceptional.
,
The
is
simply
+A = 0.
Hence the
form unless 2
solution of the third equation of (20) does not take an exceptional is an integer, or
VA
A =
It follows
f
Zj-
(j
an integer).
from the discussion above that the admissible values of A satisfying this relation are 2.25, 4.00, and 6.25. These values belong only to the point (a) or the point (6), and in no case can the congruence be satisfied
for the point (c).
225
istic
A
in
According to the results obtained in 23, the solutions and characterexponents of (20) are always expansible as power series in e except when has the special values noted above; and according to the results obtained
24, the
same
even
if
istic
equations differ by imaginary integers. However, in the latter case the construction of the solutions is quite different.
It was proved in 33 that in equations of the type under consideration here the characteristic exponents occur in pairs which are equal numerically but opposite in sign. Therefore the solutions of (20) are of the form
pt
pt
2,
11
1''
<s
+ + +a e-*[ - pu + w ^T v + a e-'^1 + a v + a e~ v Vi a, e u = a, **=* [a V~^l + + a e -^~ V~=J v + v' y[ +a e [pv +v' ]+a e- [-pv +
a 3 d* [ Pu3
3
2]
u' s]
4]
av/
'
'
i\
e"'
pt
(24)
i\
v[]
<x
2]
i
pl
pt
v[},
=c =c z[
zx
e^
vz=rY
'w 1 +c2 e~
0,v
^^
= e aV
.
T,
[wx^Tw +w' +c
l
l \
e~ Uv'=* [-o>\r
t
^iw +w'
i
2 ],
where
aj
,
. .
a4
cx
a-2 e
c,
a
p
= =
(r o
+0-ie
+
-!-
tt4
tr
uf
0)
+w
<1)
e+tl V-|<
)
p.+p.e+fte
1
8
,
=VZ+
uf, vf,
e+tt s e*+
are constants.
vl
>
can be taken equal to unity without loss of generality, and will be so chosen. Moreover, the u i} vi} and w are with the and since this period 2w, property holds for all e for which periodic n n and w\ separately is periodic with the the series converge, each u\'\ v\ coefficients of 2t. The these series can be found by the methods period 26. The a and c, are uniquely expressible in terms of the set forth in initial values of x, x', y, y' z, and z', the a, of equations (19), because the solutions (24) constitute a fundamental set, by hypothesis, and the determinant of the coefficients of the a and c, is therefore distinct from
initial
The
values of the
v,
and
zero.
We may
The
and
c,
or the a as arbitraries.
t
and o> are real for e sufficiently from the third equation of (21), now arises shall as we small, which is of the same form as that treated in 50, where it was shown that
characteristic
<r,
p,
the
trV
characteristic
1
and
exponents are pure imaginaries in this case. p are roots of an equation of the form
A(a, e)=0,
The
(26)
226
PERIODIC ORBITS.
is
where A
an even function of a and a power series in e [see 22, and For e = the solutions of this equation are particular equation (98)].
a
in
o-
1,
where
<r
and
p are real.
Now
let
and
(26)
A(<r
becomes
V=T,
e)+A'"(<r
v^T, e)0V=i -
A"(*
V=T,
e)/?+
=0.
(27)
<r
1,
and the
coefficients
of the various powers of (8 in these terms are therefore real; all the odd and are multiplied by odd powers of derivatives are odd in <r 1
and the
powers of
(er
j8
in these
0)
is
terms are
1,
distinct
from
zero under the conditions satisfied in this problem, the solution of (27) for /3 as a power series in e, vanishing with e, gives a unique series whose coefficients are all real.
Therefore
initial
<r
=a
is real.
It
is
a real constant.
Now
we have
suppose the
y
l
by
(15).
Then, since
{t)
= -y
{-t),
d
t
*tf)--,(-0,
v t (t)
altrv=~*'v
(t)
+a e-^%(t) +a
i
ef
+a
e- pt vi (t)
=c1 e
[a.e-'^^'v.i-t)
l
+a
i
e'^'i<2 (-0
"
1
+a
-"
1
(-0+a
2
p
't;
(-fl],
aV=ll
(t)+ct e-
^'w
of
(-t)
58,
= - [c e~ u /=s'w
(-t)+c
e wV=It
(-*)]
it
On
applying the
lemma
follows that
(24)
+a
+
1
[e
{+t)]+ a
[e
+ "w
pt
(+')
--*t,(+0]
i
yl
+ = -a [e-^'u (-t)-e ^'u (-t)]+a + = +ale ^'v (+t)-e-<>^'v (+t)]+a = - e-^ v,{-t)- **=* - 1) - a
1
1
[e,
[e+>
v 3 (+t)-e-
pt
vi
(+t)]
(
1'
fll
i>,(
[V*
,(
Zl
= +c
[e^^'ti^-M)
- e-"^= I
'u^(+0]
227
we have
*(-!), V3 (t)=V (-t),
t
= -u (-l),
2
^l 3 (t)
= -U
v,
,
(-t),
*(0
W^^Wii-t).
and
u =
y
+2
2
[A j cos jt+Bj
[A j coa jt
am jt], am jt],
Pj iu
m2 = u3 = u = Wt =
t
Bj
+2 [Cj
2
[Cj
v3 vA
l
Dj
ainjt],
,
= +2 = +2 = +2 = +2
am jt],
F am jt],
(28)
w =+'L[K
cos jt
sin jt]
Since the Uj vs and to, are defined by linear equations they are independent of the initial conditions. But the equations a^-a*, a 3 =-ait c =-ci hold only for the initial conditions (5).
l
108.
in the a
,
t
Second Degree. The terms of the second degree and X are found from equations (13) and (14) to be defined by
of the
The terms
x2 -2y' 2 -[l+2A+6Aecoat+
a yZ+2x' -[6Aeainl+
-]x2
[6Aesin*+
]&-X
-]y2
] ]
x2
z2
+ [A+ 34ecos<+
where
-[l-A-3Aecoat+ - Z},
=Y
(29)
=+
[-2A'+ [-SB -
GA'ecoat-]
]x,X+[~]x\
6A'eamt-\
]r/,X
12Becoat-]
+ [ -24J5esinH + [fl+6flecosf+3A'ecoat-\
],y,
-}z\
+ [+|5 + 6ecos*+
F = +[- 64'esinH
2
]j/J
]^X+[ A'+
]zj
]y,X
~\x 1 y
l
+ [-12e + [+
2
sinH
+ [3
+[
+ 12ecosH + 3esinH
9fie
sinH
]*S
]zjH
]x,z,
~\y 1
Z = +[,4'+3A'ecosH
]z,X
+ [3fi+125ecos<H +[
6Z?esin*+
1
zl
ft-
'
l
(0)3
1
Z(073
' '
+ U - +
Mo
(0)4
+
'
Mo
Zffift
'
'1
'1
228
PERIODIC ORBITS.
It is necessary for further work to determine some of the properties of These properties depend upon the form the solutions of equations (29).
and properties
The
of their right members, which are given in equations (30). of the coefficients in these equations easily follows character general On referring to the results which 106. from the properties (a), (d) of were developed in Chapter I, it is found that
. . .
[1]
solutions of (29) consist in the first place of the complementary The arbitraries functions, and they are identical in form with (24). a, and c, which "appear are uniquely determined by the conditions
The
x, (0)
=x
(0)
=y
(0)
y2
(0)
=z
(0)
=z
(0)
= 0, where
x2
z' 2
are
There are terms arising from those parts of the right members which contain X as a factor. These right members consist of sums of terms which are periodic with the period 2-k multiplied by one of the funda^ e +pt e~ pt e uvr^' and e~ uVzri1 mental exponentials e +ffv/ T( e~ aV:= Hence it follows from the results of 30 that the corresponding parts of the solutions consist of sums of periodic terms whose period is 2ir
~~"
multiplied by these same exponentials, plus t times the corresponding shall be particularly interested part of the complementary function. in those terms which contain t as a factor. They are linear in the a,
We
and
and
[3]
c,,
and
for e
stants.
The
the parts having the period 2-k reduce to conexpressions for x2 and y2 do not depend upon c and Cj
x ,
is
independent
of a t
a4
of X
multiplied
consist of
of the right members of (29) which are indesums of periodic terms having the period
of the funda-
the special case where mental exponentials. Therefore, except a or co is an integer, the exponents of the exponentials in the right members are not congruent to any of the characteristic exponents
mod. V
these
t
hence
it
follows
by
solutions consist of
sums
by
same exponentials.
and the
c,
.
as a factor.
a,
In particular, there are no terms containing These terms are homogeneous of the second degree in
the
109.
The Terms
c(
,
of the
Third Degree.
of (29).
The terms
right
in the a t in
members
The
which are
(a) linear in
y ,
y
,
(b) linear in X and of the first degree in x2 y 2 (c) of the third degree in x lf y lf and 2,; and (d) of the first degree in
xlf ylf
z,
and
in
x2 y2 and
, ,
z2
229
The
[4]
There are the complementary functions identical in form with the The constants which appear in them are determined expressions (24) by the conditions that x3 (0)=x 3 (0)=y (0)=y 3 (0)=z 3 (0)=z' 3 (0)=0.
.
[5]
The
(a) consists of
of periodic terms, period 2ir, multiplied by the squares degree products of the fundamental exponentials.
[6]
terms of two different classes because x2 z and consist of terms of two different types. There are terms 2 y2 which contain X as a factor, and a part of these are sums of periodic terms, period 2ir, multiplied by t times the fundamental exponentials; the remaining part lacks the factor t. These terms are homogeneous
(b) give rise to
-
The terms
and the c, The corresponding parts of the solusums of periodic terms, period 2x, multiplied partly by t2 partly by t, and partly by f. They are all homogeneous and linear in the a, and the c, the x and y- terms being independent of the c< and and
linear in the a,
.
tions are
a,
coming from the other part of (b) are sums of periodic terms, period 2ir, multiplied by squares and seconddegree products of the fundamental exponentials. They are homogeneous of the second degree in the a and the c
of the solutions
t (
.
The terms
[7]
The terms
a,
t .
and the c sums of periodic terms, period 2t, multiplied by the fundamental exponentials to the first degree; and the second of which are sums of periodic terms, period 2x, multiplied by cubes and non-canceling thirddegree products of the fundamental exponentials. The corresponding parts of the solutions consist respectively of t times sums of periodic terms, period 2x, multiplied by the fundamental exponentials to the first degree, and sums of periodic terms, period 2ir, multiplied by cubes and non-canceling third-degree products of the fundamental
are
of the type (c) are homogeneous of the third degree in the They consist of terms of two classes, the first of which
exponentials.
[8]
part of the solution coming from the terms (d) consists of terms of two kinds, the first depending upon those parts of x 2 y2 and z2 which contain X as a factor, and the second depending upon those parts
,
,
The
of
y2 and z2 which are independent of X. The parts of the solutions corresponding to the first are homogeneous of the second degree and they are sums of periodic terms, period 27r, in the a and the c multiplied by squares and second-degree products of the fundamental exponentials, and some of these products contain t as a factor while
x2
{ ,
others do not.
(d)
The
other parts of the solutions coming from the terms of those coming from (c).
230
110.
PERIODIC ORBITS.
It will
be necessary to use
members of the first two equations of (13) involve only z, it follows that x and y are even functions of c and c,
t
of
Since the right member of the third equation of (13) is an odd function z, it follows that z is an odd function of c, and c 2 taken together, and
members
for
of the first
two equations
4
t
of (13) vanish
identically for
x = y = z = 0,
a
v
but not
vanish
identically
4
=
c2 5^0.
a,=
[12]
it
= a = 0, e^O,
Since the equations reduce to those having constant terms for e = 0, follows that the sums of the periodic terms, period 2-k, reduce to
e
constants for
= 0.
Symmetrical Periodic Orbits. The with the period 2t. Consequently the period of the periodic solutions, if they exist, will be T = 2nw, where n When the initial conditions are such that the orbit of the is an integer. infinitesimal body is symmetrical, as defined in 106, then sufficient con111. Conditions for the Existence of
t
a!+a = 0,
2
a3 +a4 = 0,
2/(f) -2/(0)
series in
c,+c8 = 0,
(31)
z'(f ) -x'(0)
=0,
=0,
. .
at cl} Cj, and X, and In order that they may vanish identically with 4 for c and have any solution ,aiy lt a, C;, vanishing with X, aside from a4 this one, the determinant of the coefficients of the linear terms in a,
a lt
x
,
= a,
.
= a = c = Cj = 0.
,
Cj
and
c2
must vanish.
It follows
from
(28) that
uM = -uM,
",(0)
i(f)
u3
.(f)i
=
-tt,
<(0) =
u' 3 (0)
+ u' (0),
2
*((f)- + 4(f),
u' 3
==
,(0),
(!)
(|)
= + u'M,
= -v! 2
(0),
(|)
= + u\ (|)
t\(0)
i>
(0),
e,(f)-+*(f),
vt
v[(0)
^(|)=-fKf),|(32)
vAo)=+v
(o),
(f)-+4 (|),
(|)
=
*(o)*-:(o),
u<(0) =
*(f)--;(f),
,;(!)--/(!),
wM -+*(<>),
7l
+^(|),
(o),
231
is
*
i
:
A = A A2
1
(33)
A,=
232
PERIODIC ORBITS.
where
C^aV^lV^+v'^)
The determinants become,
^1
)
C-/*(l)+*(f)
reductions,
after
some
^3
A-4
^
*(
<!)
(37)
A=-2[w V^T^(f)-^(f)]^(f).
Since D^ and Z)2 are determinants of fundamental sets of solutions of linear differential equations for the regular point t = T/2, they are distinct
from
Therefore their second factors are not zero, and equations (36) can be satisfied only by
zero.
*
-e
r-v^I* 2
=0
>\/_i-
- u vC
or
[e
=0.
(38)
If either of these
equations is satisfied, A is zero. In order that one of equations (38) shall be satisfied
it is
necessary
that either
aT = 2N
Since
ir,
or
wT = 2N
2 tt
(N
integers)
(39)
T = 2nir,
where n
is
an
become
(40)
or
co
=
n
Hence the conditions for the existence of the symmetrical periodic solutions in question can be satisfied only when a or u is rational. These quantiin are series in e and (25), ties, given power they depend upon /x and /*<> and
the
er
the transformation n = na +\ is made. Since and w are continuous functions of n, /x and e, the rationality of at
way n
is
generalized
when
and
co
depend upon
/*, ju
they are independent of e. In any case |X| can be taken so small that the series will converge, but the periodic solution does not belong to the physical problem except when
Theoretically Suppose the series diverge for this value of X. the values of the coordinates for this value of X can be obtained by analytic
X
= M~ Mo-
which
continuation with respect to X as the argument from the periodic solution exists for a smaller value of X. There is an exception only if the function has a natural boundary, or if X = /x Mo is a singular point.
233
Orbits.
Existence of Three-Dimensional Symmetrical Periodic and n are Suppose w is the rational number u = N/n, where
The
relatively prime integers, and take T=2nir. Suppose aT is not an integral of 2t. Then A and the first four multiple x equations of (31) can be for a a solved and X, vanishing 4 uniquely as power series in c lt c 2 lf
^0
a 4 are even property [11]. taken together, by property [9]. When these results are substituted in the last two equations of (31), they become power series in Cj c2 and X. These series are of odd degree in c, and c2 taken together, by property [10], and therefore vanish identically for c = c2 = 0. The a 4 does not change the linear terms, substitution of the values f or a x for the first four equations were even in c and c2 alone. Let c2 be eliminated by means of the fifth equation of (31). Then c, is a factor of the result, which has the form
C!
2
x
.
= c = 0, by
c2
The a
and
=c
The
solution c t
[a 01
X+a c+
20
].
(41)
is trivial and we are interested only in those obtained by the other factor equal to zero. The second factor set equal to zero setting = = X 0. If o 01 is distinct from zero, solutions for c x as power is satisfied by c x
powers of X certainly exist. If a y is the first aJQ which does not vanish, then the solutions are expansible as power series in X v\ In particular, if a20 is distinct from zero the solutions are expansible as power series in X*. If the number of solutions is odd, only one is real; and
series in fractional
,
even, only two are real, and these are real only for positive or negative values of X according as a01 and a,, are unlike or like in sign.
if
number of solutions will be even, for there nature is nothing of a dynamical by which to distinguish the two sides of the xy-pl&nc. Consequently, if any initial projection gives rise to a periodic orbit, a symmetrically opposite one with respect to the xy-plsme will also
It
is
produce a periodic orbit. To prove this the It remains to show that o 01 is distinct from zero. terms of the second degree in the a c,, and X must be considered (108). It follows from the form of (29) and (30) that a 01 depends only upon the 2-equation, for it is not changed by the substitution of the solutions of the
t ,
Hence ai
(42)
z"+[A+3Aecost+
where
It follows
-]z 2
= [A'+3A'ecost+
,_
-} Zl
^_
of
is
w lt w
a power series in
equation (26), of Chapter VI, where it being distinct from zero, e can be taken so small that the series for o ul
distinct
u and the general theory of 29 e. For e = it was given was shown to be {-VfAaA'. This
is
from
zero.
234
Similarly, a^
small.
e
is
PERIODIC ORBITS.
made up
e,
of
converging series in
and
is
a constant part distinct from zero plus a therefore distinct from zero for e sufficiently
x
and let the value of c obtained from solving be the resulting equation cf; then let c = c{ (l + y). It is easily found from (41) that dy/de is a power series in e and y which is distinct from 0> zero for y = e = Hence the provided X has such a value that cj ?^0. solution of (41) can be written in the form
in (41)
0)
x
Suppose =
Cl
\*p(\),
(43)
e.
where p^X')
is
this result in the solutions of the first four equations of (31) for the a, as power series in c, and X, we have a, , . a 4 expressed as power . ,
On substituting
X1
.
a t contain only even powers of c they have X instead of X* as a factor after Cj is eliminated by (43). The expressions for the coordinates become, since x and y are functions of c\,
series in
,
. .
But
since at
x ,
x=
\P^;t),
i
,
|f-XP,(X*;0,
are
= X*P
(X*;0,
(44)
where
P ,P
l
and
power
series in X*.
Since the problem is dynamically symmetrical with respect to the xr/-plane, a solution symmetrically opposite with respect to the xy-p\a,ne
exists for all
X*,
|X|
sufficiently small.
series in X*.
2/
series in
X-X^OjO,
where
Q,,
= X&(X;0,
= XQ
(X;0,
(45)
power series in X. We suppose that e>0 and therefore that the finite bodies are at their The infinitesimal body crosses the z-axis perpendicularly perifoci* at 2=0. at t=0 and at t-T/2. It follows from the symmetry of the motion that it can cross the z-axis perpendicularly only at the end and middle of the true period. Therefore if n and N are relatively prime, T = 2nir = 2Nir/u)
2
,
Q and Q
are
is
The two
cases
I,
n even, and
II,
n odd, merit a
little
further discussion.
the second
it
n even, odd. The infinitesimal body crosses the z-axis = at t and also at t = T/2 = mr =2nV, where n' is an perpendicularly Since at both of these epochs the finite bodies are at their periinteger.
Case
I,
the infinitesimal body crosses the z-axis perpendicularly only when the finite bodies are at their perifoci. It follows from this that the infinitesimal body crosses the z-axis perpendicularly at the same point but in
foci,
= T/2.
To
'Perifoci will be used to denote the points at which the finite bodies are nearest each other, and These points correspond (o perihelia apofoci those at which they are most remote from each other.
and aphelia
in
planetary motion.
235
suppose the two points were different. Then we should have four solutions corresponding to X* at each of the points, and it is known that are but two. The values of z'(0) and z'{T/2) are opposite in sign thei* because otherwise the period of the motion would be T/2. It can now be shown that in this case the orbits obtained by taking
prove
it
the two signs before X* are geometrically the same one. Consider the orbit defined by the positive sign before X*. At the time t - T/2 the infinitesimal body crosses the x-axis perpendicularly at the point at which
it
crossed at
It has
= 0, but
as an origin of time
larly.
in the opposite ^-direction. may consider T/2 for defining orbits which cross the z-axis perpendicuis
We
of motion, and at t = T/2+ T/2 = T the infinitesimal body will again cross the x-axis perpendicularly with the opposite z-direction, viz., with that which it had at t = 0. Since this orbit was unique, it follows that the two orbits
which correspond to the double sign before \ are geometrically the same, but that in one the infinitesimal body is half a period ahead of its position in the other one. That is, changing the sign of X in the solution is
h
equivalent to adding T/2 to t. When the solutions are actually constructed and are reduced to the
trigonometric form, they involve sines and cosines of (jo:-\-k)t, where j and k are integers. Since x, y', and z' are even functions of t, they will
involve only cosines, and since x', y, and z are odd functions of t, they will involve only sines. Since x and y are even series in X*, it follows from
[(;+*)] s
These
identities are satisfied
if,
[0*+*)(+r/2)]
if,
and only
cos
therefore
(n'
an integer)
(jw+k)n'=j
Since
+kn' = p
(p
an
integer).
this relation that j is necessarily even. Similarly, in the case of the series for z the relation
is it
odd
follows
from
sin
[0+AO*]
from
- sin
[U*+k)(t+T/2)]
t
must be
fulfilled.
It follows
this identity in
2 -
that
integer).
(ju>+k)n'
=j
^ + kn'=
(p
an
This relation can be satisfied only if j is an odd integer. There are solutions in this case which have not been obtained by the intersect the x-axis analysis as given. The orbits which have been discussed
and obliquely, perpendicularly when the finite bodies are at their perifoci, = when the if at all, when they are at their apofoci. Similarly, supposing t
236
finite
PERIODIC ORBITS.
bodies are at their apofoci,
it
with the same period in which the infinitesimal body crosses the x-axis perpendicularly when the finite bodies are at their apofoci, and obliquely, if
at
all,
when they
14.
In the present case the infinitesimal body crosses the x-axis perpendicularly when the finite bodies are at their perifoci and also when they are at their apofoci, because T/2 is an odd multiple of -k. X is geometrically distinct from that is even, the solution for If To prove this, we note that |X| can be taken so small that the for X*. If e is small, z' has sign of the series for z' is determined by its first term. Now the first its sign determined by the constant parts of w and w2 aVZ ^' ws/zzlt e and e~ of these first have the
1
Case
II,
n odd.
parts Since
terms,
viz.,
N
t
is
an even
= T/2.
, period 2ir/u. follows that the integer, sign of z' is Now suppose X to increase to the value
it
belonging to the physical problem. If the sign of z' (T/2) changes it must do so by passing through zero. But in this case, since z(T/2) is also zero, Therefore z' (0) would also be zero and would z would be identically zero. change sign for the same value of X, and z' would still have the same sign
Consequently, the z-component of velocity at t=T/2 is not equal to the negative of that at t = 0. Hence, when n is odd and X* are geometrically distinct, because a is even, the orbits for +X* and single orbit can cross the x-axis perpendicularly but twice.
at
t
and t=T/2.
If
n and
the infinitesimal
pendicularly, as before, both when and also when they are at their apofoci.
body
But though in this case the sign } = of z' at t =0 is opposite to that at t T/2, the orbits for +X and -X are distinct; for otherwise the motion of the infinitesimal body would be precisely the same while the finite bodies were moving from perifoci to apofoci as This is impossible while they were moving from apofoci to perifoci. because t enters the differential equations differently in the two cases.
1
we set up the problem taking = when the finite masses are at their apofoci, we shall find similarly two solutions; but they will be identical with these, for in these the infinitesimal body crosses the x-axis perpendicularly when the finite bodies are at their apofoci. The numbers n and N have so far been taken relatively prime, and T=2w7r. We shall now inquire whether there are other solutions with the = period T' kT, where k is an integer. The determinant Aj vanishes for this
If
t
value of
t. Solutions having this period certainly exist, for they include as those with the period T. cases Proceeding as in finding equation special (41), the corresponding steps are taken, and it is found that in this case a 01
former values only by multiples of 2r. Therefore the with the period T' is the same as with the period T. Hence number of solutions
their
new
solutions
of those considered.
237
existence of the symmetrical three-dimensional proved except for a discussion of the convergence
The
which have been employed, a matter which must now be taken of the difficulty will first be pointed out. up. Equations (13) were integrated as power series in the a c and X. It was shown in 14-16 that for any preassigned T the moduli of these parameters can be taken so small that the series converge for all <! ^ T. The limits on the moduli of the a,, c,, and X are functions of n, /z a and e. But T can not be taken arbitrarily in advance, for it is a discontinuous function of w, which is in turn a function of n, ju and e. It is not evident a priori that values of and e, satisfying the relations n m = X, o=/(m, Mo> e ), exist such that fi, n all the series which are employed are convergent. The final parameters of the solutions are p, /z e, and the mode of Suppose the ratio of the finite masses is given, generalizing /z is arbitrary. that is, that n is a fixed number. Suppose also that the mode of generalhas been determined. It will be shown that values of ju and e exist izing fi
The nature
t ,
( ,
such that the series all converge for X = /x In equations (25) it was shown that
co
/v
= \/Z+w c+w e
1
where
VA
and the
co,
are functions of
/z
shows they are continuous functions of /* Suppose for any /* such that |m Mol=i>0 the series for w converges for where depends on e^ It is easy to show from the nature of the \e\ ^i7 u dependence of the series for w upon n and ju that q > 0. Take any particular /xi" such that \n /4|4 and suppose that, while e runs over the range to rj u the value of to (n e) runs over the range co 04", 0) = VA to w (/4", ij,)For brevity, let w and w H> represent the smallest and largest values of w. .are not all identically zero, and it An examination shows that u, w2 " Let mJ take all real values such u a> 0. > follows from this that Let that |/x Mo"|^ e and find the corresponding values of w and co The value of Cj can be and a>, be the least co w be the greatest co 0) a) taken so small that w, -w< >0, and hence from the continuity of a as a function of n it follows that u takes all values satisfying the inequalities r ^w ^wj". Take any rational value of w depending on p, fi and e which
tion of the functional relation
i\ x
l ,
<0>
(0)
<0)
(1)
<0)
(1)
(1)
say
= No
n
where
(46)
and n are
Then determine
T=
!F
by
the equation
7>2 7r=^Wo
(47)
238
PERIODIC ORBITS.
T Since they vanish consider the series (31), in which we put = in 14-16 shows the discussion for a, c, =0, that, for any values identically and p>0 of X and e for which the differential equations are regular, r 4 >
Now
T=
series
converge for
all
,
^ T^ T
X,
provided
The
,
r,
and
p<
are functions of p
-f X,
and
e.
We may
|
and we shall take |pp = e and are both distinct from zero and have such values and where <! e ^ Let rj and p, that the differential equations are regular for |p p ^i e<jr?i e take all values be the least values of r< and p as p and satisfying the ine^ e = ^i- It has been shown that when the solutions qualities |m Pol of (31) exist they have the form
eliminate X
77,
,
by the
relation p
77,
=p
0)
<0)
=u
t
a
t
= \p
J
t
(X
e)
!
c,=
-c = X*p
2
(X
e) ,
(48)
where the p are power series in X and e. give \a \<rf\ |c,|<p, provided 0<|X|^
(0)
t
These
2
series will
converge and
t
.
^e
1 ,
for all
e<ij
t
That
(
is,
= 0,
Choose any X satisfying the inequality |X|^j and determine p by the relation p p = X. It was shown above that for this Hence for this p /n p there exists an e < ^ such that w (p e) = w = the existence of certain are that and e all the series employed is, convergent;
,
is
proved.
The question might be asked whether the solutions exist if both p and e It is not easy to make the answer in general, but are given in advance. X opens a the mode of generalization of p into p and p clear that it is This is so unless, indeed, the realm of validity wide range of possibilities.
independent of this process. conditions are given such that the motion may be represented by
of the results
is
Suppose
is
p, e,
and the
initial
periodic.
The coordinates
x=F
{n,t),
y = Fi ( x,t),
t
t-F (p,f).
t
Consider the expansions of the functions F, as power series in X, where = ju 4-X in at least part of the places in which p occurs. It is clear that the /i realm of convergence of the series depends upon the manner of this trans-
For example, if F = Ft = F3 = sin/(l+X)(2+p), and if we put l+M=l+p +X, 2+p = 2+p, then the functions are expansible as power series in X which converge if |X| <l-fp where X and p are subject to the
formation.
l
,
if |X| <2+p where X and p are subject to the For + \=n. example, if p = l/3 in the first case the scries converge if |X|<2/3, and in the second case if |p|<7/6. Now it is clear from the variety of ways in which X can be introduced that convergence of the series can be secured in many, if not all, cases when p and e are given
in
advance.
239
16. The Existence of Two-Dimensional Symmetrical Periodic Orbits. The last two equations of (31) are satisfied identically by c = c = Q, in which
y
have to consider the solution of the first four equations for a a 4 in terms of X. The determinant A, equation (33), now becomes simply A = A It follows from (34) that the condition A 1= can be satisfied only by <r=N/n, where and n are integers which we shall suppose are relatively prime. We shall solve the first three equations of (31) for a2 a 3 and a 4 as power series in a and X. This solution exists and is unique provided the determinant of the coefficients of the linear terms in a2 a and a4 is distinct from zero. It follows from (34) that this determinant is
curves.
.
become plane
x
We
D=-A
which
If
is
T
3
distinct
3
[e^-eA is zero.
3
-?],
were zero we should use the first, second, and fourth equations of The determinant of the coefficients of the linear terms of a2 a,, (31). and a 4 in these equations is found from (34) to be
,
d
t
*(j)|V*-.Tf],
3 3
which is distinct from zero unless v3 (T/2) is zero. But A and v (T/2) can not both vanish, for then D of (37) would be zero, whereas it is distinct from zero. Therefore a 2 a and a can always be eliminated by means of three of equations (31), leaving a single equation of the form
,
=a
[ft 1
t
X+/310 a +/3 11 a
1
X+&
aJ+
(49)
This equation carries a as a factor because equations (31) are identically = a4 = 0. satisfied by a, = Now consider equation (49) The trivial solution a = will be rejected, and we shall attempt to solve for a as power series in X I/J where j is an
.
from zero such a solution certainly exists, and j The coefficient /301 is determined by the first & which is distinct from zero. is a power series in e whose term independent of e was found in Chapter VI, Consequently for e sufficiently equation (26), to be distinct from zero. small /301 is distinct from zero, and the solutions exist.
integer.
If
j801
is
distinct
Now
consider
/3 10
It
was shown
of e
is
in
Chapter VI,
It will
zero. of this coefficient independent first three equations of (31) for of the solutions The zero. identically first of the no terms ai} a3) and a 4 contain degree in a x alone, for, if we
eliminate a2 and a 3
by means
of the first
A
It follows
T
3
[e"
from property [3] of 108 that the terms in the solutions of the second of the degree in the a, do not contain t as a factor. Hence, the only terms second degree in the a which are not periodic with the period T are those
240
PERIODIC ORBITS.
which contain a\ or a] as a factor. Hence the solution (50) for a, has a term of at least the third degree in a as the term of lowest degree in a,
t
alone.
a2
a3
first three equations of (31) are solved for in the last one, the result contains no term
is, /3 10 is
That
identically zero.
VI,
not necessary to consider j3 u if /320 is distinct from zero. In Chapter 100, it was shown that the part of /3 20 which is independent of e is
distinct
from
zero.
It
is
also distinct
from zero
y=
x=
\ i P(V;t),
\*Q(\i
One value of the double sign belongs to the orbit when the body crosses the axis in one direction, and the other when it
the other direction.
the expression a
1 1
infinitesimal
crosses
it
in
is
even or odd
in
Case
t
I,
dicularly at
n even. The infinitesimal body crosses the x-axis perpenand at t=T/2 = mr = 2n'ir, where n' is an integer. Since
2nV
and
is an integral multiple of the period of revolution of the finite bodies, since the infinitesimal body crosses the x-axis perpendicularly only at
the beginning and middle of the period, it follows that it crosses the x-axis It follows, perpendicularly only when the finite bodies are at their perifoci.
as in the case of the three dimensional orbits, that the orbit belonging to X* is not geometrically distinct from that belonging to +X ! in particular, that
;
one orbit can be obtained from the other by increasing t by T/2. Consider the terms which are even in \K They are not altered by changing the sign of X*. Consequently in these terms
sin
cos
an even
integer.
sin
cos
[U*+k)t]
ft
Z[(j<r+k)(t+T/2)]
from which
If
body perto the when the x-axis finite bodies at their are pendicularly apofoci, wc should find similarly two geometrically identical orbits in which the infinibody crosses the x-axis obliquely when the That is, when n is even there are two perifoci.
tesimal
finite
we
follows that j is an odd integer. should set up the problem starting the infinitesimal
it
classes of geometrically
distinct orbits of given period which intersect the x-axis perpendicularly; in one, the periodic orbits intersect it thus only when the finite bodies are at
their perifoci,
and
in the other
only
when they
241
Case
II,
n odd.
If in
n is odd
the infinitesimal
body
crosses the x-axis perpendicularly at t = and the finite bodies are at their f peri oci, then it also crosses the .r-axis perpendicularly at t = T/2 when the
finite
bodies are at their apofoci. If is even, the infinitesimal body crosses the x-axis in the same direction at t = and t = T/2. Hence the
X are in this case geometrically distinct. If is odd, the orbits for -f-X* and X* are also distinct, because otherwise the motion of the infinitesimal body would be the same while the finite
orbits for
+X* and
bodies are moving from perifoci to apofoci as while they are moving from This is impossible because t enters the differential apofoci to perifoci.
body
crosses the x-axis perpendicularly when the finite bodies are at their These orbits, therefore, apofoci, and also when they are at their perifoci.
when
It
1=1
2 y y= 1=\
2J X',
z= 2 ^,X l=i
(51)
where the xj} yS) and zt are all periodic with the period 2ir/w. It has been shown that symmetrical orbits exist, two for each value of X, and their coefficients are uniquely determined by the periodicity conditions and It follows from this last relation and from the fact that the series z (0) = 0. for x, y, and z converge for all X sufficiently small, that each Zj(0) sepa|
rately
is
zero.
It follows
i
120. Coefficient of X
z"
from
term
is
defined
by
+ [A + 3Aecost+^Ae (l+3cos2t)+
is
-Jz^O.
and
its
(52)
This equation
is
53,
general solution
of the
form
z,
= ciV^'u^e; t) +c
2 s
( l)
a e- ^=ll wi (e;
t),
(53)
where
w = v/ A+a)
while
tit,
e+w e
w
x
-|-
w = l+w? e+w ?e +
) (
only in the sign of V^i, and where each w[ is separately periodic with the period 2ir. Since (53) is unchanged by changing the signs of both t and l, it follows that in the expressions for w and w 2
differs
from
J)
242
PERIODIC ORBITS.
from the condition ,(0) =0 that c[ +c^ = 0. The coeffiIt will now be shown that co is a series in even powers of e. 3 Now the cient of 2, in (52) is derived from the expansion of r^ and r2 \ e and expressions for ^ and r2 are unchanged if in them e is replaced by
l)
the coefficients of the cosine terms are real, and those of the sine terms are Since w\(e; 0) = wi (e; 0) = 1 for all values of e, it follows purely imaginary.
a solution of (52), then also M) ( Since any solution can be expressed e e; t+w) is a solution. in in of the two solutions terms (53), and since the solutions now linearly under consideration hold for all e sufficiently small, and, for e = 0, differ only ^ltT it follows that for any e they differ only by a constant by the factor e^,v
t
by
t-\-ir.
Consequently,
if
eP^^^'w^e;
t)
is
factor.
Therefore
-*^=>,
This relation
the period 2t,
ef
(e; t)
- Ceu( -" v^
Tc,+,r,
w; l
(-e; t+v)
e.
= 0.
1
is satisfied
it
and
Since
is
periodic with
{e;
t)-e
wi -')V=Ti
'C^- t)VzrT,,+ir) w
(-e; t+v)
l
= 0.
Hence we have two homogeneous linear equations in efaV>v^Tl w {e; t) and t-)N^To+x) w? (; e ^_ e t-j-ir), and as these quantities are not identically zero,
.
the determinant
*
'
*
I
w(-e)\f=T2K
-i(i)v^iJ
definition
w must reduce to
2
.
VA
for e
= 0, we
we
have
Upon
53,
find
+ f_
,
9A(l-15A "
+ 18A* + 8A _
s
9 A'
8(l-iD(l-4A)
!
!
(l-4A)
iC0St+
,, 184
(1-iAY
V^T Smt
(54)
+ 9A(l-34-2A 8(l-A)(l-4A)
,
82t
It will
coefficients of cos,;7
have
as a factor
This
is
equations, and an examination of the process of integration, as explained in 53, shows it is also a general property of the solutions. If we had solved the differential equations for x and y, we should have
v ,
found that these quantities are identically zero because of the periodicity conditions to which the solutions are subject.
243
from
(13)
and
are defined
by
-]a;2
2 X2-2y' -[l+2A+6Aecost+
-[6,4esin<+
-]?/2
-]?/ 2
=r
2 y2+2x'
[6
Ac sin t+
]a; 2
[l
(55)
SAecost+
=Zj,
where
X =[^B+6Becost+
i
]**,
]**,
F = [3Besin*+
2
r>
_ + +
1 'l
Mo
Mo
(0)4
'
r (0)l
'2
the signs in 5 being the first, second, or third pair according as the point By means of (53) and (54), we find (a), (6), or (c) is in question.
z\
+(c-)^
+2cl
,
,
C
f,
'{l-[ 1
t
-^z (l-cosi)]e+
(\A
}
j
(I)N2
-2o,v
.,
12A' A
,1
(56)
+ 6Bcc
+3B(c2 '>)'e
{i
(2
+ [1
^A + f5|4co
,)
S(
(;+
j
v^Tsm^-f
]
j,
F,=
35(0
2w,/r7
'{[sin]e+
j+6c'
c2
1)
j[sin<]e+
1)
+3flc2
V ^'{[sin*]e+---i
2
.
character of the solutions of equations of the type to which (55) was It was there shown that they consist of the discussed in 29. belong function plus terms of the same character as X, and F2 complementary It follows from the hypothesis that the imaginary characteristic exponent crV l are incommensurable, l, arising in the solution of (55), and wV that the constants of integration must all be put equal to zero. The particular integrals can be found most conveniently by assuming their form with undetermined coefficients, and then defining them by the conditions that the equations shall be identically satisfied.
The
244
PERIODIC ORBITS.
need the following properties of the solutions of equations u>/^rT and (55). They are homogeneous of the second degree in c"'e -u> %/=7 The terms in the solutions multiplied by (c^') 2 differ from those c e 2 l, because this is a property of the multiplied by (cj ) only in the sign of V
shall
'
We
<i>
l)
right
If throughout equations (55) of the differential equations. into we change y2 into -yt v/^T, and t into -t, the equations are in the expression for xt the coefficients of the cosine unchanged. Therefore,
,
members
V^T
terms are real and the coefficients of the sine terms are purely imaginary. The terms in x, having c^Cj" as a factor involve opposite is true for y2 li v^7' and e~ wVZrr while only cosines and are independent of the exponentials ef those in yt having c"'^" as a factor involve only sines and are also indepen^ It follows from these properties dent of the exponentials ewv/trT and e~ wv/ T and (? = - d? that x2 (0) = y2 (0) = 0.
The
',
'
'.
Certain divisors are introduced in the integration of (55). When the right members of these equations are omitted, the solutions are of the form
(24).
On
using the
method
we
find
a;
M)
F
2
,
(t-i,
...,4),
is
where
(t)
and where A
It follows
the determinant
(24).
of
The expressions F,(t) 18 applied to this case that A is constant. contain terms of the types given in (56). Consequently the a, contain terms of the types
dt;
or,
A a, =
+
-
@^ ***=
^
J
1-4^
'
[jV=Tcos./*+28in;*]
[
+ ^jj^ e"*"^7 +^
;
-2a> y/^\cosjt+jsinjt\
sin
jt]
ff.4
[jyT^T cosjt+2w
+ SlS.
C, sin jt.
2 4 2 appears in terms involving t in the form Therefore, the divisor j e *2a/=T/cos^ an(j ^he di v i sor j j n those involving t in sin^. It was seen that in the expression for z l the coefficients of cosjt and sin jt carry e' as a factor. Consequently it is true also for z\, and therefore
and
245
(51) that
X'\
It is
found from
(13), (14),
and
z"+\_A
+ 3Aecost-]
pw
-\
]z3
= Z,,
H
\x 2
z,
Z =
3
+{pa
1
k +| 3
if ]
(57)
+{l[
The
^+
'
}^+{6^sin<
j&Zi
and
coefficients of z,, a^Zj, and z\ are series involving only cosines, the coefficient of y2 zx is a series involving only sines.
Now
member
The
=C
e uV=T,
w + C, i~"v=t
1
where
C and C
1
variation of parameters, the conditions on C, and satisfied when its right member is included are
By C
2
the
method
of the
e"^1 w
'
C[
+
1'
grwsn w
t 2
^=
2
o,
e"^71
'
v^MWj+wJlCl+e-"^ [,
(58)
3
.
Upon
we
',
get
A C( = - w Z e""^77
t
3
A C2 =
+w Z
1
e uV
=~ lt
,
(59)
where
A=
It follows
UVI W
-\-w[
WV
1^2
+ ^2
w and w
x
it
from the results of 18 that in this case A is constant, and since are 2 power series in e the determinant A is also a power series in e. In order that the solutions of (57) shall be periodic with the period T is necessary and sufficient that the right members of (59) contain no
constant terms.
We
must
w Z and
2 3
wZ
x
e'"^
'
respectively,
and
Z Upon substituting the value of z from of the coefficient of e wv/=T coming from the constant see that part (53), WjZ, is the constant part of the product
x
we
'
-c^Wittdpa
pn +3fiecos<+
(60)
<J-
246
PERIODIC ORBITS.
uV~' coming from part of the coefficient of e~
The corresponding
+w
Z,
is
+cj
It follows
;,>,
J pai
pr +35ecosf+
3
1
(61)
-J-
from the properties of w and wt that their product involves only Hence the cosines and that the coefficients of their product are all real. constant parts of (60) and (61) are power series in e which, aside from the coefficients c[ and Cj ', differ only in sign, and the parts independent of e
l) 1
are respectively
Ci
yr
(o>3
r (o)ij,
t^
\r
m)i
mi
Consider
xt zx as
in the
now the terms coming from a factor. The constant parts of the
that part of
which contains
coefficients of e"""^*
and
e~ Uy/=u
products
-w^z^SB+UBecost
and
+w x z
l
[ZB+\2Becost
It follows
to,
that
u>,[3.B
+ 12Bcos*+
and
w [3B+12Becost+
2
differ only in the sign of l, which is a factor of all the sine terms, while the coefficients of the cosine terms are all real. These products In the product x2 z x the do not involve the exponentials e wV it and e~' vr=Jt 2 ^ coefficients of e""^' and e~ wVZ= are multiplied by (O'cJ" and c"' (c"') respec-
tively.
It follows from the properties of x 2 and z that in this product differ only in the sign of the coefficients of (c^Y 4" euV^' and cf (cj )** l, which is a factor of all the sine terms, while the coefficients of the
x
"^'
all real.
The
^
)
tv i
i'
['3B+ 12/?ecosJ
and that
AjQO&jt
lBiSWJt,
(ci )
c? [aJ cosjt+
lbjBUijt].
+ \2Becost +
x% z containing e~
x
us r=ru
+AjCOsjt
- -f^lBjsinjt,
c^ic^fiaj cos jt
- V^Tbjsinjt].
247
The constant
Since these properties hold each term individually, it follows that the con~ stant parts of the coefficients of e"^1 and e~ Va=u in w2 x2 Zy[%B+ and
'
-\-w l xi z 1 [ZB-\-
'
-(<f)*4
[Ci+C+
of
+c[
(cmC +C e+
a
1
].
(62)
The corresponding
third
same
(57) shall
O-JTcf+IKcfW,
where
= Kc?+Lc[
l)
(c i
y,
(63)
and
corresponding work
are constants and power series in e. It follows from the in Chapter VI that both and L have terms independent
from
zero.
The
-e -0,
}
K+Lc?c? = 0.
(64)
The former
sidered.
Since
= cj"
x=y=z=0
c=^| =
and
(
[X,+X
],
(65)
and c " are determined except as to sign. The orbits which correspond " are geometrically identical or distinct according as to the two values of c T is an odd or even multiple of 2w (see 113-114).
ej
{
After the conditions (63) are satisfied, the solutions of (57) are
z3
(e
t)
+P?\t) +P?(0,
(66)
where
Pf
is
and homogeneous
in
l>
c{
{
WN/^rT '
and
ce~ uV=l
',
and where Pf
uVr=lt uV=It and In the right is homogeneous of the third degree in c "e c^e~ member of (57) the coefficients of all cosine terms are real, and the coeffi.
cients of
all
multiplied by
multiplied
3)
by c,
(O (O
3
3
,
(c"')
^"
2
differ
and moreover those which are from those respectively which are
c"' (c"')
P{
and
Pf
3)
that
(53)
and
(54).
The P[ and
3)
are entirely
known
3,
functions, while cf
and
c"'
are
2,(0)
=c5
248
It
PERIODIC ORBITS.
follows from the properties of the expressions Pf (t) and 3) u that Pr(0)=P; (0)=0. Hence ef and <f are subject c| =-c; condition
l)
(t)
and
to the
*(0)-<f+<f0.
(67)
Therefore but one undetermined constant remains in (66). The divisors introduced at this step can be found as they were in the preceding step. They are the determinant of the fundamental set of 2 2 2 solutions of the z-equation, f ui andj 9oj the j 2 co 2 appearing with
,
'
jt,
and the j2
*
9
e'
The
coefficients of
jt
contain
as a factor.
123. Coefficients of \\
The
a-2yl-[l+2A+6AecosH
y"+2x[
where
]z4
4
[64dsin*
-\
]y
=X
i ,
[6Aesint+
-]x
[l
A
^
3Aecost +
~\y i
=Y
(68)
,
are of even degree in ea,v/ T( and e~ wV=Tl considered together. It follows, as in the case of x2 and y2 that in the right member of the first equation the coefficients of all cosine terms are real, and those of all sine
A
and
terms are purely imaginary the opposite is true in the right member of the second equation. The coefficients of cfej8 and (c"') 4 differ from those of c" cf and (<')* respectively only in the sign of V l. The coefficients of (0*(0* are real, independent of e?" 1 and tT^r=u and involve only cosines.
;
1
'
The
It follows from these properties. = = that c[ x[ (0) properties yt (0) 0. (%*, The divisors introduced in the solution are the determinant of the
same
and
1)
fundamental
and j in connection with the terms involving e* 4WN/TrT'a? jt, e* "^11 jt, and ljt respectively. The coefficients of these terms contain e'as a factor.
3
*
x and ^-equations,
16w2 j 2
,
4co
2
,
to be able to
make
It is necessary to consider this step in order the general discussion. The differential equation defin-
ing z5
is
g+[A+SAmo6t+
where
z\z3
it is
]z6
(a)
=Z
(69)
found from
(14)
and properties
multiplied
and
,,
(d) of
106 that
a
1
&
is
made up
,
of real cosine
series
into
x2
z3
^,
z,.
x\z
y\z ly
z\,
from
in e"
v
~' and
of real sine series multiplied into y 2 z 3 and the properties of z n z3 x2 yit xt , and y t that 5
,
,
and
*/ 4
It follows
is
of
- "^^'
odd degree
249
of the sine terms purely imaginary; that cf and cf enter 1 that the coefficients of cf\ (e?>)*f ffc'V, (e?*)*^) differ from linearly;
,
and those
the. coefficients of cf, (O'cf, c'?c?c? the sign of l; and that the terms
respectively only in
cosjY
and
sinji
have
e'
consider the solution of (69). The discussion proceeds as in the case of (57), and it is found that, in order that the solution shall be periodic, wi Zi e~ uV^T and Zh e uVtrT must be zero. It the constant terms in x
'
Now
+w
<
[A.+A^^^+AAc^^+A^yic^^O,
where
last
(70)
J
Alt
e.
Upon
equation of (64),
we
get
[A,
[
Since
3)
^] A ~ ^]
c=
<f
c
+ 4l
<f
"
+ Af <f = 0,
"
)+
^iF c
)+
^
l
(71)
c
"'
=0
c=c
"
and
and define
c[
provided
Ax
is
K
e
A*K
L
=
of Chapter VI the corresponding from zero. Since the coefficient in the present case = 0, it is a power series in e and reduces to that of the former case for e follows that it is distinct from zero for e sufficiently small. When equations (71) are satisfied, the solution of (69) has the form distinct
from
zero.
In the case
coefficient
was
distinct
zb
= c? (T^'Wt+c? e-^-^Wt+PrW+P^W+PFit),
>
(72)
Uv/ T and cf e _Wv r taken where Pf(i) is linear and homogeneous in c" e together, and P(t) and P(t) are homogeneous of the third and fifth in e w ^~' and e~ uVrrTl taken together.
' >
'
degrees respectively
It follows
'
from the
facts noted
above that
coeffiare interchanged with c, 4, c v^7, -f cients of the cosine terms are real; hence the solution (72) has the same = - c{", cf = - c,0> it follows that Consequently, since Cj" properties.
f
V^T
if
M
c}
,
= P?(Q) = P? (0) P^
6,
(0)
= 0; and
since z6 (0)
= 0,
that
(73)
if+cF-O.
250
125.
PERIODIC ORBITS.
lt
z^-i,
_,,
yt
-2
x and ^-Equations. Suppose z, xt have been computed and that they have the
for the
,
following properties:
(a)
The x2J and y2} are even functions of e"^77 and e -0"^*' taken
'
together,
and the
(b)
z 2J+x are
odd functions
of e?"^'
and
_Wv'^T
'
taken together.
In the x 2l and z2J+i the constant parts of the coefficients of all cosine terms are real, and those of all sine terms are purely imaginary.
are purely imaginary,
(c)
In the y2 , the constant parts of the coefficients of all cosine terms and those of all sine terms are real. In the x2j and y21 the highest powers of
ef
aV
Tt
(d)
and
e _Uv/=T
'
are 2j,
in the
and
(e)
2j+l.
The
expressions for
(?)H<f)*(<f)*
z2y+i differ
'
(C)*'
from the
coefficients of
(0''(<f )''
(f)
The
cf
+1>
only in the sign of V^7. constants of integration arising at the step 2j 1, viz., must the relations and c\ j+l (j=l, v2), satisfy
*
(<?)*(?)*(?)*
.
.
(cD
l-
c w+ m
<2J+
^*fU =
e,
(2y+i)
(i)
= _ M w +0
where
a power series in the step 2,7+3 shall be periodic. remain arbitrary at the step 2 v.
is
^
and
c2
ttJ+l)
(g)
divisors introduced at the step 2j are the determinant of the fundamental set of solutions of the x and ^/-equations and
2
The
k,
A;
-4
2
,
&2 -16V,
k*-jW
in the coefficients of
c
a ?*kt;
and at the step 2j+l they are the determinant of the fundamental set of solutions of the z- equation and k, k2 u? A;2 9co 2
,
...,k*-(2j+l)W.
The
differential equations
2
]*- [6Aeeint
[1
ZAecost
from the properties (a), (b), (c), and (d) of 106 that 2V z 2v -i taken together; that in functions of Zj, even and Y2V are 2y _ the coefficients of all terms which are of even degree in y2) y if 2v y t
It follows
.
X X
taken together are sums of cosines of integral multiples of t; that in ly the coefficients of all terms which are of odd degree in y2 y it , ylp _t taken together are sums of sines of integral multiples of t ; that the last two
, .
. .
if
in
iv
or
tv
the
then
2\+
+2j\ 2] +2fit +
i-2Kii u +v l
{2l-\)vv _ ^.2 v
l
251
It follows from these properties and (a), (g) that iv and Y2v are even functions of ew%/=T and e~ wV=I taken together; that in iv the constant parts of the coefficients of all cosine terms are real, and those of all sine terms are purely imaginary; that in F2 the last property is reversed; that
.
.
'
'
X X
in
2v
and
iv
coefficients of coefficients of
(0
(0''
(f)*(D*
(<?)*- differ
(0'"' only
the sign
of
</=T.
It follows
(74)
solutions
of
(74)
(i.
properties (a),
(g) so far
and these properties that the periodic the particular integrals) have the e., as they pertain to x2j and y2] and then
;
from
a, c,
= y(0) = 0.
for the z-Equation.
126.
The
2
differential equation
defining z2 ,+i is
+ [A+3Aecost+
(a),
. . .
-~\z 2v+i
.
=Z^
(75)
It follows
(g) of
106 that
it
Z 2y+l
is
of
odd
Vtt
'
degree in z1} z3 ,
and that
contains z2v -\
linearly.
t
In
V*
the coefficients of terms which are of even degree in taken together are cosines of integral multiples of t, and the
.
Z 2y+]
terms which are odd in the same quantities are sines of integral multiples of t. 125 that Z 2v+l It follows from these properties and (a), (g) of wV=*' and e~ wV=ri taken that the highest is an odd function of e together; "-" _uV=T wV=Tl enter linearly; and e is 2j/+1; that cf~" and c2 power of e that the constant parts of the coefficients of all cosine terms are real,
coefficients of those
.
.
'
'
(cf) only in the sign of v^T. The conditions that it shall be solution of (75). w2 Z 2v+l and +tc,Z H 1 the constant periodic with the period T are that in Wv ^ =I/ and e~ aV=It e of coefficients the of respectively shall be equal to parts the same coefficient that z3 has The z 2).-i enters Z2v+i linearly and has zero.
(cl ')'
(O'
1-
differ
(cD
4*
in
(70),
Hence, from the relations of the preceding paragraph and equations we have
[^
+^
' )
]cr +^3(0
i,
i,
cr-
+p
2 , +1
(c 1
,o=o,
j
a polynomial of odd degree in c"' and c2 taken together. It 2 v and c2 ' +l) (j = l, that is supposed 2) have been eliminated C?'" If the at the successive steps by the equations corresponding to (71). the then and relation k is P in term satisfy (0'(c2'T, ; 2, +i general
where
P 2v+
1}
is
" 1'
i-*+l.
(77)
252
PERIODIC ORBITS.
reducing (76) by means of (64), making use of (77) and (f) of 125, and the fact that cj = c[, it is seen that equations (76) are equivalent, x) v> and that cf~ = cf~ is uniquely determined as a power series in e. Since
l)
On
in
+l
the coefficient of
(cj
/
-
(cj
)''
(O*
from that
of
the solution has the same property, and from this that ziv+l (0) = 0. For small values of v there will be no other terms than those considered
(0''
Wic?)*
(cD*
v^T,
it
follows that
and -{-WyZiv+i, which are constants multiplied by the ^ e~ >v and When there are no other terms respectively. exponentials and when equations (76) are satisfied, the solution of (75) is periodic and z 2v+
above
in
t
wZ
e
+i
wV=Xt
it
has
all
125.
But
there
is
a value of
v for
which
other terms of the type in question can arise. It follows from the properties Z2v+1 contain respectively the terms wt Z2v+l and of Z2v+l that 1
+w
a
-C M (cfT +l e + iv+1 ^ v^'[a +a cost+ V l&jsintf +C \c?y +l e-'- +1 >" v=~u [a +a co8t+ + V^T6 sinJ+
<
+a
coskt+
],
(p
V ibisinkt + a* cos +
Art
(78)
].
+V~^Tb smkt+
t
Now e 9* v>ay/=i> =
e uV
=~ u
[cos2vwt
+ \r=lsm2vwt].
Consequently terms
of the type in question will arise if 2vu = k, k being an integer. Upon substituting the value of w, this relation becomes 2vn = kN, which can
when 2v becomes a multiple of N. Suppose the integer n is In this case when 2v = N, the smallest A; satisfying the relation, odd. The term in which this occurs is multiplied by viz. A; = n, is obtained. JV+,/ == if x e". But n is even and iV odd, the relation is satisfied first y*+Vt e for increasing values of v when 2v = 2N, and then A; = 2n. The term in which 2" 2 iN+ 7 = occurs is e \ e ". this relation After terms of this multiplied by X
be
satisfied
'
2 ""1"' '
'''
all
subsequent steps of
(78) are respec-
The
tively
coefficients of e
2"
"^' and
)
e~ aV
-Ciy+iic^y^ and
or e
in place of
+Cw+l (c
is
by
e"
according as iV
where C2V+X is a constant multiplied 2 even or odd. Therefore, when these terms
,
.
+l
arise
we have
equations (76)
[4
4-^ c r cr]cr
-"
l,
+A (o cr '+^ +1 (^^c;
^
l
v^
arise.
+I
(^^ +l =o.
2"
-"
253
Form
dt
of the Integral.
Equations
2
(1)
dt
d'
dt
df
(80)
After the transformations (10) and (11), these equations have the form
*--V-g.
where now
t/ is
/W-f.
x, y,
^-f,
(81)
and contains no terms lower in second and z. the It contains than terms independent of x, y, degree X and others, for the particular transformation (11), multiplied by X to the The coefficients in the series for U are power series in first degree only.
a power series in
z e
and
whose
2ir,
and which
= 0.
seen from equations (10) and (11) to be
2
The
first
terms of
U are
U = + U+2A+Aeco$t+
]x
2
+[~6;lesin*+
~\xy
+ ^[l-A-3Aecost+ +
[6(^3
<6T 3
^A+3Aecost+
x2X
J
]z
|_-(o)3
+ pn ~ 3 ^3 - r mjecost+
- T sin Z+ r )e
(
J
-(or 3
xy\
"
(82)
+ g [r - r m + 3 (j.m + g [^3 - r m + 3
(r
<,3
)ecosi+
'.
2/
Toy 3
)ecos<+
Jz
>.(+
y+
explicitly, is
+z'
(83)
so far as
occurs
This partial derivative enters through x, y, and z. e it contains as a factor. zero for e equal to zero, and therefore
254
128.
PERIODIC ORBITS.
In the periodic Integral in Case of the Periodic Solutions. solutions which have been considered, x, x\ y, and y' are expansible as
The
power
series in X, while z
and
follows from property (a) of when the expansions of all these variables are substituted in (83), the result is a power series in integral powers of X of the form
.
are expansible in odd powers of X*. It 2 106 that is a power series in z Therefore
z'
F = F \+F \*+
1
+F,\>+
=C,
(84)
where, of course, F and the F} involve the integral sign which arises from the right members of (83), and where C is the constant of integration. Since (84) converges and is satisfied for all |X| sufficiently small, it
follows that
F.-C,
the
(-l
y,
l
...
ao),
(85)
being constants.
and
'
z
'
x = \[x2 +xi \+
+x
2J
y^[y +y^+
t
-fv
z
it
~K
[z 1
+z \+
3
,
+z 2J+1
x' 2
,
+ v _1 + \ +
\'~
i
.
.
L
],
follows that
.
Fv
involves x2
3 2"-i-
zl ,z[,
Vrts
y2
-i,
y' 2 -2
*t>{%2jy%2j!
y 2 )i y 2 ii
z 2j+i> z 2i+n
~dt~(
= Cv
t
(86)
where Pv and Q y are polynomials in the indicated arguments with the coefficients in sines and cosines. The subscript j runs from
entering
to
,
v1.
and
The
z 2i-\ z 2-t)
x' derivatives enter (86) only in the form x' 2] ^ 2 2v . 2) term of or the is Q v general Suppose
y 2] - 2 y' 2v - 2l
P
'
x
2M,
>
g
2^'+l
2/1,
"2ll[
*%flf
2M"+1
+ [K+
'
'
+K] = 2v.
t
The partial derivative of Q n with respect to enters explicitly in the coefficients of x2j
, .
.
is
zij+l
be computed for t entering both explicitly and also implicitly through the xtJ . . . , z2 j+i It was shown in 125 that the x2J the y2J and the z2J+l have the form
,
, ,
x2j =
tj x^V"^',
y2J = 2 iff*****,
tj
z2J+1
t=-j-i
2 ag^e****^',
(87)
255
where the x%? the yff\ and the z^+i are power series in e whose coefficients are periodic with the period 2w. In x and z+i the coefficients of the cosine terms are real, and those of the sine terms are purely imaginary; the opposite It follows from these properties and those of F v enumerated is true in on page 254 that P, and dQ/dt can be written in the form
l)
y.
P.- 2 P'V"^
at
.
i=-v
(88)
Since in U the coefficients where P and R are periodic with the period 2t of odd powers of y are multiplied by sine series, it follows from the properties of 2 the Zj /', the y^f, and the z^; that in P the coefficients of the cosine terms are real, and those of the sine terms are purely imaginary; the opposite is
1'
true in
-~^
V=lj
yjgjg^
f-iWci
i
coajt
+
H
^}p
a ?-4k?u*
rinjf,
(89)
/'
Therefore
sin jtdt
tto
cos.?*
.^
smj*.
we have 2
S?V*=*
(90)
where the
are periodic with the period 2r. Moreover, the coefficients of the cosine terms are real, and those of the sine terms are purely imaginary. It follows from these properties that (85) can be written in the form
(*)/,2twv^=T/
"'
c,
(91)
are periodic with the period 27r. The coefficients of the cosine If we let real, and those of the sine terms are purely imaginary.
4-*=**m 9t
and make use get from (91)
of the fact that the
(92)
2ir,
ff
we
2
t__l
tt
F,
VtuVrrT
'<r t
= CV
2
tm-r
Ff^^'al = C
{k=-v,
*=-
2 F?e"^ ^fff-C,.
(93)
if
either
FT-C,
or
FT-0
1
,
.
-i,+i,..
+ v),
(94)
256
PERIODIC ORBITS.
This determinant is the well-known product of the differences of the a+, taken in all possible pairs, and is distinct from zero unless a relation of the form <r = <rj is satisfied. Since <r, = <r\, at = a[, this relation can be satisfied J only if ffJ~ =l; or, because of (92), only if
t
Since u = N/n, this equation can be fulfilled only if {ij)/n is an integer. But then two or more of the exponentials of (91) are equal in value, and the number of terms under the summation sign is reduced by combining
similar ones.
understanding as to the reduction of (91), A can not vanish and equations (94) must be fulfilled. It is clear that when these equations are satisfied all relations of the form of (93) are satisfied. The F"' are explicit power series in e, and they also involve e implicitly in w, which is a power series in this same parameter. Now o> enters in two as a factor of certain terms either to the first or ways. It is introduced second degree by the derivatives which occur in (83), and to the first degree
With
this
by the
integral, as
is
shown by
2
(89).
The
4kW. We denominators in the form j in the first way. This will not change the character in e wherever it enters But where co enters in the second way we shall regard of the convergence. Then equae as an independent parameter and leave it implicitly in w. tions (94) can be written in the form
F?'
2Fe = C=
J
CV,
.-i.+i,..., +).
e,
F? = 2 FZ e'=0
{k=-v
we have
(95)
F?,=CV
The F") are sines and
F%~0.
t.
On making use
F=
S
J>-0
[a,cosp+\^&,sinp*] = C,.,,
[a?J ,
F = i
cos pt
+ V^T 6?)
t,
sin pt]
= 0.
we have
finally
a ^.o = C./>
ai),
= by] =
iP
(p=l,
,j),
(96)
257
when e=0.
Equations
(96) are relations among the coefficients of the solutions and may be used for checking the computations. The control is very effective because at each
step
the preceding coefficients are in general involved. But equations (96) can also be used, step by step, for the determination of the coefficients of the expansion for z when the coefficients of the expansions for x and y
all
are determined, alternately with those for z, from the first two equations of Before taking up the general problem we shall treat the case of e = 0. (13).
When
x'
2
this condition
2
is
becomes
2
2 +y' +z'
=
[x
+y +A (2x -y -z +3Bz
2 2
2
-]+C,
r>
~M
'1
**<>
's
In this case w
= Va and
x 2 ,= 2
o^e"^"',
2
y21
= 2 ^fe *^^1
2
',
zS)+1
( a Y,Sa e
+OVIV=i
',(98)
where the
<4 /',
/?*',
and y
+l)
are
constants.
We
shall
show how
(97)
to
compute the y%+? for successive values of j. Terms for j = 0. In this case, since x =y = 0, equation
as a consequence of the last of (98),
becomes,
1)
7<
Since C, is an identity in t, we get 4^7, yf = C _1) and yf. But since unknown, this equation imposes no relation on 7j -1> = z (0) = 0, it follows that t{ yf, and there remains the single unde-
-l>
is
It follows
from
(13)
and
(14) that x 2
and
'+e*^-><-2],
(99)
The
_ r2 ~
.^A
2TT=7;L
J
-
+ 18A
(100)
y*-
+ -(r-7A+isA
[e
)
258
PERIODIC ORBITS.
Terms in
z for
,
j=l.
It follows
1
from
}
are
+ (l+2A)x\+(l-A)yl
(101)
= 7f
3)
e~wXv=s, +yi~^
rise to
F^^+iQ e^ ^ '+7
v
^^',
(102)
the relations
(103)
another equation which is useless for present purposes because Since z3 (0) = 0, we have also involves the unknown constant C2
There
is
it
y^+yl^+yr+y?^.
It follows
(104)
Y;3)
= -7,0)
l>
,
Yi-
=-7
a)
3
.
(105)
In order that yl~ shall be the same as determined by both equations of (103), we must impose the condition
27(1
-ZA + UA*)B%yl
l)
2(1
+ 24) (1-7A+18A)
=
9
2
t0y >
+ (1 Krf
,
\
(
ifV
l ~ s)
(Yl
_ Wy } _U
'
lom (1Ub;
and yf are uniquely which determines y except as to sign. Then y 3 determined in terms of 7"', but 7"' remains so far arbitrary.
The problem
for e
was treated
in
Chapter VI.
If
equations (100) of the present work with equations (42) of page 211, find that 7"' and Cj are related by the equation
c\
we compare we
= 4A(y?)
(1)\4
Upon making
use of this relation, it is seen that equations (106) of this chapter and (44) of Chapter VI are identical.
It follows
from
(13)
and
(14) that
we have
x!-2y'<-[l+2A]x = SBz
i
zi
+P
, i
,
2/ ;
+2^-[l-^]t/ = Q
4
(107)
where
and
2ir/V~A.
We
are entirely known periodic functions of t having the period wish the details of the solutions of these equations only so
A
259
l)
Tj
I)=
fl~
So
far
is
as these terms are involved, the right member of the first equation 3By?y?[e-"iv=I <P* v=Tf]. Hence the solutions of (107) are
Xi
1-7A + 18A
[e
+C
6
J+
T+2A
'
+^
'
(108)
!/where
+
P
K
i-7^ + 18A
and
[e
i+Qt
t.
are
known
periodic functions of
Terms in
2z
1
z forj
zi
= 2.
2
It follows
from
zl+2Az
=-2x
+6Rt
where form
^ +2 (^m - ^3)21*3 + 3 C
3
(109)
z?z 3 +fl 6 ,
is
known
periodic function of
t.
The
(110)
On
substituting this expression in (109) and equating the coefficients of the several powers of e^ >/=r', beginning with e~ 6 *a N/=T', we get
8 4 t!
of
,,
3,
5,
l,
8^7[7r = [
3)
"1
(HI)
There is also an equation, coming from the terms independent of e vlvZT *', which involves the unknown C, and need not be written. The first equation b y. In order that the second and uniquely defines yl~ \ which equals third equations shall be consistent, we must impose the relation
[
Therefore 7
positive;
T^i-fisA
is
+ ^o] (7i
1,
7f = known
function of T
(112)
of (111) defines 7s
which
equation is is the
negative of
y.
condition that z5 (0) =0,
we
7"',
and
All succeeding steps are precisely similar to the one which has just The parts of the equations which contain undetermined been explained.
coefficients differ
in the subscripts.
260
1
PERIODIC ORBITS.
30.
It will now be shown General Equations for z^ x and y are determined from equa(14), the coefficients of the expansion for z can be determined We shall need the partial derivative of U with respect from the integral (83) We find from (82) that to t so far as this variable occurs explicitly.
(j^=-[3Aesmt+
+ |"|AesinJ+
lz
-lz
2
+[6Aecos+
Jxy
+ ^AesinH2
]y
Jx
\
(113)
2
-[6(pi>j--<wi)ecos+
|J (^ojs-pajesinf-f-
-jy
-[^(r
Since x
t
m-^ )esmt+
3
.]^X-[f(^+ r^)esin<+
(82), (83),
]*+
integral
is
= y = 0, we
2
find
from
2;
=-[A+3Aecos*+
-]z
-/[3Aesin*+
2
^zldt+C,.
(114)
In the notation of equations (87), the expression for z has the form
Z =jz
(-1)
o-uV^ltjigto gwv'-i*
where
t-0
*-0
m-
s Wr^cosp<+-/=T/3i:^sinp<],
p-0
(115)
& - S [Ccospi+V^TiCsinpt].
p=0
On
it is
found that
+ [-u
+ [2a;
2
<
- 1)
1 (z[-y+(z'
y- 2a>V=iz[+2o V^l2l
J
1)
l)
- l) ]
z[
e-
,-2uv=it
+ [-* (O
^8
+(2i'
H) 2 ;
]+"=*
") ] =
+2z;1,
+2z[
'
~ l)
z[
a)
(116)
1)
-[A+3AcosH([ZAeaint
][(zii)
ee-
2 'oVirT
i)
'+(z;
2wV=T
i)
2Uv '~ 1
z[
1)
]
1>
][(zr
'+(z;
2a,v
^ +2z{,
zi
]^-|-C
261
Before the integration the series for z[~ and zf must be substituted from (115). Consider the coefficient of e~^ s/=lt under the integral sign.
It is
sum
of cosines
and
coefficients of cos jt
Suppose
iAj
It follows
from
we have
sin jt,
terms
j^
are
cosjt
Vf=$
(H7)
'
respectively.
The corresponding formulas for the coefficient of e2Wv/;rr The terms obtained from these simply by changing the sign of
it
independent of the exponentials which involve the cosine and sine are divided by -\-j and j respectively. Consequently, in all cases
easy to write down the explicit equations for the identities. In the notation of (91) the coefficients of er*v=* J e +i " v=Tl ,
(116) are respectively fj
of jt
is
and
in
-0
,
F <- =F = 0)
Since these functions are power series in notation of (95)
e,
F?L<!lt
we have
in
V-*5-A
And
these functions in turn have the form
*Z-C*>
(us)
2 [a^cos^+v^Tfc;Ft? = J>=0
J>=0
J
1'
sinpZ]=0,
F% - s [CfCosP'+v^TCsinpflsO,
2 [a h ,coapt+>r=Tb,,f&np(\ = Ch ,. FZ - J>=0
Since these equations are
(0)
all identities in
t,
we have
ia)
262
PERIODIC ORBITS.
whence
l)
(z[-
= (z^y+2z^z^e
+[2aig> +()]*+
z\-z'r'= tf-y =
+zWr
l)
0+0
2
1 '.
(O
=(0 +2**S
=
(Cy
=
_
(-1) _(1)
+iS^f< +0
,
+?)*> +
,
,
/-(-
-<1>
_(U ,(-1) \ p
(120)
l)
w
z[
z[-
=0+0
o
+c;>
+*gi&-*
+2co
1
2[ 2 1
l,
,<
- 1)
o
co*
=
Equating
+[ +[ +[
$?*$+*&*&]*+
JW^+fSifi*]*+ w +2a> w,]e +
s
!!
we
get
(4-a>
)(zl-
(_l)
(z.'
- 2o>V=Tz{-
l)
'- i)
z' 1
-(z[-y[SAecoat+
(^- w )(0
!
T[3^esini+
i,
JGO**
]
+(2i
a, )
+2coV"^l2;
i,
21
' a>
=-(z;
[3^cos<+
- i)
2(A-w')zi-z? +2z[
-22l,)
)
(121)
f[3^6sin<+
l)
](z?ydt
- l)
z[
-2o,V^[zl-
m
z[
)
-z?
z[
)^
]
,>
l)
2j
[3^ecos<+
]-2Jz[-
z?
[3Aesmt+
+C
lt
where the
under the integral sign / must be transformed by instead of forming the ordinary integrals, and where Ct is an equations (117) undetermined constant. These equations are power series in e, and setting
coefficients
we have equations
(119).
131. Coefficients of e.
On
referring to (120),
2afc::
we
,)
-^(z^r= -A(z^y,
-*(&- -aw,
C= -t^c+e,.,.
+ z = 0.
,
Since z^O) =0, we must add to these equations z[~" from these equations that
<*\
It follows
= A,
z^=-zZ,
since
4A(zY = CU0
CM
is
(122)
and
2["
an unknown constant.
263
(115),
On
and
we
get
at this step
(
-2u>lz
[a^.+o^cosH- V^T/^sin*]
-2co
^ C^
')
-2a>
l
= -%(/-?)* Aco&t
V^T/^cosZ]
1
-2zi!U[a;: .+ai
1
u
1. 1
cos<+^^^^,
sin<] +
^^[cosi-2co ^^Tsini],
(123)
+4 WoWl 2<:
zi:
+*R [
&+&
2g{*?
cos<+
-jft [
[a<:i.o+a,
i.i
sin*
+ V^TjfiB., cos*]
sin*+ V=T-fl?,
[oiJi.,+0.! cos*+
=
cos*]}
-ta^agU cos*-2a{z{:
V^TigL, sin*]
>,
A eM#+CM
we
find, after
making use
of equa-
-2AzXU^+2VAz^fc=-3A(zXy-2Az^a<-+ ^jjfffl
-2Az[^^+2VAz^a^=-2Az[^^+ **^p'
-2A3a. - 2^1^(0*- -24z<C.,
+2A^^+2ii2W^+2^U ^-2v^^T.V*+A(^'
-2Az^aZ,-2Az^a^-6A(z^y,
=
-2Az^tfl-2AzX.
264
PERIODIC ORBITS.
the
first
From
i
two
we
get
z i,o-
= U,
a i.i,i
= ~a
i,i,i
=
i
_4
>
Pi.1,1
=+Pi,i.i
=-
j^^
- ^Vila
1 , 1 1 .(v
,
liS4J
The first of
the last three equations imposes no condition upon the unknown coefficients since it involves the undetermined C u and the second and
,
become
identities.
The
coefficients
ajjjjj
and
undetermined, are not involved in these equations. The o>, is zero was known in advance, for it was proved in 120 that w is a series in even powers of e. It has also been shown that 1. 2," and zj", aside from constant factors, differ only in the sign of V Since z(0) = these constant factors differ only in sign, from which it follows that a{~^ differs from a{", only in sign, while (t 9 and #" are equal for all Applying the condition z(0) = to the terms under consideration at j and p. = 0. On making use of (124), this equation present, we have z^ (0) + z," (0)
still
which are
fact that
l)
l>
<--C>
and
o"J
(125)
Of course, it should be alone remains undetermined at this step. noted that a",,, and 0"}., are expressed linearly in terms of the undetermined
when we
133. Coefficients of e
and
e*.
From
we
can infer the character of the process in general. Because of the relations between aj^,", j8^ and a{", #" it is sufficient to equate to zero the
^T coefficient of e2&,v/
'
in (116).
Since
unknown
we
out the equations explicitly only so far as they involve these unknowns. Upon equating to zero the terms independent of t and the coefficients of 2 cost, eos2t, sint, and sin2Z in order in the coefficient of e we find
,
-2VJ(z!!i)
co 2
= r(z;)
2
,
-iVItfttfL-gf^,
-2VJO&-/&,
where / is known and where g?^ the second degree in z"i = a['o, and
(I>
;
,
(126)
J
/{j i2 are
homogeneous functions
in aj"
of
a"J,
alone.
, .
The first
. . ,
a"2il equation uniquely determines co2 the remainder determine 0^ uniquely when z" and a"| become known, as they do when the coefficient
of X
2
is
considered.
The
left
coefficient
a{|20
is
The
place of
P^'i.t,
The
first
has w t in
. . . ,
and the
. .
members
a"i,,
2p -s/ZzJ'J
of a| o,
fl
a"^, the numerical coefficient of /3['i and a"i being The right members are homogeneous second-degree functions
,
o"-i.o-
The
aj'i
remains arbitrary.
265
1.
from
F = 0.
2
From
(82), (83),
and
2z[z' 3
+2[A+3Aecost+
+2[6^4.esinf+
~]z
z3
.]*j
-~\y\
+[
is-^r3
3(^i-^)
)ecos<+
] $+[f(fp50 +jjb)
2
. .
+f
(j^r
+ $m)ecoat +
-~\x2
]zl+2j*[3^esin<+
yt dtz3
j
.]x 2
dt
-2 \[6Aecost+
-2J [34esinJ+
r
[34esin+
^\y\dt
]z
dt+
[]^(^-^)e&mt+
.]*}*+.
.+(?,.
\\dt
3J
[(^ + ^)esin<+
.
The 2 and i/2 are determined from equations (55), and it is seen from these equations that they are homogeneous of the second degree in the Other properties of the solutions are given in 121, coefficients of z x
among which
The
that they are of even degree in ewV=il and e~ uVZZIt expression for z3 has the form
is
23
= 2^
,)
e-
w=
I,
+2
,)
"^=I<
.
(128)
The
coefficients of z 3
~ 3)
and
zl~
differ
from those
of
zf and zf respectively,
l,
Hence
it is sufficient
tf
= S
e\
zf
= 2
t=0
ig e\
ig = 2 =
J>
[a 3
l,cospt+V=i ff^smpt].
135.
Terms Independent
of
e.
and e**^ set equal to substituted in (127) and the coefficients of zero. These terms are power series in e whose coefficients separately must
be set equal to zero. We are now interested in the terms which are independent of e. These results were worked out in 129, where the parts of x2 and y2 independent of e were derived. The explicit results were given in equations (103), the relations in the present notations being
_(
tt
l,0,0
f*^
_,< Tl
>
a3,0.0
>
73
,<
a3,0.0
<
3>
/3
/>
266
PERIODIC ORBITS.
The
C And
condition for the consistency of the two expressions for y," is equation = except as to sign. Therefore, by 132, (106), which determines 7"' and 0gj are also determined except as to sign, and is defined.
Co
by
133
it is
as to sign, while the C, remain as yet undetermined. The equations correto determine but remains so far arbitrary. (103) 73" 7"' uniquely, sponding
136. Coefficients of
e.
Co
#"
(p^O)
are
all
determined except
We
C,
shall write explicitly only the terms which SSi tT ft f . and flfij which, at the successive
unknown. The quantity z3 is involved in the right member of (127) under the integral sign, but since this term is multiplied by e, it as an unknown coefficient, and this undeintroduces at this step only termined constant enters linearly. It is determined from the terms which
-0
are independent of e when v = 2. Consider first the parts of the coefficients of etuV=It are independent of sint
and
e iWy/=lt
which
and
cost.
We find
4^4m^--^&2
where
/jfr"
+/&?>
Since
-24eCC =2Aa{
w
o,
is
and
f^
known
(130) shall be consistent is detailed discussion shows uniquely determines the coefficient equations (130) uniquely define Co-
bjJ =
C Co
M +/{,,
(130)'
which
Then
Now we
e*" v=T 'cos<,
^u"/=rit
s\nt, e*"*^' s \ n t f
and
'cost.
The
V=l VI [+
a*, + 4 VJ
V=T VA [
- 2 VJ aSa
- C. a, - * -* {&
, J
] 3
l
(131)
,
where
<p t
<p<
are functions of
linearly.
the equations are derived for v = 2. The unknowns in the left members of and 0j&A , which enter linearly. On making use of (131) are a"},, $'},, = the fact that C. the determinant of their coefficients becomes
Co
which enters
known quantities and the arbitrary This constant remains undetermined until
d,
C"
A=-A(C)
[4A-1],
(132)
which is distinct from zero. Therefore these quantities are uniquely determined as linear functions of the arbitrary C, This illustrates sufficiently the method of determining the coefficients from the integral. The complexity of the details makes it unprofitable to
.
267
Terms
in X v\
It
was shown
in
as
from equations
lfc-0,
-Jyi-O.
The
known from
the general
1,
p,
pt +a'? e- ui ,
)
=a +r,
=vl
m
e'+
tt.-tf-Hipe
vi
+v
( l)
+ e +
(i=i,
...
,4),
where a{,
a 4 are arbitrary constants, and where the u, and the v are periodic functions of t with the period 2ir. In order that the solutions shall be periodic we must first impose the
l)
. .
conditions
a,
1)
= a = 0.
1)
(135)
The constant
since u[
is
j)
a continuous function of n, n , and e. It will be supposed that these parameters have such values that a- is a rational number. Then, a
is
are periodic with the period 2t, the solution at this step periodic with the period T, where T is a multiple of 2x and 2x/<r. In the symmetrical solutions, x' (0) = y (0) = 0. Since these relations
v,
X*,
and
J)
are identities in
1
we have
x[ (0)
=y
(0)
= 0.
t
1)
from
268
PERIODIC ORBITS.
restricting the
Without
we may suppose
that
=v Vy (0)
= v, (0) = v (0)
=1 (0)
=
Therefore
we have
a}"
= =
l>
Oj
*i
+<
aV~'
[e
- e-"^
1'
ut
~]
+a
[>'
u3 - e' pl wj
(136)
We
shall
initial
1 as to give the symmetrical orbits. Then, since changing the sign of in (133) does not alter these equations, with the same initial conditions
Therefore
we have
4
m
vl
u3 (-V=i)=u3 (V=T),
v3
(-v^T)=w K^T),
4
(-V=T)=v
(V=T),
(-V=D -(V^D.
If
we change
the sign of both t and yx , equations (133) are unaltered. initial conditions as before, which this transformation does
affect, since
(0)
0,
we have an
is
changed
in sign.
Therefore
t*i(-0
,(-<)
- -u
t (t),
w,(-0 =
vs
~u.it),
i {t).
-+*C0,
ft
it
{-t) = +v
Now
if
l,
t,
and
i(-v^=T1 -0--i(v^:T,i)j
^(-v^=T,-<) = +w (v =T,0,
A
2
>i(-V=T,-0 =
r
-t> 1 (v'=T,0,
4
w3 (-v/^T,-<) =
4
w4 (a/^T,0,
3
^(-*V
=T,-4)+# (V^T,0,
= +y3(vr=^T,0w
4 , . .
w (-v^T,-0 = -w
It
v,
,
.
(v
^T,0;
:rT,-<) 4 (->/
vt
when expressed
as Fourier series,
u = '2[+ajCOSJt-\-\^^ibjsinjt],
l
ut = 'S[+CjC08Jt+d
sin jt],
w,
=S =S Vi =2 u,
V^T 6 sin,/*], + V^T ^ cos;'* + & sin #] + v^T a, cosjt A sin ji]
djcos jt+
;
,
,
(137)
,
vt
5;
sin jt]
where the
series in
e.
a,,
b,,
and
5,
and power
269
-e-#y=BttJ
y^a
\f
v=li v
(138)
In the case of the periodic orbits it follows from equations (137) that the numerical coefficients of the cosine terms in x are real, and that those of the sine terms are purely imaginary; and the opposite is true in y
x
138. Coefficients of X.
coefficients are defined
It
is
(13)
that these
by
-fo -$Aemat+
-]y2 -\y2
y2
+2x -[ GAesint+
2
-]x2
-[l-A- SAecost+
]x]+[-24:Besmt+
=X =Y
(139)
,
[-3B-12Becost+
-]zi*/i
+ [f + 6ecos+
Y = [-12Besint+
t
2
-]yl,
]x +[3fl+125ecos+
(140)
-]yl,
+ [9flesin*+
1 Mo + Mo r _ + zm' + r (ox P s.
character of the solutions of the equations of the type to which (139) belongs was determined in 30, where it was shown that they consist of the complementary function plus terms of the same character as 2 and
The
F,
Hence the
is
x2 = a{2)
*=
e,
+a a> r"*= u +f
s
2
yt = a?
2)
(141)
>
aV=li
P,
+af
-**=* v 2
+g
where oJ and aj" are constants which are as yet undetermined, and where /2 and gt are the particular solutions. We shall need certain properties of /2 and g2 It is evident from (139) and (140) that they are homogeneous of the second degree in a{ and in *v=n an(j e-#v=* it follows from (137) and (138) that in x\ and y\ the e 2aV=il and e~ 2 " v=Jt coefficients of those cosine terms which are multiplied by e are real and identical, while the coefficients of the sine terms are purely imaginary and differ only in sign. The terms which are independent of In the e <rv=rt an(j jr*v=t con si s t only of cosines whose coefficients are real. which are those cosine terms of the coefficients multiplied by product Xj y 2<rv=ii e are purely imaginary and differ from the coefficients of those cosine terms which are multiplied by e~ aay/=lt only in sign, while the coefficients of
.
270
PERIODIC ORBITS.
ancj
e -<rv/=r< are
The terms which are independent of on iy g j ne terms and are real. Therefore it follows and Yt have the form
!)
o>
the coefficients of the cosine terms are real and identical, and the coefficients of the sine terms are purely imaginary and differ only
In
and
X ~v
l
in sign.
In Fj
2)
In XJ0) there are only cosine terms and their coefficients are real. and F,~ 2) the coefficients of the cosine terms are purely imaginary
only in sign, and the coefficients of the sine terms are real and In Fj0) there are only sine terms and the coefficients are real.
and
differ
identical.
from the properties of X2 and F2 which have just been derived, and from the form of equations (139), that/2 and gt have the form
It follows
g,
=
,
<tf'
#"=*
(
+)V-;
2)
r"*=
(142)
and g are periodic with the period 2r. from the properties of X2 and Yt which have been found that l is changed. equations (139) are not changed if in them the sign of V Therefore this property is true of the particular solutions, and we have
/,~
where ff,
It follows
2)
/, g,
~ 2)
gt
2,
/i
0,
/<
(- V^T) -/T(V=D,
It also follows
sign of
from the properties of X, and F2 that if we change the and y2 equations (139) are not altered. Therefore
,
/.(-<) - +f?\t),
g?(-t)
i)
-g?(t).
It
/i"
(- v^T,
-0- +/i-
2)
(v^T,
0,
271
when
f,
gf =2[+v^a,cos.#+ftsin#],
gf = 2 [- v^T ajcosjt+^sinjt],
2)
(143)
gf =2y>smjt,
a,
b Jt c Jt a t
ft
and
and power
series in
e.
since yt {0)=0,
we have
a=
= 0.
Therefore,
(144)
where both
a"'
and
2)
a}
139. Coefficients of X Vl .
It is
(13) that
by
x3 - [64esin*+ y3 =
y3 =
x"-2y' 3
y
[l+2A+6Aecost+
]
]
X
Y
3 yi +2x' [64esin*
3
x3
[l-A-SAe
[
cost+
sint+
X =+[-2K-6Kecost+
+ [-65-245ecos<+ + [3fl+12fle cos* +
Y =+[-6Kesmt+
3
-]x,+
]z
1
-6Ke
-]y,
x2
+[-24flesinH
^{x^+x^y,)
-Jrtfjjl
(
]2/ 1 2/2
+[4C+
].r?+[-6C+
-]y,
]x
~]x l
+ [K+3Ke cost+
+ [-24Besmt+
+ [l8flesin*+-tV
jyr
(0>8
(0)3
'
x2
fl ^
1
'1
Mo
(0)6
T
I
Mo
(0)5
'
'l
'1
's
from the results of 29 that in general the terms of the first vz:T will introduce non-periodic terms into the solution. degree in e"*^ and e~" We must determine a" if possible, so as to make their coefficient vanish.
It follows
1'
'
1
272
PERIODIC ORBITS.
The
are zero,
first
two equations
of (145),
when X, and F
\+a
"j
pt
e
p
Ui+a
'v3
e~ 'ut
f
(146)
t
,
vx
+a? e
v2
+a?e
+a?e- "v
where a '\
variables
and
are arbitrary constants. Now, on supposing they are subjecting them to the conditions that equations (142) shall
.
af
be
satisfied
when
their right
a
members
are included,
p
we
p
get
e'^' u (a?y+e1
^\(a?y+e 'u
xv=TI
(a?y+ei
'ui (a?y= 0,
(dry.
)'
[W=r
**=*
1
,+;] (a?y+e"
3
2 [-<rv=Tu +u' ]
=X =
3 ,
w
(a 1
)'+-
rV=I
'
(147)
0,
v,{ay+<rv,{ay+e-'
v l (fi?y
(,*)'
3
,
+e" [pk +
where
3)
'
(af )'+e-
- p + ()' - F tQ
4
.
The
(af)' are the derivatives of af, 3) solutions of these equations for (a{ )',
(aj )'>
.
3>
,
(a 4
)'
are
A(a?)' A(ar)'
= =
[A.
1
AW)' = [/> X
+Z> ,
3
F]
3
"*
,
[A X +DM
Fje*^
ACO'^Z^X.+A, F ]e
3
+P
',
(148)
where
if.
A=
Mj
w,
*>-w
Uj
u,
*>3
A,= +
^2
oV^T W, + Wj,
pMj
+ Ug,
-pW
tt 4
273
from Da and D respectively, by changing the suband by changing the sign of V^T and of the whole expression ZX. and D3i are obtained from D n and Da respectively, by changing the subscript 3 to 1 p to a V 1, and by changing the sign of the whole expression; and D41 and D 42 are obtained from D u and D u respectively, by changing the subscript 4 to 1, p to <rV l, and by changing the sign of the whole It follows from the discussion of 18 that A is a constant, and expression. in this case it is a power series in e.
Dn
and
Dn are obtained
1
12
script 2 to
In order that the solution shall be periodic it is necessary that the We shall show these right members of (148) shall contain no constant terms.
conditions are sufficient.
right
When
member
they are satisfied the general term of the two equations has the form
~
[alk
stants.
where j and k are integers distinct from zero and where aJik and Consequently aj and a are sums of terms of the type
3)
a,j, k
(2ka
V-
cosjt+jsinjt)
bj, k (
-jcosjt+2 ka \r^\
sinjTJ
,_ 4
(149)
The
right
member
of
t,
[aJtt
cosjt+bJjk sh\
-p]<
jf\
of
,, t
is
sum
be no exception to
this form.
There-
W-|L(2*+l
I"
fp
"I
Wk+l)<rV=Tt
-pJ eoBJt+jmajt
-,
^^
-pt
-pi
"1
|-
(2*+l)<rv
=T(
(150)
for a," only in the sign of p. There It follows, from is an additive constant of integration with each of the af\ if of and that we the constants of integraform the (150), (146), (149), put
for
differ
a"
from those
and af equal
}
T.
to zero, the resulting expressions for They may be written in the form
)
It
y*
= a?
e''^'v
+a?
e-<
^%+g
(151)
where af and o
274
It
PERIODIC ORBITS.
remains to show that
a["
of the first
two equations
of (148) shall
'
can be so determined that the right members have no constant terms. Let us
consider the first of these equations. are to set equal to zero the conYs It follows, from the stant part of the coefficient of e"^1 in n 3 l2 Y form of 3 and 3 equations (145), that the term which must be made to
.
We D X +D
vanish does not depend on af. It also follows that the conditional equation which must be imposed has the form
W-Q.W-O,
x
x
(152)
where P, and Q^ are power series in e, the former coming from those terms of X, and Y3 which are linear in x and y and independent of x2 and y2 and the latter coming from those terms which are of the third degree in or which involve x2 and y2 Xj and y The solutions of (152) are 0^ = 0, which leads us to the trivial result x = y = 0, and
,
l ,
(153).
The
116-118 in connection significance of the double sign was discussed in with the existence of the solutions. The expressions for P and Q are power series in e and both of them contain terms independent of e, as was
l 1
shown
e = 0.
in
Chapter VI
l)
Therefore
a[
is
in the discussion of the corresponding problem for a power series in e having an absolute term.
remains to be shown that this value of a{" also satisfies the equation which is obtained when the constant term of the right member of the second
It
equation of (148) is set equal to zero. We shall show that the constant aV^Tl in n i -\-D n Y3 is identical with the conpart of the coefficient of e stant part of the coefficient of e~ a,yzr:i in D2l 3 Let us first consider 22 F3 the term [2K6Kecosl-{]x oi 3 which contributes to P, of equaSo far as this term is concerned, we have tion (152).
D X
1
X +D
A
A(a2
( 3)
}xl
-]x1
,
e-'rV
=' u
,
21
' ^ e + v ri
(154)
.
On
referring to (138) and the values of parts of the right members of these
Dn
ut
and D2l we see that the constant two equations are respectively the
constant parts of
w,
-a
[-2K-QKecost-\
],,
oV=\v
-a
(i)
+v' 2
pv.+v' z
u,
-pv
W
+v[
[-2K-6Kecost-]
]u2
+(^f=lv +v[,
i
pt>,+t>3,
-pv
+v' t
275
in
f^U,
U3
v3
11,
+ Ui
+v
i
v2
-i\
v3
(t),
W^T
_aj_
vMM>i, pfa+pJ--H*-i
,
p{v3
-v<)+v' 3 +v[
u^
u2 v
t
v3
-v
v3
+w +v
Fi(t),
+<rV=Tu
where
vl
+u
2 v' 1
p(v 3 +v A )+v' 3
-v' 4
p(v3
-v )+v' t 3 +v'
t
F(t)
= [-2K-6Kecost+
y
parts of these expressions containing u u2 as a factor are identical and need no further consideration. The parts multiplied by x^v2 and u 2 v so far as they appear in the second lines of the determinants, are respectively
t ,
The
,d)
u3 -u
u3 +ut
P {v3 -v i )-]rv 3 +v' i
-v' t
[-2K-6Kecost-\
],
+ --u v
2
u3 -u
u3 +u
p(w,
p(vt +v 4 )+v' a
-v' t
-)+!+!
[-2K-6KccoaW
].
from (137) that u3 ut is a sum of cosine terms, that u3 -\-ut is a sum of sine terms, that v3 +i\ is a sum of cosine terms, that v' v\ is a sum 3 v is a sum of sine terms, and that v' of cosine terms, that v3 3 -\-v[ is a sum the of sine terms. Therefore determinant parts of these two expressions are sums of sine terms, which, multiplied by a cosine series on the right,
It follows
i
Hence, to get the constant parts of these expressine terms of the products u v2 and ut vl It is seen at once from (137) that the sine terms of these products are identical, but that the cosine terms differ in sign. Therefore the constant terms coming from the parts of the two expressions which are multiplied by u v2 and u2 vlf so far as they come from the second lines of the determinants, are identical. The parts of the expressions which contain u^v2 and u^, so far as they come from the third lines of the determinants, are respectively
t .
a v + -fc*y/~=lu
l
-\
<s
s/~=\u2 v
276
they are identical
PERIODIC ORBITS.
Therefore the constant parts of the two expressions, so
far as they arise in this manner, are identical. It remains to consider only the constant parts of the
two functions
[-2K-6Kecost+
in
W,
V,
W4
W3
^3
*>4
,
+ W4 + [-2K-Keco&t+
^4
.
we need only the cosine terms of u y v2 and v^v[ that the coefficients of the cosine terms in these products (137) Therefore the constant parts of the right members of equaare identical. tions (154) are identical.
It follows, as before, that
We see from
The
is
discussion for the other terms of X, and Y3 which are linear in x in a similar manner, and it is thus proved that the P, which is l obtained from the second equation of (148) is identical with the one which
l
and y
made
depends on the
It is
first.
now
linear in xt
and y
necessary to consider those terms of X, and Let us treat in detail the term in
x
.
XjX2 as a factor.
of (148)
So
far as this
term
is
which are not which contains concerned, the first two equations
3
Y
3
become
A(af)' = Ai[-6P-24Becos<+
A(af)'=Ai[-6P-24PecosJ+
x x2 +oV=it
x
f!l
(155)
On
referring to equations (138), (144), and the expressions for n and sl ,it is seen that the constant parts of the right members of these equations are respectively identical with the constant parts of
u,
-Crt V (a?)*
a)
[-6B-24Becost+
][i/, -w,/,
a9
e)
],
aV^\v +v2
2
pv3 +v' 3)
-p+4
uA
a
-(a?)
v,
[-6P-24fiecos*+
][,/,">
-uJJ
*]
+ <rV^Tv +v[,
l
pV3 +v' 3
-pVt+v'i
Since/
is
is
a cosine
series,
of
it
and
6P 24Pecos+
0)
x
also a cosine series, the discussion for the terms multiplied by w, /j and 0) Uj /J does not differ from that given above for the terms multiplied by x
.
277
*
2
27S
PERIODIC ORBITS.
Let us consider, for example, the term of 3 which is multiplied by x y\. Then, so far as this term alone is concerned, the first two equations of
x
(148)
become
-U^e-^A
{
J JfcfB ?"=.-}
members
u
u3
+W)
-ff-s/^Tvj+tfj, pv3 +v' a
,
[-6C+--'][2u
v l v,+u t v[],
-pv
-\-v' t
(158)
+W)
+ ffV^Tl) +
1
[-6C+---][2iw,+u
1 ',
ifl.
Pt>j
t>s,
-P^ +
*>4
a cosine series having real coefficients, the discussion for the the terms multiplied by 2m ?;,i>2 and 2u2 v v2 does not differ from that given above for that part of X, which is multiplied simply by x
Since
t^f, is
1
l
refer to equations (137) and (138), we see that v\ and v\ have the % properties of /" and ff \ as regards the relations existing between their Therefore the discussion of these terms of (158) is respective coefficients.
If
we
identical with that of (156), for which the proposition was established. In a manner similar to this the identity of the constant parts of the right members of the first two equations of (148) can be established for all
of the elements of
140.
which
and
are composed.
Members
of
the First two Equations of (148) are Identical. We shall treat first the We shall need the following properties of s 3 parts which depend on
(1) (2)
(3)
a polynomial in x1} y xt y2 Those terms which are of even degree in y and y2 taken together are multiplied by cosine series having real coefficients. Those terms which are of odd degree in y and y2 taken together are
It
is
x ,
,
.
(4)
having is an odd y2 then ^1+2^ the one or case present three). integer (in The parts of the first two equations of (148) which depend on A'3 are
is x[ x'i y\>
l -
sine series
real coefficients.
,
+ ^+2^
X
A(a?)'
It is
= Dn
e-
A(ary = Dn
e'^'.
(159)
obvious from (137) and properties (2) and (3) that those parts of ^JI and e"^^' which are independent of the exponentials e~ aVZrJt 3 3 and e" s ~' are sums of cosines having real coefficients and of sines having purely imaginary coefficients, and that the real coefficients in the two exe~ av
279
pressions differ respectively only in sign, while the imaginary coefficients are respectively identical. Hence, referring to the expressions for n and n we may write these parts of equations (159) in the form
ilj
u -ux
3
u3 +ut
v3 +v,
v3
-v
2 [A
cos jt
+ V^i B
sin jt],
Wj
u3 -u4
v3
-v
v3
+u +v
2[A j cos jt
t
Bj sin jt].
from these expressions, as in the discussion in 139, that their constant parts are real and identical. Now consider the terms depending on Y3 which has the properties (1) and (4) belonging to X3 and Those terms which are of even degree in y and y2 taken together are (2)
It easily follows
,
,
multiplied
(3)
by a
by a
sine series of
having real
coefficients.
odd degree
in
The
two equations
e-
of (148)
which depend on
are
(160)
A(a?y = D Y
12
<rv^T<
A(a?y =
DY
from (137) and properties (2) and (3) that those parts of Y3 er a rli and F, e<rVTrT which are independent of the exponentials e-"^=li and effvCrT are sums of cosine terms having purely imaginary coefficients, and of sine terms having real coefficients, and that the purely imaginary coefficients are
It follows
' '
respectively identical while the real coefficients differ respectively only in M these parts of the l2 and Hence, using the explicit values of sign. form the of are found to have (160) right members
I/,
u3 -u,
v3
+u
2 [\/~-iA, cos jt
v2
-v,
i
v,+
vt
- <tV^Tu +u
2
p(ui +u
)+u
u+, p{u 3
,
u^+u' 3 +u[
2 [V^TAj cos jt
W,
1
-W
U3 + U,
03+04
P (m3
0.3-^4
>
+
real
first
(rv/
^Tw +^,
1
p(w,+tt4 )+Ws-wI>
-W )+W3+M
4
It follows
from (137) that the constant parts of these two expressions are and identical. Therefore the constant parts of the right members of the two equations of (148) are identical, and when one of them is made to vanish
l)
by a special determination of
a[
the other
I'M)
PERIODIC ORBITS.
141.
Form
X'
It follows
3
from the form of X3 and Y given in equations tion (151) have the form ~ 8 +a?f3 e-^^'+a/ /,-aJ /"^
l
(145), that/,
and g
of equa-
i)
.,<7v/=7
I
(-1)
<-3)
+/;
~"+tt
/>
-30\'=7<
fc
<
,
08
ft
e^'+af
_s>
.
</<-
e-^+a ^
(
(161)
e~
of these functions.
functions of Z We need certain properties from the properties of X3 and Y3 and of the left members of the differential equations, and from certain considerations of changes of sign of v^i t, and y3 in both the differential equations and
,
<73
are
known
It follows
/^'(V^T),
/!<-
/r
(-o=+/r
w,
(j
= 0,
1, 2, 3,
from these relations that the /"' and the g^ have the form
/<
g?
gl-
= 2 [a* cosjt+V^ib
,,
sinjt],
gf
g<"
/</,"
-2{+y=Taf cosjt+pf sinjt],} = 2 [- V^Taf cosjt+(3? sinjt], =2[+V=iaf coBJt+$f sinjt], .2 [- V^Ta* cosj<+,8? sin#],
=2[+
n)
(162)
=2[<cos^+v^T6;
= 2 [a;
1'
sin#],
,
93
/ J-
/f =2 a? cosjt,
where the
since
t/ 3 (0)
gT =2/3;
are real constants.
sin?7,
af,..., $f
It follows
=0, we have af =
from equations (161) and (162) that g3 (0)=0. af, and equations (151) become
1
Therefore,
x3 = a
3l
[e*^
1
m,
- e~" v
~
~"
uA + afW
(-3) s)
^-'+/<- e-*^'-r-/f
2)
<1)
+/V^'-'+/<,i
l
c e -3ff>
o-,
"-+/,"
"+/^
1)
e-^-",
711
(163)
+g?\
-ff\-i<
142. Coefficients of X
2
.
It
is
X are defined
by the
differential equations
x"-2y' -[]
t
+ 2A+()Aecost +
]
y"+2x[-[6Aesint+
-k = A'
4 ,
(164)
-]y<=Yt
281
where
+[-2K-6Kecost+
~\x2
+[-6Ke
1
sint+
]
1
]Vt
+ [-3B-12Becost+ + [-24J5esin+
]
][xl+2x
[x2 y2 +x
l
xi
3
ij3
+x
y3
~\
+ [#+6ecos*+
Y =+[-GKesint+
i
][yl+2y
~]x2
]+X
i!
(165)
+[K+3Kecost+
1
]Vi
+ [-12Besmt+ + [3B+2Becost+
+ [9Besint+
where
][aZ+2x x
~\
][x
y2 +x y3 +x3 y
l
']
][y +2y
2
y3 ]
+ Yif
3
and
are independent of x3
and y and
linear in x2
and y2
In
X, the terms which are of even degree in yx and y2 are multiplied by cosine series having real coefficients, while those which are of odd degree in y and y2 are multiplied by sine series having real coefficients. In the case of Yt
x
the cosine series and sine series are interchanged. or F4 then i +2 /2 is the general term in t 3j 3
If
x\ x 2 x 3
3
k'
y\
2 y"
y3
ft 1
+2A-2 +3fc
=4
or 2.
When
the right
members
is
of the equations
xi = a^ y4
e aV
^l'u
'
+a^
-"w2 +c44
Vu +<
3
e- p 'w 4
1
=
.
<
,
e'^1 Vt+a?
e- v =~>' v2
(166)
+a?
ef v3 +a? e'" vx
where
shall
a[*
subjecting
them
are arbitrary constants. Now, on varying them and to the conditions that (164), including the right members,
af
be
)'
satisfied,
we
find
= [Ai X A(a?y = [D X
(aJ)'
4l
(167)
Dn
same
as in
139.
Necessary conditions that the solution shall be periodic at this step are that the constant terms in the right members of the first two equations of as given in (165), It follows from the form of i and Y4 (167) shall be zero.
and from their properties, that these constant terms are independent of Therefore the condition that the constant a linearly. a'i' and involve of the first equation shall be zero has the form member term of the right
<+& =
0,
(168)
282
PERIODIC ORBITS.
are power series in e. It was shown in Chapter VI, in the = treatment of the case where e 0, that P2 has a term independent of e which is distinct from zero. Therefore for |e| sufficiently small af is uniquely
where P, and
determined by (168) as a power series in e. The equation obtained by setting the constant part of the right member of the second equation of (167) equal to zero is of the same form as (168); it is, in fact, identical with (168), as will now be shown. It follows from
XA and Y that the parts of the right members of the two equations of (167) which are independent of the exponentials ov=ii have the form c -,v=Ti anj e
the properties of
t
first
+Ai 2
[Aj
}
-D
tl
2 [A
B,
sinj^-D^ [- v^T
C,
cosjt+Dj
sinjt].
of exactly the same form as those encountered in 140 in the preceding step of the integration, and the conclusion follows in the
same manner.
Consequently
if
af
is
if the additive constants arising with the integrals of the last two equations of (167) are taken equal to zero, then the solutions of (164) are periodic. It and 4 that they have the form follows from the properties of t
xi = a *
i
[ef
xV - 1 'u -e1
^=it y^d^le'^'Vi _
(170)
,
J>
,
g\
f\"
and
S)
g\
of equations (162).
143. Induction to the
General
.
.
Term
of the Solution.
We shall
,
suppose
Xj yn -\ have been computed and that their x-i) y -2) coefficients have all been determined except a{" and of-0 which enter in z2 Vn-2 Zt and ?/_, in the form
the
xn _ 2 =+a?yn - 2 =
t)
[e
2
+<+a?-
V
,
v=I( v -e-'^=il v 2 ]+
l
i)
[f?- i
^ v=n+7^? er
^'+f :L +
(
(171)
283
We
shall
(p
= 1,
1)
-2
/;"
/
)
y,= 2
)
( -*>
j/:
= 2[a^ cosW+V^T6?-' sin<*] /J-^SK'" cos^-V^Tft?" sin?*] = 2[+^3Ta?-' cos^+^'sin^] pj" gf = 2[-%A=7 a-" cosrt+jS*" sin?*]
)
(jVO), (jVO),
(172)
(j^O), (jVO),
J)
0)
/i
=2 a^
0)
cos?*,
d"-2 #*
which define xn and
on**.
?/
The
(14) to
differential equations
and
be
-]x n -[6Aesmt+
xl-2y' n -[l+2A+6Aecoat+
-]fe-Xj
-]yn
y:+2x:-[6Aesin<+
where
J* +[l-A-34ecos+
=Y
(173)
,
xn _z+[-Kemit+
1
-}yn _ 2
+ [-242tesinH+ [3 + 12ecos*+
Y=+[-QKesint+
]
][x
a]
n-1
'][*V-t+1h1l-d+Xn
(174)
xn _ 2 +[K+3Kecost+
]
yn _ 2
+ [-12Besin*+
-f-[35+2ecosJ+
+[9flesin*+
[2x
x. +2x xl
2]
][xl
][2y ym. l
l
tt
-2
.
y2 +xn - y1 ]
1
functions Xj, and Yn do not involve xn -i or yn ^ , and are linear in _ and yn - 2 In n the terms which are of even degree in ft , ... , t/_ 2 are multiplied by cosine series having real coefficients, while those which are
The
of
odd degree
coefficients.
in y In the case of
x ,
. . .
y- 2
are multiplied by sine series having real the cosine series and sine series are interyt-'i
changed.
If x'i
_ a;i"_ i t/J'
is
or
n,
then
(175)
h+2j +
t
+(*-l)i._1
members
of
ii
+M-2+
+(n-l)^_ = n
1
or
n-2.
shall
be periodic are
(176)
It follows from (174) that these constant n_2> involve aj The coefficient of and terms are independent linearly. v of a'"~ is distinct from zero for \e\ sufficiently small, for in Chapter VI it
no constant terms.
a~
284
PERIODIC ORBITS.
to be distinct
was seen
for e equal to zero. Therefore a{"~ 2) is uniquely determined as a power series in e by setting the constant term of the right member of the first equation of (176) equal to zero.
from zero
It can be shown, precisely as in the discussion when w = 4, that the constant parts of the right members of equations (176) are identical. There"~ a fore a is uniquely determined by the conditions that the solutions
(
It follows
_,;
y- u and from
x
(175), that
when
B)
xn = af e*^' w
+< e-"^
[ff> %e
-
1'
^+a<
<-
e"
+a w e"" u,
4
+aJ
yn =
M,
l(+/
v2
e -*rv^
+/ ]+ S
<0>
e >*V=i
<
n)
(f
+ae>"
v3
+a? e~
+
g
(177)
+<-"
where
a[
,
.
gf
f^'+g*-* e-*=-+g?] + 2
e"^'
(n-i)
gf
ft *,
gT,fn\ and
rf
/,
_2)
/^
/n
(0) ;
In order that (177) shall be periodic it is necessary and sufficient to impose the conditions a = a=0. Then it follows, from the properties
of v lt v2
,
g?, glr",
n)
that a
=
xn = a
aj
Since y (0)
= 0,
it
follows
e"
v=Xl
- e~' v=l
'
u2
(-
+ a (-.) [f V ^v^I, +/
aV=T
2)
g-Sffv^Ti
(178)
+a?- [g?
l)
<?
'+?n
-*
e~*^
+g?\
2 g + J=-n
e"^-
satisfied for
p = n.
In picking out the constant part of the right member of the first equation rV=0 of (176), in general only those terms in n and n which contain e as a factor to the first degree will be used. But because a is a rational
powers of the exponentials, multiples of <r which are integers, and constant terms in the right member of the first of (176) may occur from these terms, but their presence does not prevent the determination of the constants so that the solutions shall be After such terms once appear, they occur in general at each periodic.
succeeding step of the integration.
in the higher
Chapter
VIII.
fl
BODIES.
Statement of Problem.
showed
In his prize memoir* on the problem of that, for any three finite masses mutually
distinct configurations such that, under proper initial projections, the ratios The bodies describe similar conic of the mutual distances remain constant.
sections with respect to the center of mass of the system, the simplest case being that in which the orbits are circular. In three of the four solutions
and in the fourth they remain at the vertices of an equilateral triangle. This memoir is one of the most elegant written by Lagrange, and its mathematical form does not seem capable of improvement. But the method which he employed can not be extended conveniently to the case of more than three bodies, and it has not led to
lie
the masses
always in a straight
line,
practical results in applied celestial mechanics. This chapter is devoted to the solution of two closely related problems I. The number of straight-line solutions is found for n arbitrary positive masses; that is, the ratios of the distances are determined so that under
:
initial projections the bodies will always remain collinear. This is the generalization of Lagrange's straight-line solutions to the problem of n For each straight-line solution of n finite masses there arc n-fl bodies.
proper
points of libration near which there are oscillating satellite orbits of the -VII. Therefore the results of this chapter types treated in Chapters
lead to generalizations of those of the preceding three chapters. II. The problem is solved of determining, when possible, n masses such
they are placed at n arbitrary collinear points, they will, under proper initial projection, always remain in a straight line. The first problem, in a somewhat different form, has been considered by Lehmann-Filhes.f The method of treatment adopted here J, though originally developed independently, has considerable in common with that
that,
if
of Lehmann-Filhes,
German astronomer.
masses
may
286
PERIODIC ORBITS.
In the second problem it is proved that when the number of arbitrarily chosen real collinear points is even, the n masses are, in general, uniquely determined by the condition that it shall be possible to place them at these points and to give them initial projections so that they will always remain
collinear
in orbits of specified linear dimensions. Or, if it is of revolution can be taken as the the arbitrary in place period preferred, In general, the masses will not all be of the linear dimensions of the orbit.
positive,
and revolve
pretation.
When
not always have a physical interthe number of points is odd, it is not possible to determine
will
the masses so as to satisfy the solution conditions unless the coordinates of When they satisfy the points themselves satisfy one algebraic equation.
this condition,
all of
any one
of the
masses
may
be chosen
arbitrarily, after
which
I.
145. The Equations Defining the Solutions. Let the origin of coordinates be taken at the center of gravity of the system, which will be supposed This point and the line of initial projection of any mass to be at rest.
determine a plane. All the other masses must be projected in this plane, for otherwise they would not be collinear at the end of the first element of All the bodies being initially in a line and projected in the same time. Consequently, if solutions plane, they will always remain in this plane. exist in which the n masses are always in a straight line, the orbits are
plane curves. Let the plane of the motion be the i^-plane.
noted by
(&>
V2),
m
is
1(
>
mj,
,mK
and
(,
Vn)-
Then,
their respective coordinates by (,, 77,), choosing the units so that the Gaussian
constant
(Ph
dt
2
dU
'
mt d&
n
287
n bodies remain collinear, the line of the resultant acceleration to which each one is subject always passes through the origin. Therefore, in collinear solutions it follows from the law of areas that, for
each body separately,
*" 1
'
'.*)
collinear
we have
also
d6 1= '
dt
= djh.
dt
'
from which
it
follows that
i
j
\m
Kc,
t
(3) '
Cj
from
this
if any collinear solutions exist, the from the origin are constants, and it that the ratios of their mutual distances are also
is,
That
constants.
They are therefore of the Lagrangian type. If the n masses remain collinear, the ratios of their distances from the
all collinear
origin being therefore constants, the ratios of their coordinates are constants.
Therefore in
solutions
<
= Z,,
Upon
Vi
= X,v
(i=l,...,n),
(4)
substituting in equations (1), as necessary conditions for the existence of the collinear solutions,
x,
where the
are constants.
we have,
dt
t
l
[(x i
-x )T'
J
(J
*'
*>
(5)
de
xiLix,
- xj)T'
r"
r" v* +'
$ and rj as defined by their initial values and equations be the same for all values of i, the coefficients of /r* and rj/r3 must 2 be set equal to a constant independent of i. Letting -co represent this i constant and rij = -\/(x these conditions, which are sufficient as well Xj)
In order that
(5) shall
become
-a
run
xi)
|
p-
(6)
mi (x
3
'nl
nh(xn
^3 '
~-\
Xi) -
m (x
3
x3 )
"r
-1
zr~
'
-r
n u
_. x_
idj,
Hi
288
PERIODIC ORBITS.
;
< x Suppose the notation is so chosen that in any solution z, < then the terms of the left member of the last equation are all positive. Since the origin is at the center of gravity, x is positive, and therefore or is For every set of values of a;, xn positive in all real solutions. satisfying equations (6) the solutions of (5) are the same for all values of t, and these solutions substituted in (4) give the coordinates in the collinear
, .
. .
configurations.
Since equations (5) have the same form as the differential equations in the two-body problem, it follows that in the collinear solutions the orbits are
always similar conic sections. In case the orbits are ellipses, the coefficient 3 3 of 77/r is the product of the cube of the major semi-axis of the %/r and If the orbit and the square of the mean angular speed of revolution.
is
the
numbers, and the problem is any value of n. For each of these solutions there
velocity of revolution of the system is co. that w 2 and n are real positive to find the number of real solutions of (6) for
is a six-fold infinity of collinear configurations, the six arbitraries being the two which define the plane of motion, the one which defines the orientation of the orbits in their
which determines the epochs at which the bodies pass their apses, the one which determines the scale of the system, and, finally, the
plane, the one
eccentricity of the orbits.
146. Outline
of the
Method
of Solution.
The method
of solution
involves a mathematical induction and consists of the following steps: Assumption (A). It is assumed that for n = v the number of real x v is whatever real positive values o> 5 solutions of (6) for xt , , . v and m, may have. It is known from the work of Lagrange that when v = 3 the number is AT, = 3 = \ 3
. ,
N
.
If to
m v+l
the system m w of positive masses an be added, then the whole number of real solutions
x
, .
.
(H-l)tf,.
Theorem
to
(C).
As the
.
infinitesimal
mass
ni r ^
any
finite positive
number
remains precisely (v+l)N Conclusion (D). From successive applications of theorems (B) and
(C)
it
number
n = v+n
.
is
N
Since
v+lx
=(v+n){ v +n-\)
+2)(v+l)Nv
.
A =
r
3!
it
Let
n+3 = n
it
and we have
(7)
N
To complete
=~nl.
remains only to
289
and the
infinitesimal
Theorem {B).~ When there are v finite body m +l equations (6) become
,
. .
.
bodies
l}
.-.
,m,
Vi
^
_ -
^ X2
fniJXi-Xi}
|
|_q_|_
IS
+ m,(x m
i
-x,)
-a;y)
+ rn,+L(x -x
1
]!+J )
^
_
(8)
y (Xit
~3 '2w '2c
T mH-ife-rc+i) f+1
~3
'
9 2,
#v
m^rc-rcg)
-f-
'lv
,
,
3 'SK
t
i
n tui
,
/n y+1 '(a; y
is 'iM'+l +1
a; y +i)
=u,
<Pv+i=
w(x
-3
-x,)
-f
n=n U. hU
The
tesimal).
last
column
of these equations
is
zero because
w +1 =
,
(is infini.
x Consequently the first v equations, which involve x2 alone as unknowns, are the equations defining the solutions when n = v. By (A), it is assumed that there are precisely y real solutions of these Let any one of these solutions be x1 = xf\ x = x equations.
.
.
Then
Vi/+i=
~~
,.2-v u x v+i~r
1
7n\\X v +\
(o)3
'
1.
1 I
TThK
) Xy+iXi _ w)3
'
-r
v [x v +i
xv
r+l
2.
c+1
zws '
v,
n V.
(V)
-,
J/+1
The number
is
required.
Consider
as a function of x
+1
It
is
*v+
i(+<)=-5,
lim
""o^ +1 (*r+)=+*>
lim
,
y-i.
ti-1, ...,")
*r+l(->) =
Since ^
+>,
0^-*)= -00
(10)
and continuous except at xy+1 = xf\ it follows that there is an odd number of real solutions 00 to xf, where x is the smallest xf, x intervals which are adjacent, and xf and are two xf x, any x'f re is the largest xj". But we find from (9) that
+1 is finite
.
. .
xf,-\- 00,
<0)
00,
in
<0>
to
+00, where
d<p+i a~ "Cy+l
_ =
2m
(0)2
2/na
CO
ZMi
'2,
'
'
'
2?n y
r 'o>2
'v,
'
'l,
v+1
c+1
v+1
= x +1 = x where it is infinite. negative except at x +1 Xj function in each of the intervals, Therefore p +1 is a decreasing monotonic and consequently vanishes once, and but once, in each of them. Since
which
is
,
.
. .
290
there are
first v
PERIODIC ORBITS.
of these intervals, there are, for each real solution of the 1 real solutions of the last equation equations of (8), precisely v
v+l
of (8).
tions,
Since the
first v
real solu-
+ l)N,
real
solutions.
v Let x } =xf fj-1, \) be any one of the {v-\-l)N v real solutions of equations (8) which are known to exist when w +l = 0. It will be shown that as m +1 increases continuously to any finite positive quantity whatever, the xf can be made to change continuously so as always to satisfy equations (8), and that they remain distinct,
148.
finite,
and
real.
From
this
it
+ l)N
,
m,
m v+l
It will also be shown that no new solutions can appear as m, +1 increases from zero to any finite value. Consequently, it will follow that the number
is
exactly
.
(V
\)N V
the masses
m v+l
roots of algebraic equations are continuous functions of the coefficients of the equations so long as the roots are finite and the equations
The
do not have indeterminate forms. Consequently, the xf are continuous m v+l if no xf becomes infinite and if no xf = xf The real roots of algebraic equations having real coefficients can disappear only by passing to infinity, or by an even number of real solutions becoming Therefore we have to determine conjugate complex quantities in pairs. whether finite can become (1) xf any equal to any xf, (2) whether any xf can become infinite, and (3) whether any two real solutions can become
functions of
.
ra +1 complex for any finite positive values of m, mp+1 by hypothesis, are (1). The masses m the notation be so chosen that for any values of m,
, .
.
all positive.
.
Let
the
xf
are
xf changed the difference xf approaches zero in such a way that xf and xf remain finite; that is, rtj which occurs only in the expressions <p, and <Pj approaches zero. Suppose i < j. Then the term involving ru becomes negatively infinite in <p and Consider <p = 0. Another r(t must approach zero positively infinite in <p s and the term in order to restore the finite value of the function <p r Therer infinite as must become u approaches zero. positively involving lk
satisfied.
, , t
.
xf
<xf <
But rrt enters besides only in <pt and similar reasoning shows k<i. In this manner we are that rm where I < k, must also approach zero. driven to the conclusion finally that an rM where one of the subscripts is
fore
,
, ,
zero.
Then
consider
its
^ = 0.
viz.
r(J
r vt
291
.
.
that some rt] can approach zero for finite values of xf, xf +l and finite positive values of m, leads to an and it is y+l impossibility, therefore false.
,
. .
.
(2).
On
and adding,
it is
(8)
by m
m +1
respectively,
cf(ml x1 +mt zt +
It follows
if
from this equation that no xf alone can become infinite, and that one of them becomes negatively infinite then some other one must become
Suppose the notation
is
positively infinite.
Z<X?<
Then,
if
<*<*&.
any xf becomes negatively infinite xf must also become negatively infinite, and from the equation above it follows that x +l must become Now suppose this occurs and consider <p = 0. In order positively infinite. that this equation may remain satisfied, xf must also become negatively infinite in such a way that xf xf shall approach zero. But now it follows = from <?, 0, since orxf and m^ixf xf)/r\ are both positive, that xf must also become negatively infinite in such a way that xf xf shall that xf must become approach zero. Then it follows similarly from <p
x
negatively infinite in
infinite.
such a
it
x| xf +l must all become at least become But must xf positively infinite. negatively +1 Therefore xf can not become negatively infinite, and similarly xf +l can not become positively infinite. Hence no xf can become infinite.
xf
0>
,
.
shall
.
approach
,
zero.
This
+I approaches zero, equations (8) remain always determinate, and that there are accordingly no solutions besides those obtained in theorem (B), consider a solution Zi,..., xv+l in which the x are all distinct for a set of positive values ra +1 and then let m, approach zero as a limit. of ?!,... In the first place, if x, approaches neither xi+l nor x ^ as a limit as m approaches zero as a limit, then by the reasoning of (1) and (2) above no Xj can approach any x t as a limit. In the second place, x can not approach x i+l as a limit as m approaches can zero as a limit unless x ^i approaches xt+l as a limit, for otherwise cp = not be satisfied. But if x,_i approaches x i+ as a limit as m approaches zero and <p, + = can not be satisfied unless _ 2 and x(+3 as a limit, then <p,_i = This shifts the difficulty to ^,_ 2 = respectively approach x as a limit. and <p, +2 = 0, and so on until <pi = and <p +1 = are reached, which can not be satisfied under the hypotheses it has been necessary to make. In the third place, x, can not become positively infinite as w, approaches = can not be satisfied unless ,_i becomes infinite in such zero, for then <p
In order to prove
now
that, as ra
and
their solutions
292
PERIODIC ORBITS.
x,
Continuing through <p _ = 0, we are led to the conclusion finally that x xv+1 all become positively infinite, but then the center of gravity equation can not be satisfied. Consequently the solutions all remain regular as m approaches zero as
a way that
x,_i
approaches zero.
a limit.
Since the solutions of (8) are continuous functions of = can ever follows that no two solutions which are real for +l
(3).
m +1
it
become
any m, +l without having first become equal; and, conversely, no two solutions which are complex for m y+l = can ever become real for any real value of m +1 without having first become equal. Consequently, if a multiple solution of (8) is impossible for
conjugate complex solutions for
real value of
every set of
finite positive
values of
any real solutions should disappear by complex solutions should become real. The conditions that x = x m shall be a multiple solution of /(#) =0 are w that /(a; ) =0 and df(x )/dx = 0. The corresponding conditions that a set of simultaneous algebraic equations shall have a multiple solution are that a set of values of the variables shall satisfy the equations and that the
co>
m +1
it is
Jacobian of the functions with respect to the dependent variables shall vanish for the same set of values. That is, the conditions that
Xj
= xf
(j
= l,
,i/+l),
(8) are
dXi
dx2
'
dx+i
d<f>2
A=
dXi
dXt
dx, e+l
= 0.
(11)
d<pV+i
t
'
dXi
dXt
dxv +i
Consider two solutions of a set of algebraic equations having real coefficients. As they change from real to conjugate complex quantities, or from conjugate complex to real quantities, for some value of a continuously varying parameter, then for this particular value of the parameter they arc not only equal but they are real. Consequently, it is necessary to examine A only when all of its elements are real. It will now be shown that it can not vanish for any set of real values of the x } when m +1 are positive, and consequently that it can not vanish for any particular set which satisfies equations (8). When this is established, it will have been proved that all the solutions of (8) which are real for w>+1 = remain real
when mr+l
increases to
remain complex.
293
From
v*
equations
(8)
and
M,
A=
294
PERIODIC ORBITS.
same form as the original determinant, and the sum of the elements of its (J row i s mjr\ Consequently every term in the expansion of the w2 mji* minor which does not vanish will contain at least one of the u But these elements appear only in the main diagonal of the as a factor. minor. Hence all terms in the expansion of the minor which do not vanish depend upon at least one element of the main diagonal. In considering our
tft x
.
particular term it may be supposed, without loss of generality, to depend upon the first main diagonal element of the minor. In the expansion of the
original determinant the product of these two diagonal elements will be multiplied by the co-factor of the minor of the second order of which they
This co-factor has the same properties as the first minor just considered, and in the same way it is proved that at least one of its diagonal elements must be involved in the term in question; that is, the term under consideration depends upon at least three elements of the
are the
main diagonal.
continuing in this manner it is proved that any term in the final expansion depends upon all the elements of the main diagonal, which are all of the same sign in every one of their terms. Consequently,
main diagonal.
On
all
the terms which do not cancel out in the expansion of the determinant v+l and it has been seen that there is at least one such l) (
;
Therefore the determinant not only can never " can never be less than a> 2( +1> in numerical value. Since A can never vanish for real distinct xf when all the m, are real and either zero or positive, it follows that no real solutions can ever be lost or gained as the nij vary so as not to become negative, and therefore that
1)"
.
+1
w2(
" +1)
vanish, but
it
the
number
is
2
.
(v-f-
l)N y = % (V+
finite
values of
l ,
w +1
and w
Computation of the Solutions of Equations (6). There are wellknown methods of finding the roots of a single numerical algebraic equation of high degree, but they are not readily applicable to simultaneous equations of high degree. However, when the order of the masses has been chosen, will become polynomials in x x after they have been equations (6) Then by rational processes n 1 of the x can be cleared of fractions. eliminated from these equations, giving a single equation in the remaining unknown. The solutions of this equation can be found by the usual methods and the results can be used to eliminate one unknown. By repeated application of this process to the successively reduced equations, the soluThe one satisfying the conditions of reality of tions can all be found. x n and their order relation is the one desired. x The solutions of (6) can also be found by a method closely related to that by means of which their existence was proved above. Suppose for = a solution of mi = raf (t = l, ft) x, xf equations (6) is known. The mf are supposed to be zero or positive. Suppose it is desired to find the corresponding solution, that is, the one in which the masses are arranged
149.
t
,
. .
.
295
on the
. .
line in the
(
same
order, for
x satisfying (6) be the are functions of = be to determined. On m Mi,*-. & a,ndm = mf+ij, substituting #, xf in (6), making use of the notation of (8), expanding as power series in the and n (which is always possible, since it has been shown that no xf can
set of the
, , 4
become
is
infinite
a solution of
(6) for
m = mf
t
(13)
where the
are the symbolic powers used in connection with the power-series expansions of functions of several variables.
(13)
The determinant of the terms of the first degree in the %t in equations is the A of equation (11), which has been proved to be distinct from
Therefore equations (13) can be solved by the method and 1, by 2 the solutions converge for |ju|>0, but sufficiently ^ r. Keeping the fi within this limit, Suppose they converge if
|/z<|
t (
explained in
small.
can be used as a starting-point computed. which can be repeated as many times second of the for a process, application as may be desired. m n have any Hence, to find the solution in which the bodies m finite positive values and lie in a determined order on the line, we may start with m m2 and m3 and solve the Lagrangian quintic* which defines their Then an infinitesimal body m 4 is added and its distribution on the line. is found position by solving the single equation (9), in which v = 3. This
a solution x
l
=x
x)
is
Then
this
mass m x is made to increase, step by step, to the required finite x4 are computed. It value and the corresponding values of x, follows from the fact that the dy^/da;, are less than fixed finite quantities 2 while A is not less than w 8 in numerical depending upon w and reached in this way by a finite 4 can be value, that any finite value of
infinitesimal
.
m m
number
of steps.
Not5 etc., to any finite number of bodies. process can be repeated for withstanding the fact that this would be very laborious if the number of bodies were large, we must regard the problem as completely solved both
,
theoretically
and
practically.
I, p.
290
PERIODIC ORBITS.
II.
150.
the n
Determination of the Masses when n is Even. Suppose co and x xn are given, and consider the
l ,
.
mn
exist.
There
will
be no
,<,<
necessary and
<,. With
which are
become
8*-+ '12
+^
'23
'l.n-l
ft
BODIES.
297
(14), to the set
+m
1
xi
+m
= 0, which
is
a consequence of
of equations.
if,m
,...,m.
1 linear homogeneous equations in In order that they shall be consistent their eliminant
o,
E=
298
PERIODIC ORBITS.
as a function of x
Then
From
for
the inequalities x2 < x3 and r]2 < r13 it follows that dF/dx is positive oo <x t <x2 Therefore there is but one solution of (17) for x,<x2
, l
.
when xt and a
If Xi is
positive x3 are chosen. By symmetry, there solution of (17) for x3 >x2 when x2 and a negative , are chosen.
is
but one
negative and x3 is positive, but both otherwise arbitrary, considered as a function of x2 gives
2
/
*i
2J
/
?*<<).
/jj
d9)
U^/j
13
Therefore there
is
(17) for x 2
which
satisfies
defined
by
(17).
a positive x3 and m2 are given arbitrarily and Then equations (14) give
,
m-*B [+>*-j?}
If
m^rl^-^xIf
^]2
(20)
is
negative,
and
is
positive
and
As m 2 increases, m^ and m, sufficiently small, both m^ and m 3 are positive. decrease. Suppose a;3 > x, Then, for a certain positive value of ra2 the mass m3 vanishes while m 2 is still positive. For a certain greater value of mj the mass m is zero and m3 is negative. For still greater values of m, From the fact that x, must be negative both mj and m3 are negative. and x3 positive, and from equations (20), it follows that not all three of the masses m,, m 2 and m3 can be negative simultaneously.
. ,
Chapter
IX.
EQUILATERAL-TRIANGLE POINTS.
By Thomas Buck.
153. Introduction. This chapter is devoted to an investigation of certain periodic orbits which an infinitesimal body may describe when attracted according to the Newtonian law by two finite bodies revolving in
been shown by Lagrange that three bodies placed at the vertices of an equilateral triangle can be given such initial projections that they will retain always the same configuration.
circles
It has
The
orbits here considered are in the vicinity of the equilateral-triangle points defined by the two finite bodies. The infinitesimal body is displaced
from the vertex of the equilateral triangle, and its initial projection is determined so that its motion is periodic with respect to that of the finite bodies. The existence of the solution is established by the method of analytical continuation. The construction is made by the method of undetermined
The
using the properties obtained in the discussion of the existence. solutions are given in the form of power series which converge for sufficiently small values of the parameter employed.
coefficients,
The motion of the infinitesimal body 154. The Differential Equations. be referred to a rotating system of axes, the origin being at the center of mass, the ^77-plane being the plane of the motion of the finite bodies, and the The masses of the finite rate of rotation such that they remain on the |-axis. and 1 n so taken that m^s> their distance bodies will be represented by unit of distance, and the unit of time will be the will be taken as apart 2 Then the equations so chosen that the proportionality factor k is unity. of motion for the infinitesimal body are
will
/j.
<tl_
df
dr L
dt
_dU "
dt
d?v_
'
dt
"'
l9 c%
dt
= dU "
'
tn " = dU
dt
2
dr,
df
'
m
K
where
r^Viz+nr+f+F,
and r, being the distances from the and m respectively.
r,
infinitesimal
body
to the bodies
299
300
PERIODIC ORBITS.
The Lagrangian
I.
&-J-M,
Z
*- + jVS,
W
*"
f,-0.
II.
=2~
fl ,
yVf,
= 0.
The two
points in the rotating plane defined by these solutions will be The question of the existreferred to as point I and point II respectively. ence of periodic solutions of equations (1) in the vicinity of these points is
to be investigated.
in question
is
g-i-H-sr,
After the transformation
is
n=\Vz+y,
made
the right
f-#.
(2)
When
point I
is
given
by the values
of x, y,
and
2
-Kx +f+z--{-x-\-\^y<+l,
This region consists of the
-Kx +y +z--x+Vsy<+l.
2 i
portion of two spheres, excluding their For point II the centers which are at the finite bodies, each of radius V2. is defined the of by convergence inequalities region
common
-Kx'+y'-hz'+x-VsyK+l,
-Kx-+y
-\-z'
-x-Vsy<+l.
Since the origin in this case is at the second point it follows that this region is the same as that found for the first point. As the two cases differ only in the sign before the Vs, it is necessary
The discussion will be given for to consider in detail only one of them. that the I with understanding by changing the sign of Vs the correpoint sponding expressions for the point II are obtained.
Then, parameters, e and 5, are introduced as in Chapter V. denoting derivation as to r by accents, the differential equations become
Two
[X
...
.
J,
.
],
],
(3)
z"
= (l-H)
[Z1
+Z,+Z *+
l
301
where
X1= + iaH-|V3(l-3M)z/, X
2
F = + fV3(l-2 M )o;+fz/,
1 2 2
=+
],
Y = + ^[V3x?+22(l-2iJL)xy+3V3tf-4:Vzz%
2
Z,
= -z,
3 [
Z = + f[(l-2 M )xz+V%2],
2
(4)
2/x)*/
F = + ^[-25V3(l-2M)a;
3
-r-123x
2/
2
+ 135V3(l-2/i)a;?/
2
+37/
-60\/3(l-2/x):cz +132i/z
],
^ =3
[x
For
z,
and
all
convergent.
Equations
(1)
The corresponding integral of equations (3) can be series in e. The terms independent of e are
4[x +2/' +z' ]-(l-r-5)
,2
2 2
[3a;
+% -4z -r-6V3(l-2
2 2
u)a;t/]
= const.
(5')
These terms
will
155.
The
Characteristic Exponents.
For
=5=
equations
(3)
become
x"-2y'- f x- f
Va(l-2^ = 0,
2/
= 0,
(6)
first
immediately, giving
=c
302
PERIODIC ORBITS.
To
integrate the
first
two equations,
XT
,
let
x = Ke
y = Le \
x
On
substituting in the
first
two equations
of (6)
Xr
,
we have
from
In order that these equations may be satisfied by values of and L different the determinant of the must vanish. This zero, system gives for the
V+X +f M (l-M)=0.
2
(8)
Each
(7).
of the four values of X satisfying this equation gives a particular solusatisfy equations Since these equations have a vanishing determinant the ratio only of and L is determined. In what follows will be considered as arbi-
The corresponding
K and
K
L must
the
and L will be determined in the In order that a solution shall be periodic, the corresponding X must be a purely imaginary quantity. Upon solving (8), we have
trary,
form L = bK.
M (l- M ). v= -lVl-27 2
For small values of n the roots of (8) are pure imaginaries; the limiting value of ju for which this is true is given by the equation
1-27m(1-m)=0.
root of this equation which is less than \ is /i = .0385 .... For X the of are and the values .0385 purely imaginary corresponding (i4* Let o- and <r2 be two numbers defined by particular solutions are periodic.
The
2
1
_ 1+Vl-27 M (1 -n) ^
<t x
Ot
,_
1-Vl-27 M (1- M
r.
do not exceed unity for n ^ .0385 and that oy^aj. Then the roots of (8), which are the characteristic exponents of <ra \/ the problem, become 1 and 1(r,V
It follows that
and
<r2
303
is
The Generating
x= a e" v
1
Solutions.
The
- lT
a 2 e- aiV
2
-lT
a3
- iT
+
4
at e""/=lr
a4
e
y=6,
1
a,
= e* v Ir
2
+^
e-r'^1T +&3 a3
e*^ T +6
- ff
T
,
z=c sinT+c
The
,
cosr.
quantities a x a 2 a 3 a t c t and c2 are arbitrary, while b x b2 63 and 6 4 are determined by equations (7) when the proper values of X are substituted. Thus it is found that
,
, ,
_.
+8(r
+8<y/^I-3 V3(1-2m
-
h =
^ V^l3V3(l
l
-8o-2 \/^ri-3V3(l-2M)
4<rf+9
4a|+9
Various periodic solutions are obtained from this general solution by assigning suitable values to the arbitrary constants and to the quantity n- For
x = al
e'*
v- lT
+a
- lT e-'riV ,
e-'" v
y = b1 a l
- lT e lV
e
a'
+b
a 2 e- asV
- 1T
,
= 0.
= 0.
II.
~ x = a3 e aW lT +ai
-lT
,
y = b3 a 3
v- lT
+b
a i e- (nV
- lT
,
These equations represent ellipses in the xy-pl&ne with centers at the origin. The major axes of the ellipses coincide and make an angle 6 with the positive
x-axis defined
by
tan20=-V3(l-2M),
with cos 20 positive.
and
less respectively
The major and minor axes of the second are greater than those of the first. The periods are 2ir/<r and
1
If
/x
= .0385
it
follows that
<r x
<rt
and solutions
For
III.
all
values of m
we have
y
x = 0,
= 0,
This solution defines an oscillation on the z-axis with the period 2t. It is possible to give n values such that <r and <r2 are relatively commenLet m2 (r = m (r2 where m and m, are integers. Then, by using surable.
x
the definitions of
<r x
and
cr2
we
find
ml _ ml V l-27 M (l- M )= m2 + w2
304
PERIODIC ORBITS.
. . .
the expression on the left takes all values on the interval 2 By choosing m, and m2 so that 0< {m\ 2 )/(w 2 ) < 1, and for we have a of the value and <r2 comn <r, n, making solving equation For such of we the mensurable. values have additional periodic solution /x
+m
x=
IV.
at
1
* e' v-*
1
+
t
Oj e-
v~l T
+
3
a3 e"
+
+b
i
y=b
a1
e'
w-lT +b
a 2 e-"lV
=lT +b
a3
e'*
v-lT
^lT
,
is
By
may
it
can be shown
the characteristic exponents (rt and a2 separately be commensurable with unity. We have then the periodic solutions
c'iv-it
V.
*=* r y = b a e"
1 l
+ +b
a 2 e-"lV 1T
a2 e""^-1 '
,
z=c
VI.
x= y=b
a3
3
e u,v
<r%
- 1T t Vl
a3 e"
^
2
+ +b
a4
t
e~-ffiv'-^Ir
T
,
a 4 e- *=
z=c sinT+c
1
cosT.
The
and 2mr = 2n 2 ir/a2 respectively. a and a2 may be commensurable with unity for the same
2m
ir/<r 1
value of
n.
c<r2
= d;
then
\/i-"27m(1-m) =
262 -cr
or
<?-W
b
2
If now we give a, b, c, ). values that this relation is satisfied and such that integral 2 2 (26 a?)/a lies between and 1, the corresponding value of fi will make a x and <r2 commensurable with unity. For example, such a choice is
From
this relation
it
and d such
a = c=13,
6=12,
d=
5.
p.
we have
<r,V-ir_|_
^g-nv-lr^.
2
a4e
t
-<"V=lT
VII.
y = 6,
z
<*!
e <r,v/=i
+ &, a
is
e"''^
+ 6, a e*^ +b
1
'
at e-" v
T
,
=c
The period
of this solution
2rmr
= 2m
Tr/(r l
= 2m
Tr/<r2
305
These periodic solutions are the generating solutions for the general problem. We shall now suppose that e is not zero and consider the question
of the existence of the continuations of these solutions
parameter
e.
The
The
period in
157.
For
of equations (3)
x=
e Via+i)V
~lT
y = b^e*
z
1
a+5) v~' T
ai
'
*>
vZZi T
,
(9)
=c sin(l + 5)r
-fc2 cos(l
+ 5)r.
by the transformation
Normal
X y
=b
(10)
<r,
xt)
The
differential equations
then become
x[- Vl (l + S)y^Tx
x' 2
= A (X e+X
l
e'+
)+5 )+B
(F2 e+F
),
+v
(l
+ 8)V=ix = A
2
(Y2 e+Y3 S+
),
xg-rt (l+l)V=T%- A (X
3
e+X
e*+
i
-)+B (Y
3 4
t+Y e+
3 J
),
(11)
x4 + cr2 (l
+ 5)v^ix = A
4
(X
t+X t +
3
2
").
e+Z e+
3
A =(l + 5)*%,
t
B =(l + 6)*%,
t
the determinant of the transformation (10), and AJt the minor of an element in this determinant. The first subscript indicates the row and the
A being
306
PERIODIC ORBITS.
For
=5=
xt
x3
and xt are
the initial
Od
we take as
^,
= 0,4-0,,
,
x3 = as +a3
3J4
2=0,
j
= a,+o2 x,
Since there
is is
=a
+a
z'
= c+y.
a component of force always directed toward the xy-pl&ne, it Hence we have supposed that clear that at some time 2 must be zero.
at t
= 0.
,
14 and 15 equations (11) can be integrated as power According to in the series parameters a 1} 04, a3 a 4 7, 5, and e, which converge for |aj| |e| sufficiently small, and for O^t^T, the value of T, which in These solutions have the is the this case period, being given in advance.
,
. . .
,
form
x^ia.+a,)
z3
(a,, a,,
a3 a3 a3 a3
,
a4
y, y,
(a,, a,,
Oj,
,,
o4
a4 a4 a4
,
xA
z
2'
(13)
, ,
7, 5, e; r), 7,
,
=(c+t) 8^(1+5)7+6^^0!,
Oj, a3
5, e; t),
= (l + 5)(c+7)cos(l + 5)T+ep,
(a 4
Oj
a3
a4
7,
5, e; t).
The
are
x (7 )-x (0)=0
4 1
(t=l,
...
,4),
(14)
z(T)- 2(0) = 0,
These equations are
2'(T)-2'(0)=0.
On
solving
them
a,
a3
a4
7, 5 as
power
series in
e,
a determination of these quantities is obtained such that the corresponding solution is periodic. Hence, on substituting these series in (13), the resulting
These expressions can be expressions for the x 2, and 2' are periodic. in e which will as series converge for e sufficiently small, power rearranged
t ,
o and for all <! r ^ T, provided the values of a 7, 5 obtained from (14) lie in the domain of convergence of (13). The convergence can be secured by imposing the condition that the expressions for the arbitraries The obtained from (14) shall be power series in e, which vanish with e.
t
, .
.
solutions are then analytical continuations of the generating solutions. The periodicity equations will now be set up for each of the generating
solutions,
I
and the
will
possibility of solving
required form
point
only.
be considered.
in
the
for the
The
307
The First Generating Solution. The now to be determined for T = 2ir/a and
1
one of
= a = c = 0.
we
let
x l = ale +n '=lT +y
*
*b%
t*2
z 3 = 0+2/ 3
Z
z'
= Q + W,
*4= 0+2/4,
3
= 0+w',
where y (Q)=ya (0)=y (0)=yi (0)=w(Q)=w'(0) = 0, we find that the partial derivative of the integral (5') with respect to y2 is
1
Mcr\(2<j\-l)a
(4^+93
for r
= 2ir/<r and
1 ,
= y = y = y = w = w' = 0.
i
l
The
solved uniquely for y2 in terms of y l; y3 y i} w, and w'. quantities are periodic it follows that ya also is periodic.
,
Therefore the On computing the second equation is redundant and can be suppressed. the terms of it is found that (13), necessary remaining equations have the
form
8
(2*V-li)+
308
PERIODIC ORBITS.
zero.
If the coefficients of a,
from
and
a, in
respectively are also distinct from zero, it solution for a 3 a t and 5 as power series in e, vanishing with
, ,
It
is
neces-
<r x
jWEI "
l
27T
,
1,
*'
2ir
1,
sin
cos
two vanish if <T^ = ma xl where m is an integer. this occurs only when <r, = <r 2 Since <r, and <r 2 are positive and <r ^ <r2 This case will be treated later in the discussion of the commensurable The last two It will be shown that orbits exist in this case also. cases. of an if a is the vanish Suppose, then, reciprocal integer. only expressions
shall not vanish.
The
first
solving the first three equations of (15) for a3 a 4 5 2 and substituting in the last two, there remains, after dividing out e and y, a term in each independent of the arbitraries. There can be, then, no solution
that
<r l
= 1/w.
On
of these equations in the required form. They can be satisfied only by = Hence equations (15) have a unique solution of as before. putting 7 the same form in this case also.
tions will
of the existence of
an orbit re-entering
is
after
m
1
revolu.
considered.
The
2mx/o-
The
periodicity equations
such that
have a unique solution, as before, except when n is In this case the second and third equations do not m^ a^. This case will be treated later in the discusadmit solutions for a 3 and a 4 It will be shown that orbits exist for sion of the commensurable cases. Hence there is a single orbit re-entering after m these values of m also. But one such orbit is obtained \>y m repetitions of the orbits revolutions. It follows, therefore, that this orbit is the re-entering after one revolution.
.
only one.
The
periodicity equations
remaining arbitrary. arbitraries is obtained by fixing the origin of time. that x' = at r = 0. This gives the relation
<t x
Since
have now been satisfied with a and a? still we now have z=0, one relation between these
x
It will
be supposed
(a x
Oj
+a
a,)
<r,
(a 3
a4 )
= 0.
,
The same
=a
We
have then
(i
- o) +
<r2
(a,
- a = 0.
4)
x
.
This equation may be regarded as determining c^ in terms of the arbitrary a There will then be in the final solution the two arbitraries a, and a besides Since a, and a, occur always in the combination a, + a, the parameter e. When the solutions of the they can be replaced by a single arbitrary. periodicity equations are substituted in (13), the desired continuation of
x
is
obtained.
)+(**+
310
PERIODIC ORBITS.
second equation of (14) can be suppressed because of the integral. The required terms of the series for the x are found from equations (11), and the explicit forms of the periodicity conditions are
t
(a 1
+a )5[+2m 7rv^T+
l
]
t
i
i
i t
+ (a +a )?^[8
l
1
li
(a 1
=0,
(a,+a,)5[+2w 2TrV^T+
]
(17)
4
+ (a +a )^[^(a +a )(a
s
+a,)+033(a3 +a,)(a4 +a )]
]
=0,
(a 4
+a
4)
[-2ra2 7rV^T+
42
+ (a +a )^^[0
4 4
(a 1
+a )(a2 +a )+^(a,+a3)(a +a )] +
2
1
=0,
where the
The
are functions of m which will not be given explicitly. first equation of (17) is solved for 8 and the result substituted in
d tJ
by
2
,
+[
=0,
(18)
+a )[^,
4
(a 1 +a,)(as
+a )+^
2
/x.
22
]=0,
Au
are functions of
exist, it is
necessary that
ls
aa [A n a 1 at
+A
at ai ]
= 0,
al [A 2l a l ai -\-A 32 a3 al ] = 0.
(19)
satisfied
by a3 = a 4 = 0.
When
imposed, equations (18) can be solved uniquely for a and a, as power series The generating solution is reduced to that conin e, vanishing with e.
sidered in
cases which
158,
possible to supply the proof for the exceptional were not covered by the previous discussion.
it is
and
now
after
1
when
cr^o-,
When m = m =l
By
putting
this deficiency.
after
m, = m and
m =
i
1,
we have
the
The corresponding
can be treated by
cases arising from the second generating solution so combining the periodicity equations that the equations
and a 2
respectively.
The
discussion
is
then the
same
In order that equations (19) may be satisfied by values of the a different from zero, it is necessary that the determinant of the A u should vanish.
This determinant can be developed as a power series in Vju. If it is identiThe coefficient cally zero in ju, each coefficient of this series must vanish.
311
to be different
it
from
zero.
may be possible to make this determinant vanish, but because of its complicated character this possibility has not been considered. The question of the existence of these orbits is thus left open, but it seems improbable that the necessary conditions can be satisfied.
values of fi under consideration here
Fifth Generating Solution. The values of ft in this case are such that the period of the generating solution is 2rmr = 2m 1 7r/o'1 The last
161.
.
The
equation of (14)
is
suppressed.
form
(a 1
+a )5[+2m 7rV^=T+
1
]
i
3
+ e[a u (a + o )a
1
I
+a n (a +a )a +a n (a +a )a +a u (a +a )a +a
1 1
ls
al+a
1
li
alj
2 i)
+t*[a il (a
+a
(a t
24 (a2
+a +a
3
;!
!!6
(a 1
]h
=0,
+a )d[-2m Tr\f=T +
t
1
']+e[bn (a1 +a
)a 3
+ b a (a +a )a +b u (a
l l
-r-e
[6il (a 1
24 (a 2
+ a,) +6
3 3
1B (a 1
+a
2
1
(a
2
!!
+a )+6
2
23 (a 1
+a )(a
1
-|-a 2 )
+6
it
<
+a2 +&
)
26 (a 1
]+
2)
=0,
[{ fi~*"=*-l)+
]+e[c u (a +a0 +c
1
12
(a 1
+a
)(
+a
+c
a4
22
(a2
+a +c
2
2)
23
(c+7)
+a
(-
}+,<
I 1
>]+
1)
-0,
[(e-*<w^-l)+
.]+ e [d u (a
2
+ a )*+d M (a + a )(a,+o
1
+d
[2wirH
]
2i (a i
+a Y+d (c+yy+a
i3
3 2 4
3
3(
)+*<
1
1
)]+
)*
=0,
a + e[e,a + e a +5(e H )]+[e,i(a + +f (a + a +e 3(a +a )(a +a )+e (c+7) + a (---) + a (---) + 5(---)] + -'-=0,
s 2
22
2)
!!
24
are functions of n which are readily determined. The last three equations are solved for a3 a 4 and 5, and the results thus obtained are substituted in the first and second equations. After dividing
where the a tJ
eu
by
2
,
4(a
+a CCa. + a.)
1
(a 2 (a 2
+*( +<(
= 0,
2)
)=0.
312
PERIODIC ORBITS.
exist, it is
= 0,
a2 [Ca1as
+Dc = 0.
i
(22)
With a and a 2 having this value, But the generating equations (21) then have a unique solution for a^ and o2 in The orbit obtained is, 159. solution has reduced to that considered
satisfied
x
.
by e^ = a2 = 0.
therefore,
after
the continuation of the third generating solution re-entering = 1 revolutions, and moreover the value of n is such that ma^
where in, is an integer. Thus we have a proof of the existence of an orbit in one of the exceptional cases omitted in discussing the third solution. In order that equations (22) may have a solution for which (^,0^, and c are different from zero, it is necessary that the determinant of the A, B, This determinant can be developed as a power series C, and D shall vanish. If it is identically zero, each coefficient in this development sepain Vyt. The coefficient of Vju was computed and found to be rately must vanish. For special values of /x it may be possible to satisfy different from zero. Because of the (22) by values of a lt Oj, and c which are distinct from zero. this of the coefficients, possibility has not been complicated character
established.
As
lacking. discussion for the sixth generating solution differs only in notation from that just given. No new orbits are found, but a proof is obtained of
The
the existence of the continuation of the third generating solution re-entering This is another excepafter revolutions, when n is such that ma2 m 2
tional case not treated in the discussion of the third generating solution.
The values of /x for this case 162. The Seventh Generating Solution. = = 2m As in the previous is 2mir 2m the such that are 2 ir/<r2 Tr/<r period The remaining periodicity case, the last equation of (14) is suppressed. form: have the following equations
1 1
.
+ a )5[+2/W 7rV^T+ -]+ 2wMr [pn(<h + *i) (fl*+t) + Vu (a + a )(a +a )(a +a + ^ 3(a + a )(c+7) y+ (a + a )a[-2ra irv^T+ ]+2mir[<p (a + a )(a + a +V? (o +a )(a + a )(a -|-a )-|-v s(a +a )(c-|-7) ]e + (a,+a )5[+2m 7r\^rr+ 0+ 2 * [?n(i+i)(i+ (* + + (az+a Y(at+a )+<P (az+* )(c+yyy+
,
(a l
4)
=0,
2l
2 )-
22
=0,
)
i)
'
<P3i
3i
=0,
(23)
(a 4
+o )5[-2w
4
7rv/ "=T+
+V(a3 +
2mir5
a3 )(a4
2
-|-a 4 )
.
.
4)
=0,
+ mir
[<pa (a,
+ Oj) (a
+<p*(c+yyy+
where the
<p tJ
=o,.
are functions of m-
313
The
last
tuted in the
first four.
equation of (23) is solved for 5, and the result obtained substiAfter dividing by e2 the equations have the form
,
(a l (a 2
12
(a 3
]
] ] ]
=0,
+ai )[^P
il
(a 1
)+^P23 (c+yy]+e[^
2
= 0, = 0,
=0,
(24)
]+ e[-
(a 4 +a 4 )[^ 41 (a 1
+a )(a +a )+^
1
42
(a3
where the
ty iS
are functions of
ju-
of the required
form
shall exist,
it
In order that solutions of these equations is necessary that a3 [yp3l a a2 -\-^3i a3 ai -\-^33
l 1
2
}
= 0,
i
]
= 0,
c*]
= 0,
t
= 0.
These equations are satisfied by a = a2 = a3 = a 4 = equations (24) can be solved in the required form for alt
.
With
<%, a 3)
these values
and a
The
orbit obtained
is
re-enters
after
the continuation of the third generating solution, and revolutions. Moreover, n is such that ma1 = 1 and
m<r2 =
and m, are integers. This is the only remaining 2 x exceptional case not considered in discussing the third generating solution. It has now been shown that the continuation of the third generating solution
, ,
where m,
revolutions exists for all values of n. re-entering after In order to obtain the continuation of the seventh generating solution, it must be possible to satisfy (25) by values of the a t and c which are different
from
eliminating these quantities, two functions of the obtained which must vanish if the non-vanishing solutions exist.
zero.
On
\pu
are
These
functions can be developed as power series in VjL If they are identically zero each coefficient must separately vanish. The coefficient of V/Z was for one of the and found to be different from zero. computed developments
equations (25) can not in general be satisfied in the of p this may be possible, but on account required way. of the complicated character of the \pt) the possibility has not been proved.
It follows, then, that
it
In constructing the has been found convenient to use the normal variables
in the discussion of the existence.
The
differential
equations are the first four of (11), and the solutions are given by (13), when the quantity y has been put equal to zero. It has been shown that the quantities a2 a3 a 4 and 5 can be determined as power series in e,
, ,
314
PERIODIC ORBITS.
vanishing with e, so that the corresponding solution shall be periodic, while the quantity e^ still remains arbitrary. When these series are substituted in (13), the expressions obtained for x1} xt x3 xi can be rearranged
, ,
as
series in
which converge
e.
We
x1
Xjo -(- Xn 6 ~r x l }
fi'
* *
*j
*
j
x3
3l)
-T x 3l
+ X&
-p
~\
xt
t %a
2
(20)
= 5,6+
5, e
series (26)
(a)
(b)
They satisfy the differential equations identically in e. Each xi} is periodic with the period 2^/0-! for one set of
with the period 2ir/a2 for the other set. We have supposed that x' = at t = 0, and therefore
|
orbits,
and
(c)
it
follows that
<T l
(x 1
xi )-\-<Ti (xs
xi )=0 at t = 0.
(d)
The
arbitrary aa occurs always with the arbitrary a, in the combia, ; there will be no loss of generality if a t is specialized. It will be supposed, then, that a t is taken so that at r =
nation a t
^1+^2+^3+^4 = Gi+a, = a.
The
(27)
=A
(X e+---)+B (Y e+---).
2 i 2
The
substituted in these equations, and the coefficients The resulting equations are of the corresponding powers of e are equated. solved for the xtJ and the periodicity conditions are imposed. The construction will be made first for the orbits with the period 2-iv/a^
series (26) are
,
.
now
of
are given
by
the equations
x[
o-2
'i^
Tx = 0,
10
a/3Tx = 0,
30
315
The
is
^lO
w10
^70
""20
^O
u30
''40
u 40 c
applying condition (6), it is found that a30 = a40 = 0, and from (c) and that a 10 = a 20 = a/2. The solution satisfying the conditions is, then,
On
(d)
t'lo
~
2
'
20
'
30
'
40
'
= acos
ffl r,
2/
=-
^~Q lq
of
e
1
coser.T-
^fg sin^r.
(28)
The
power
are given
by the equations
1
l ,
+A X?+B Yf = -a.V^T S iXm +A X*>+B Ff, a4+*i V^T 4 - ^/"="T z = + r, V^T x +^3 Xf + Ff, a4+.r,V=Ta- -WV^I 5x^0+^, Xf +B Yf
x'n-^V^Tx^+a.V^T 8
.t21
x 10
31
5l
30
and F2 represent the expressions obtained by substituting x In order that the solution of the first for x and / in X2 and F2 the coefficient of e' lV=iT in the right member equation shall be periodic, ~ must vanish. Otherwise non-periodic terms of the type Te aiV lT will be ~ lT v~1T and e"' For the same reason the coefficients of e~ aiV introduced. e -nv=ir m ^g r ight members of the second, third, and fourth equations respectively must vanish. All the terms of this type come from the first
where and y
0)
.
in
terms of the right members, since the other terms are of the second degree Since x30 = xi0 = 0, these conditions are satisfied in x w x20 x30 and x m the third and fourth equations. Since we have at our disposal the undetermined quantity d the desired result is obtained in the first and second
.
= 0. now
(6)
(c)
and
(d) are
imposed. The details of this work will not be given. in the original variables are
#i
The
results expressed
12
+ b'
i2
sin 2a\r,
(29)
*,-0,
316
PERIODIC ORBITS.
where
"10 _ a,!
_ ~
^+8^8(1-2,0*,
27u(l
u)
un
,
\
u io
} "is/
>
wn =
.
_ ^ u u>
,
1I +3VS(l-2M)g,i+16<r 1 (l6<il+9)ii ~
gM
12rf(5o?-l)
ai2
,
12<r
,
(5<rf-l)
.
"~
a(,
/ 0,1
"
,
-3>/3(l-2M)an
4cr
+9
ii
l
l
h12
# fc
12<r (5cr
_,_
-l)
2
16<r 1
12
"*"
12a (5<r*-l)
1
^,o=
12
^; 2 =
10
2)
&2
]
2
^' 2
=+
)]
It will
is
now be shown
that this
method
. .
= 0, n 1) have 4; j (26) Suppose xv and 8j (i=l, general. been determined and that the xtJ are periodic. For the determination of xin and b we have equations of the following form:
,
. .
/=o
+i
/=o
_
(30)
[0
x^-^V^T x
= +<r V^T
2
5 n x30
2 + J=0
"+1
1+1
****+*, e-*'v=ir]
1
xL+^x^J x = -a.V^i
in
8 n x i0
[6t ,e
__ " vlr
+r,ti e-
_lT '^]
,
where the 6 (J and tj<> are known constants. Since x30 = xto = 0, no nonIn order periodic terms can enter the solutions of the last two equations. that the solutions of the first two equations shall be periodic, it is necessary ~ v~ 1T and e~" lV 1T in the first and second that the coefficients of e"' respectively
317
This gives for the determination of 5, the only undetermined = 0. Since two 1 a5+20 u = O, 2rj 21 constant, equations c^V a^V 1 a5 the existence proof has shown that 5 is uniquely determined, it follows that these equations must give the same determination of 5 An additional proof is obtained by means of the integral (5'). The terms of this integral which are independent of e are first expressed in the normal variables. Then the variables are replaced by their expressions as power series in e and the terms are rearranged so that the integral remains a power series in Since the integral is an identity in r and e, it follows that the coefficient of each power of e must reduce to a constant identically
. .
in
r.
We
e".
When
xl} as functions of r are substituted, this coefficient consists of a sum of The coefficients of each of these linearly independent functions of r. functions must then vanish. Let <py and ft denote the coefficients of e"lV=iT and e~ ffW^ T in the first and second equations of (30) respectively. Then, on integrating these equations, the terms in x ln and x 2n carrying ft and <p2 are found to be
The terms
in the coefficient of
\_(X 10
e"
in the integral
-X 2n ) + (M M) (Mi. - Ms..)] - 6(a; +a; o) (x ln +x2n ) - 18(Mio+M2<>) (b x ln +b x2n ) - 6V3(1 - 2m) +z ) (Mi+M2) + (x ln +x2n (b x +b
-8o*
-X
20 )
(x ln
10
(31)
2
[Or 10
20
10
xj~]
When
ia v~ lT and r, Te
'
the expressions for x10 xt0 x ln and x2n are substituted, terms carrying All other terms entering this T e~ iaW=iT are obtained.
, ,
,
coefficient contain
only xtJ
(i
consequently periodic. Since the coefficients must vanish, periodic terms are obtained from (31). The coefficients of relations are obtained which ft and v2 must satisfy. 2<r.v^iT -taw-iT Te anci Te van i s h identically. The coefficient of r gives the = 0. Since <r\>\, the coefficient relation 32o-j(4o-' 1) (ft ft)/(4o-J 9)
Hence
1),
and are
of
Hence if 5 is determined so that either of The determination of d is therefore unique. that the other must vanish also. Equations (30) are now integrated. By means of conditions (6), (c), and (d), the new arbitrary constants are uniquely determined in terms of the The results when expressed in the original variables original arbitrary a
.
xn =
> cosjr+a nJ
sin jr]
2fl u
yn
=^[b
21721
n)
(32)
318
PERIODIC ORBITS.
From the character of the differential equations it is readily shown that x n and y n carry the factor an+1 that & carries the factor a", and that a enters in no other way. Recalling the transformation by which t was introduced, we have the final series
,
x=x
the Xj y s the series,
,
e+x '+
1
,
y=y
e+y
f?+
= a e+5
i
(33)
being given by (32). From the way in which a enters it is seen that the arbitraries a and e occur always in the combiTherefore we can put a = 1 without loss of generality, and the nation ae final series contain the single arbitrary e.
and
dj
Y
^
\^
^--_--
319
2-k.
, ,
The
,
discussion of
the existence has shown that the quantities a1} a2 a3 a4 and 5 can be determined as power series in e, vanishing with e, so that x t x2 x3 x A and z
, , ,
,
will
the power series obtained in this way are substituted in (13), we have, after re-arrangement, xx x2 x3 x i and z expressed as power series in e which converge for e sufficiently small. By the
,
,
When
use of equations (10) x and y can be expressed in the same way. the solution has the form
,
Therefore
x=x
+x e+x e +
i
1
V = Vo+Vi*+y^+
(34)
The
(a)
(6)
(c)
series (34)
They Each
X},yj,
and
z, is
z(0)=0;
therefore z,(0)=0
(j=0,
1, 2,
oo).
J(0)=0
(j=l,
2,
oo).
The
property follows from the fact that the arbitrary y occurs always with the arbitrary c in the form c+y, and can be put equal to zero without
last
loss of generality.
The
(l
3e
(36)
z"
+ (l + 5) z=(l + 5)
2
[Z2
e+Z
eM----}
be substituted in these equations and the coefficients The x)t yt and z t are determined by solving the of the powers of e equated. obtained and thus imposing on the results the conditions (35). equations The terms independent of e are given by the equations
The
4-2y' y':+2x'
f
(1
V3 (1-2 M )
ifc-0,
(37)
J V3
-2 M
x,
f y,-0,
The
is
T
-a
01
e<"
*=* T
0l
02
e~" *=* T
<T
-<72 \/
1
0l
0l
02
(38)
=c
'
01
320
PERIODIC ORBITS.
condition
(6),
By
we must have
Ooi
= Oqi =
a,*.
= <4 = K = Ki = K = b' - 0,
and conditions
(c)
and
c01
= u,
c01
c.
The
is,
then,
z
= 0,
of
e
=c
sin t.
(39)
The
power
are given
by the equations
8
3
/5M_9..W_
= 3V3V n
__ c9 ^
(40)
321
, ,
and
it
For the general terms we proceed by induction. Suppose that xs y, zs = 0, 1, n Sj(j 1) have been determined, and that for j even has been found that
. .
to
*#-0,
while for j odd,
Vj
= 0,
z]
='[c
coB(2k+l)T+c*8\n(2k+l)T];
it
X]= 2)
t=i
a+ti/s
2/>=
2)
*=i
[&*
cos2A;r
+ &t
sin2A;r],
z,
= 0,
1,-0.
is
It
even the
coefficients of
n e
are
given by
=-2^
sinr
+ 2[^+1 cos(2i+l)T+C^
,
sin(2.;-+l)r].
1=0
In order that the last equation shall have a periodic solution the conditions
we must impose
-2c8 n +C'
l
(n)
= 0,
to exist
CT-0.
5
.
The
first
Since
by the
existence
known
it
C?
is
zero.
An
Cf
is
zero
The series for x, y, z integral (5')as a power series in e. Each coefficient of this series re-arranged
to a constant identically in
of this coefficient
t.
must reduce
e".
The terms
which carry
are
2z'J+2z,zn
xn y Suppose xa been determined and that the
. . ,
(45)
.
yn ys
Sl
5_i have
z,
are periodic.
The equation
z'+z^ysinr+C
On =
\t\t
co&T+
cost
| Cf
t sinr.
322
PERIODIC ORBITS.
When
the expressions for x J} y,, and z, as functions of r are substituted in the coefficients of e" in the integral, the only non-periodic terms obtained come from zn Since the They are of the form t, t sin2r, and r cos2r.
.
a constant identically in t, it follows that the coefficients Those for r sin2r and r cos2t vanish of these non-periodic terms are zero.
coefficient of
t
is
identically.
The
C = 0.
(35),
Since Ct*Q,
find
it
follows that
CT = 0.
and imposing conditions
Upon
integrating (44)
we
zn
,^gs
The quantities Cj"Vi and CJJJ, are known from the differential The constants of integration cnl and c' nl are found by (c) and (d)
have the values
c
equations. of (35) to
CSV,
,'<>_
y (2,/+i)0
(B>
When n
is
coefficients of
e"
are
=
fl
2
.7=0
[4
cos2ir
+ 4'
<
>
sin2ir]
Cn+l)/2
l/:+2*:-fV3(l-2/i)x.-fy.=
w
[Brcos2ir+5;
sin2ir]
,
(47)
+* " -25
From
the last equation
it
n csinr.
= 0,
for otherwise z
will
not be periodic.
equations,
we
find
1
"["
-Sin^r 16/(4/-l)+27 M (l- M ) - (!6J 2 +9) Af + 3 V3 (1 - 2ju) Bf + 16,/g'"" 9 "I COS^rj, l6/(4/-l)+27 M (l- M )
.
(48)
Zf
I6j(4/-1)+27 M (1- M )
2
-sm2jr
9
.
3 V3 (1 - 2 M A? ~ 16j AT - (16 j + 3) Bf T i6
)
"I
/(4/-l)+27 M (l- M ) z n = 0, 5 = 0.
COS^rJ,
323
e
we make use of the transformation by which introduced, we have for the final series
If
the parameter
was
+ Xie + x + y = y e+y +y e +
e
2 2e
x=x
=Z
e+Zie
+Z e
2
+
(49)
le
the Xj, y u Zj, and Sj being given by (46) and (48). It is not difficult to show that x, yn and z carry the factor c" +1 and 5 the factor c", and that c enters these expressions in no other way. Consequently in (49) the arbi, ,
trages
c
=\
c and e occur always in the combination ce Therefore we may put without loss of generality. The final series then contain only the
.
arbitrary
e.
approximate idea of the shape of the orbit can be obtained by considering the first two terms of (49) These terms = were computed for /* 0.01 and e = 0.5. The projection on the x?/-plane is an ellipse of small eccentricity whose center is on the
.
An
negative y-axis and whose major axis cuts This projection is the positive z-axis.
The projections on the shown in Fig. 6 and The orbit thus conrespectively.
are
two elongated
loops,
one above
Fk;.
5.
Fig. 6.
Fig.
7.
xi/-plane, the
replaced
by t+7t
and
zu
then x s
and
y,
seen that
z}
changes sign.
It follows
that the loops are symmetrical with respect to the zy-plane, and that each loop is described in half the period. For positive values of e the upper loop is described first, and the motion is such that the projection of the infinitesimal
of the
body on the
is
xi/-plane
moves
The period
motion
T = 27r(l+5).
324
PERIODIC ORBITS.
Orbits about the Point II.
To each of the orbits about point I there orbit about an point II. The proofs of the existence of these corresponds The orbits were omitted, since they are similar to those for the first point. the series from be obtained orbits can these for series easily corresponding The differential equations for the orbits about for the first set of orbits. point II are obtained from those for the orbits about point I by changing the sign of V3- The periodicity conditions to be imposed are the same It follows, therefore, that the solutions for the one case in both cases. can be obtained from those for the other by changing the sign of y/%. Therefore, in order to get the series for the orbits about the second point we make
change in the series already obtained for the first point. On referring to equations (1) and (2), it is seen that the differential equations for point Hence I reduce to those for point II if the signs of y and r are changed. this transformation can be made geometrically by a reflection in the xz-plane
this
and a reversal
Chapter X.
ISOSCELES-TRIANGLE SOLUTIONS OF THE
PROBLEM OF THREE
By Daniel Buchanan.
BODIES.
This chapter treats of periodic solutions of the which two of the masses are finite and equal. problem The third body is started at the initial time t from the center of gravity of the equal masses, and the initial conditions are so chosen that it moves in a straight line and remains equidistant from the other bodies. In I the third body is assumed to be infinitesimal and the initial conditions are so chosen that the equal bodies move in a circle about the center of mass.* In II the third body is considered infinitesimal and the initial conditions are so chosen that the equal bodies move in ellipses with the center of mass In III the third body is considered finite and the as the common focus. solutions derived have the same period as those obtained in I, and reduce
165. Introduction.
of three bodies, in
to those solutions
when the
third
body becomes
infinitesimal.
I.
PERIODIC ORBITS WHEN THE FINITE BODIES MOVE IN A CIRCLE AND THE THIRD BODY IS INFINITESIMAL.
The
Differential Equation of Motion. Let and ffi, be two l and n an infinitesimal body. Let the unit of mass
166.
be so chosen that m 1 = m 2 = 1/2; the linear unit so that the distance between m, and m2 shall be unity; and the unit of time so that the Gaussian conLet the origin of coordinates be taken at the center stant shall be unity. of mass, and the ^-plane as the plane of motion of the finite bodies. Let the coordinates of w,, r 2 and m be & ij u 0; 2 %, 0; and 0, 0, f respectively.
,
If
and
and
1/2, 0, 0, respectively,
so that
they move
Si=
then
-It" |cos(*-0,
).
*When the finite bodies move in a circle, the motion of the infinitesimal body can be completely determined by means of elliptic integrals. The problem was first solved in this way by Pavanini in a memoir, "Sopra una Nuova Categoria di Soluzioni Periodiche nel Problema dei Tre Corpi," Annali di Malematica, The elliptic integrals obtained by Pavanini were later developed Series III, vol. XIII (1907), pp. 179-202. independently by MacMillan in an article, "An Integrablc Case in the Restricted Problem of Three Bodies," Astronomical Journal, Nos. 625-626 (1911). MacMillan further developed the solution as a power series The solution obtained in I has a in a parameter, the coefficients of which are periodic functions of t.
close relation to
MacMillan's solution.
325
326
PERIODIC ORBITS.
The
differential
is
i+4r )
2 s/2
w
t.
The integral of
(1) is
where
to
C is the constant of integration. If C is positive, the particle n recedes does not pass beyond a finite If C is negative, the particle infinity.
/x
is
consideration of (2) it can be shown that, origin. C is negative, the particle crosses the ij-plane. Hence the initial time can be chosen, without loss of generality, as the time when the particle It can also be shown from (2) that if C is negative, in the ^rj-plane.
From a
the motion of the particle is periodic, and that the period can be expressed as a power series in the initial velocity of n, whose limit is 27rA/8 as the
initial velocity
approaches zero. We shall, however, prove the existence of a periodic solution of (1) by a different method.
167. Proof of Existence of
this proof
it is
(I).
In
*-<
where
8 is
= VV8(1+8)t,
(3)
to be determined so that the solution of (1) shall be periodic in t with the period 2t. At t = let
f
= 0,
= a,
(4)
where
= d{/dr.
r~a*j
then when
(3)
(5)
(1),
and
(5) are
substituted in equation
we obtain
(1+S)g
(l+4aW*
where
the second derivative of z with respect to for z become, as a consequence of (4) and (5),
z is
t.
W
/
x
fi
The
initial
conditions
Z (0)
=0,
i(0)
= l.
of (6)
(7)
For
\a\
member
the series
2=-(l + 5)z{l+(~
where
3/2
1
)4aV+
+(
' ' '
_ 3/2
t
)(4aV)*+
(8)
}>
/-3/ 2 \ m
\
i
\2)\~2j
1
2~)
i
ISOSCELES-TRIANGLE SOLUTIONS.
can be integrated as a power series in a2 and for 0^t^27t, provided |o| and \b\ are sufficiently small.
327
6
Equation
(8)
= 2 S
Zu
S a".
(9)
The
initial
values of the
z it , as
determined from
(i,j
(7) are
oo),
1
m(0)-0,
4,(0) -1,
= 0, ...
KM=o
(i+j>o).
substituting (9) in (8) and equating the coefficients of the various 2 powers of 5 and a we obtain differential equations from which the ztJ can
Upon
initial
The
term independent
of 5
and a2
is
of
it
which
satisfies (10) is
The
term in
alone
is
*u+H~ -sinr,
and the solution
of it
which
satisfies (10) is
Zi,o=--|sinT
+ JcosT.
The
z
is
therefore
= sirjT + 5
[ttCOSt
-sinr]
+ terms of higher
2 degree in a and
5.
(11)
With the initial conditions (7), the variable z is an odd function of t, and therefore a sufficient condition that it shall be periodic with the period 2x in t is z(t) =0. With the value of z given in (11), this condition becomes
in a
and
5.
(12)
from zero, this equation can be solved a2 vanishing with a2 Let us denote
,
.
5= 2 Lo".
(13)
When
we
2J
obtain
z=2z
/=0
ail
(14)
328
PERIODIC ORBITS.
which converges for \a\ sufficiently small. Since the periodicity condition has been satisfied, z is periodic in r with the period 2ir. Hence
Zll
{T+2Tr)=Ztl {r)
= 0,
oo).
(15)
The
initial
%/0)-0 = l, 2o(0)
2,,(0)=0
0=1, ...oo).
Equation
(1).
Let
us substitute (13) (14) in (8) and equate the coefficients of the various 2 powers of a". Since the result is an identity in a , there is obtained a series
and
of differential equations
can
be determined.
The
term independent of a2
is
+z = 0,
and
(16)
is
of
it
which
satisfies (15)
z
= sinr.
term
2
The
in
a2
is
z'j+z2
=-5
+6z2=(-5 +|)sinT-|sin3T.
gives rise to a non-periodic term in the solution, and, in order that (15) shall be satisfied, its coefficient must be zero. Hence
Oo 2
_9 =
2
is
'
(sin 3r
3 sin t)
The
differential
a* is
3+24=-(S4+^)sinT+6sin3T-gsin5T.
In order that (15) shall be satisfied,
8*
54
=~
32'
for z4
zt
is
found to be
^[431sinr-192sin3T+29sin5r],
(16).
ISOSCELES-TRIANGLE SOLUTIONS.
329
So far as computed, it has been found that the 5 2y are uniquely determined by the periodicity and the initial conditions, and that each z2j is a sum of sines of odd multiples of r, the highest multiple being 2./+1. We shall now show by an induction to the general term that all the 82J are uniquely determined by the same conditions, and that all the zv have the properties which have been stated. Let us assume that 52 5 2i _ 2 z zv - 2 = have been uniquely determined, and that each zn (k 0, j\) is a sum of sines of odd multiples of r, the highest multiple being 2A;+1. From these assumptions and the differential equations it will be shown that 5 2] and z2J are uniquely determined, and that 2^ is a sum of sines of odd
, .
. .
l. multiples of r, the highest multiple being 2j Let us consider the term in a". The differential equation
is
Z2>
+Z =
2J
hlZo
.
+ Zi))
.
.
(17)
1)
where
2j
is
known
in
,j
and
8 it (k
= 1,
1).
The
general term
Z2J
T
. . .
=z"'
z"* h"
k
p,
(or zero)
(a)
(6)
(c)
Mi+
+Mt
or
,
is
an odd
integer,
Hi\+
q
is
1.
. .
+/x*X t +Mi+
z2 _,_ 2 is
is
+Ptl+qp = 2j,
sum
of sines of
Since each z
Zti
sum
of sines of
highest multiple
tf
is
= Mi(Vfl) +
is
obtained
(\ k
2,7
+ 1.
(18)
+a ^
2
sin(2j+l)r,
it
where a?",
ati
are
known
constants.
From
(15)
follows that
o 2y
-ai
is
The
therefore
Asm +Asm3T+
T
+AZ l
sm(2j+l)r,
where
A? - tScm>+ a&.
,a/>
330
PERIODIC ORBITS.
periodic solution of (1) in terms of the variable r
1
Hence the
is
1
+
It is a
(19)
a with sums of sines of odd multiples The highest multiple of t in the coefficient of a2t+1 of t in the coefficients. In the sequel we shall call such a series a triply odd power series. is 2k +1.
power
series in
odd powers
of
The
period in t
is 2tt,
and
r-.
in
,
it is
.
2tt
1414,35
6_i_
II.
SYMMETRICAL PERIODIC ORBITS WHEN THE FINITE BODIES MOVE IN ELLIPSES AND THE THIRD BODY IS INFINITESIMAL.
169.
The
Let m, and m^ represent the two Differential Equation. infinitesimal the fx body. Let the system of coordinates be
chosen as in 166. Let the unit of mass be so chosen that m 1 = m i = l/2, and then let the linear and time units be so determined that the mean distance from 1 to 2 and the gravitational constant are each unity. With these units the mean angular motion of the bodies also is unity.
Let n be started from the center of gravity of rax and m2 perpendicularly to the plane of their motion when they are at apsides of their orbits, which can be assumed to lie on the -axis. From the symmetry of the motion with these initial conditions, it follows that
i
%=
^i
Let the motion of the finite bodies be referred to a system of axes The coordinates rotating about the f-axis with the uniform velocity unity.
referred to the rotating axes are defined
by
z
X|=^cos<-|-r7< sin^
yt =
^sinZ+^cosJ,
=f
({=1,
2).
and y are determined by the conditions that mx and m t shall move in ellipses and be at apsides at t = t which in this case is put equal to zero. Then it follows, from the properties of elliptic motion, that
The
Xi
=
r v
x2 = rcos(t>
t),
y2 = rsm(v
t),
(20)
where
=m[l ecosf+y (1
2
cos2<)
,
]>
m=m =m =
i
>
= eccentricity
of ellipses.
ISOSCELES-TRIANGLE SOLUTIONS.
331
The
motion of
/x
is
2=
where
r1
~T"T = "^T'
2 2
.
mz
mz
2mz
/01
(21)
When we
(20),
r
2
[m {l-2ecos
Where
+ |-(3-cos20+
make
2T
2mz
,
+z ]
tU"
(22)
parameter while
we shall substitute its value 1/2, the substitution ra = ra +X, and conboth remain fixed. In order and
to obtain the solution of the physical problem the final results. With these substitutions, (22)
we must put
becomes
=m
in
2"
= -(m
+X)2#2,+1 2
J=0
w+1
,
(23 )
where
*
;
a power series in e with cosines of integral multiples of t in the coefficients, the highest multiple being the same as the exponent of the eccentricity e.
is
Determination of the Period by a Necessary Condition for a If the motion is periodic, let the period be denoted Periodic Solution of (23). by T. Since the period of motion of the finite bodies is 2t, we must have
170.
T = 2vtt,
where
finite
v is
(24)
=0,
it
z'(0)
= a.
z is
(25)
With
an odd
function of
started from the ^-plane when l and m^ are and sufficient condition that z shall be at apsides of their orbits, a necessary periodic with the period T is
Hence
if fi is
2(772) =0.
(26)
332
PERIODIC ORBITS.
In order to determine the period T from the condition (26), we integrate equation (23) as a power series in a and X, but only in so far as the The differential equation for term of the first degree in o is concerned.
this
term
is
< +m
lZl
= 0.
(27)
This equation belongs to the class of differential equations with periodic coefficients which was treated in Chapter III, where it was found that the is an integer. character of the solutions depends upon whether or not 4
Vm
Since
|X|
is separated into ra +X, and since depends upon the way in which must be taken small in order that certain solutions appearing in the
sequel shall be convergent, the value of ra is in the vicinity of 1/2. may therefore regard 4vw^ as not an integer, and when it is not an integer
We
is
= AT e'^^'u.+Af e-^ u
1
(28)
m,
= 1+ 2J 25 n=l *=i
oo
+ x/ ~ T a
B)
sin
^+C
B)
cos
^-
1)1 e ">
n
i i
(
cos
^-
e "'
.3Vmi(l
I6m
.
,
and Af are constants of integration; the af and b are real constants which depend upon the coefficients of the various powers of e in E and a is a power series in e with real constant coefficients, determined by the condition that u and u^ shall be periodic with the period 2r. From (25) we have
1 ;
The A
Af+Af^O,
Since wj(0)
(29)
is
WV^T+u[(0)]Af
+ [-<rV^T+u' (0)]Ar = a.
2
0)
(29)
wi(0), the
^4{
and
in
(30)
where A, is real, and it is different from zero because A is the determinant of a fundamental set of solutions at t = 0. The solutions of (29) for A and
A?
are
Am =
Since
A =
A,
it is
vanishes with
(23) as a
power
series in
convenient to integrate
ISOSCELES-TRIANGLE SOLUTIONS.
333
series in
t),
The form
power
)
is
= A?
[e
X,e;
where
is
a power series in
Af,
X,
and
e.
the condition
O^r^Dly^ - e-"
This equation
solution
is
is
^]+A?P(Ar
this value of
X, e;
f-)-
(31)
satisfied
by
Af = 0, but
Af
z=0. In order, then, that (31) shall have a solution for must be zero. Now different from zero, the coefficient of A
^4.{
which
uA
according as
sufficiently
v in (24) is
= 1,
or
1+a power
series in
e,
Hence u {T/2)t
x
for |e|
"^T- e
i=0.
(32)
In order that this condition for the existence of a periodic solution satisfied, T must have the value
may be
T=
where
2Nir
>
is
of oscillations
made by
(33)
the infinitesimal
body
in the period T.
Then
it
N = v*.
If
a rational fraction, v can be so chosen that iV will be an integer. Inasmuch as a is a continuous function of e, it can be made a rational fraction by a proper choice of e less than any value of \e\ which will insure the a
is
convergence of the power series. The numerical values of obtained when a has been determined as a rational fraction.
N and v can be
171. Existence of Symmetrical Periodic Orbits. Let us consider the terms in (23) of higher degree in Af and X. The differential equation
m,
- e~
*=* u ) 2
(34)
viz.,
The complementary
function of (34)
is
ha = of
^ aV=
'
1'
i+i
e- av=lt
u2
On
we obtain
>)'
^ =Z
i
'
334
PERIODIC ORBITS.
of the coefficients of (a{ 1 the same as (30), viz., A =
l)
The determinant
18,
)'
and
(c4 )'
and
is
A^O.
(
is a constant, Therefore
l)
by
(a|T .
(a l)y
V=lr"m
V^Ie+^Z, Ai
We
. _ V=T*A?B,
_ e -,.^,
^
(36)
.
=_
=_
V=l\ATB
Ai
(MiW2
_ e+s ^, MD
The
integration of (36) gives non-periodic terms as well as periodic terms shall be concerned only with the non-periodic having the period T.
V 1 E u ujb>. be denoted by P,; it is a Let the constant part of power series in e with constant coefficients which are purely imaginary, the absolute term of which is found to be 2 V l/Vm^. Hence
terms.
x x
a{
i
a2
=A +\A? = A?+\A
(
l)
[P Aperiodic terms],
x
where
that
and
A
x
z u (0) = z[,
= 0. (0)
Then
a
x
zu
=XA P
[e
**> u
+e
v^'
u,]
+ periodic terms.
term
(37)
It is necessary to obtain in addition to this only the This term is obtained from the differential equation
in (Af)'.
z'^+m
E zM = Z = - m E z
x
3
.
(38)
On
(36),
we have
(39)
The terms not written in (39) carry the exponentials e* iJaV=It = 1, 2) as a factor multiplied by the fourth power of u and u t considered together. The integration of these terms gives periodic terms with the period T. Let the constant part of 3V lm- E u\ u\/k be denoted by P a power series
t t
x
which
is
which are purely imaginary, the absolute term Then upon integrating found by computation to be 36V lVm
.
(39),
we
obtain
a? = A?
+ (A? Y [P
)
Aperiodic terms],
(40)
where
A
,
and
A
zj,
so that z3
= (0)
z
= 0.
1
Hence
v
t[e''
M = (A'
yP
~u
t
+e-''
v~'u
%]
+ periodic
terms.
(41)
ISOSCELES-TRIANGLE SOLUTIONS.
335
we
(^)+e-
^^u (^)]+
2
(42)
aV=Tr/2 u v=~ lT/2 u2 (T/2)] is different from zero 1 (T/2)+e-" expression [e = for e 0, and therefore remains different from zero for |e| sufficiently small.
The
= 0, and hence the right side carries (42) is satisfied by In order to find a solution of (23) other than z=0, it is necessary as a factor.
Equation
Af
Af
to consider
Af ^
remains a power
zero for
series in
X and
Af can be divided out of (42) There Af and, since P and P are different from
.
of lowest degree are X and C4i ) \e\ sufficiently small, the terms There are no terms in A^e and e alone, and the coefficient of (A^Y has a
.
0)
term independent of e, viz., 36V 1 Vm Hence, after Af is divided out, equation (42) can be solved for A* as a power series in X*, the coefficients being power series in integral powers of e. Two periodic solutions of (23) therefore exist having the period T. They have the form
.
\JQ(X*;*),
X } whose coefficients are power series in e. In the practical construction of the solutions it can be shown that z is a power This fact follows also from the dynamical nature series in odd powers of X
where
is
a power
series in
to the xy -plane.
motion of m is obviously symmetrical with respect The two solutions are therefore of the form
z=z \,
2i+1
1=0
(43)
where each Z2 J+ i is periodic with the period T. In 117-118 it is shown by a discussion, which is applicable in this problem, that if v is even in (24), the orbits obtained by taking the two signs before X } are geometrically the same, but in the one the infinitesimal body is If v is odd, the orbits for half a period ahead of its position in the other.
X 1 are geometrically distinct. By an argument similar to that in 115, it can be shown that so that the solutions (43) will converge for all sible to choose X >
+X
and
it is
pos-
t^ T.
Symmetrical Periodic Solutions of (23). Let us substitute (43) in (23) and equate the coefficients of the various powers The constants of integration occurring at each step are determined of X*. by the conditions that the orbits shall be symmetrical and periodic with
172. Direct Construction of
336
the period T.
PERIODIC ORBITS.
The
is
be symmetrical
condition to be imposed in order that the orbits shall = 0, from which it follows that z(0)
'21+\
(0)=0
(j=0,
ao).
(44)
The
differential
is
+ 171^ = 0,
is
of this equation
z,
(28).
When
(44)
is
imposed
where
'+m E
z3
=Z =
3
-E z -m E z
1 1
(45)
When
expressed in terms of
t,
Z =A?d?HA?)
3
e?,
of degree
(46)
where the
e
63
t+l)
(i=0,
1)
are
homogeneous and
2t+l
in e +irvCrT
'
and
it
-<rv=r<
occurs.
The undetermined constant A"' is written explicitly In ^"(t-O, i) the coefficient of [***=]* [e^*^*
aV=ll h
}
so far as
differs
1, /,
the coefficient of
[e
being positive integers (or zero) such that./1 +,7 2 = 2t+l. These coefficients 1 s\n jt and cos,;7 in the coefficients, k being are power series in e with V If the exponentials in d3 i+l) the highest multiple of t in the coefficient of e*. are expressed in trigonometric form, it is observed that the df i+l) are power series in e in which the coefficient of e" has the form
[e~
aV:=li h
\
from and jt
V=T2
cf
+,,
sin[(2t+l)r+j]<,
"
where the cf t+l) are real constants. Hence the 0f + are purely imaginary. In order that z3 shall be periodic, the constant parts of the coefficients ~ +aV=T and e~ aV= in u2 Z3 and u^Z3 respectively must be zero. From of e the form of ult u^, and Z3 it follows that, when we equate to zero the constant parts of these coefficients, we obtain only the one equation
' 1'
-AA; [P +(A{
1
i)
1,
^] = 0,
in (42).
(47)
where
is
and
are the
power
but
series
which appear
satisfied
is
by A"'
= 0,
The
Equation
different
(47)
and
excluded.
which are
from
zero are
A? = V=i Pl
(48)
ISOSCELES-TRIANGLE SOLUTIONS.
t
337
where p is a power series in e with constant coefficients which are real since and P2 are both purely imaginary and their absolute terms have the same The absolute term of p is found by computation to be 1/Vl8ra sign.
t t
.
When the sign purely imaginary the expression for z1 is real. of 1 is chosen in (48), the periodic solution of (23) which satisfies the initial condition 2(0) =0 is unique.
Since
is
The
is
= A? f^'^+A" e-^'^+V^t^+^f],
(49)
where Af and Af are the constants of integration. The particular integrals From <pf and <pf are respectively of the same form as Of and Of in (46). the form of <pf and <pf it follows that
(44)
on
(49),
we obtain
Af+A? = 0.
The
solution (49) therefore
z3
becomes
)
= A?
a
[e
^<u-e-"^'u + V^i[<pt
2 \
+<pr\,
is
(50)
where
3)
^4{
If
Af
found to be purely
imaginary this solution for z3 is real. From a consideration of the terms in X5/2 and then by an induction to the general term, we shall show that Af and all the remaining constants of integration are uniquely determined, after the choice of the sign in (48) has been made, by the conditions that the orbits shall be symmetrical and periodic with the period T. The differential equation for the term in X6/2 is
,
z':+m
<>
E z = Z =-[E z +Zm E
l
z\z 3
]-{E
zi l
+m E
a
z\].
(51)
In order that 25 shall be periodic the constant parts of the coefficients of e +v=Tt an(j e -<rv=ri Z an(j z must be zero. From the
form
i Ui & respectively follows 6 that, when we equate to zero the constant parts of these coefficients, we obtain only the one equation
Ul
of
u lt u2 and
,
it
AfPf + V^TP = 0,
3
(52)
where
Pf and P
are power series in e with real constant coefficients which 1 in (48) has been chosen. The absolute
of (52)
is
4T*V=Tft,
where
p, is a
(53)
coefficients.
power
series in e
338
determined as and has the form
Zi
PERIODIC ORBITS.
With
A?
= A? e'^'u.+Af e-'^'Ut+V^lW+tpT+cp?],
(54)
2i+1) are the constants of integration and where the = 0, 1, 2) are of the same form as the 0f'+l) (i = 0, 1, 2), respectively. It (i follows from the form of <p, ipf, and <pf that, when the condition (44) is
where
and
= A?[e + ^>u-e-<'^u
+ V=iW?+ p?+<p?\
l
(55)
where
is
Af
is
real
if
Af
purely imaginary.
Let us suppose determined as been that power series Af"~ uniquely Let us also in e with constant coefficients which are purely imaginary. z determined and that they have been 2a _j uniquely suppose that z are of the form
shall
. .
We
now make
3)
.
A,
have
tjfe-l,
n-1),
.
(56)
+1) in (46) t = 0, f the same form as the ^ k. It where the <pf t Xi are B_1) is purely imaginary and is uniquely determined will be shown that Af z that in+1 shall be periodic; also when the condition by the condition has been imposed, that zin+1 has the form ztn+l (0) =
.
=0
are of the same form as the <p+" in (56). where the <p 2 2 The differential equation for Let us consider the term in \ "+ / term is
(2
1)
.
this
z" n+i
2n - 1
+3m E zlz
s
2n _ l
]+
.
(57)
. ,
The
not written explicitly involves z x z 2 _ 3 to odd The undetermined constant which together. only degrees 2" -0 and it has the same coefficient in Z2 +1 that A has in Z6 enters Z2n+i is A In order that z2n+i shall be periodic, the constant parts of the coefficients of e+aV=r and e~ aV=lt in u2 Z2n+l and u Z 2n+x respectively must be zero. Now since Z 2n+l is similar in form to Z6 we obtain only one equation when the constant parts of these coefficients are equated to zero. The form of
part of
,
.
Z2n+1
J
when considered
,
'
the equation
is
(58)
where
a power series in e with real constant coefficients. l) solution of this equation for Af~ is
in -\
The
AT^-V^Ta*-,,
where p2n -i
is
(59)
coefficients.
a power series in
ISOSCELES-TRIANGLE SOLUTIONS.
x
339
In general, there are no other terms in UjZ2 +1 and u x Z2n + which yield But since <r = N/v, and v being integers, non-periodic terms in z2 +i. there are values of n for which other non-periodic terms than those already
It follows
of
Z2n+l
that
u2 Z 2n+l
K (A?
where
in+l
)
e<*"
*= II
[a
1
+a cos+
1
+a
coskt+
(60)
],
+V=Tb
sinf-t-
+ v^T&* sinkt+
+ V^T
Sill 2U(Tt}.
is
a constant.
e (2,.+l)<r V=It
Now
=e
<r
Consequently these non-periodic terms arise if k = 2n<r, k an integer, or if kv=2nN. This relation is satisfied if 2w becomes a multiple of v. v is odd. Since v and are taken relatively prime, the smallest Suppose values of n and k for which the non-periodic terms in question can arise are n = v and k = 2N. If v is even, is odd; and the smallest values of n = = and k are n v/2 and k N. The terms in which these non-periodic terms
are multiplied by x + /2 e or x'+'e*, according as v is odd or even. After these terms first appear they in general occur similarly at all subsequent steps. When they are present, the equation analogous to (58), in
first arise
(2v
1)
2JV
u2 Z2
,,
+i are
concerned,
2n
1
is
^rwhere
even.
^ +v^TP _ +^
3) x
2n _ 1
(A?r +, =o,
N
(6D
2n - t
The
the sign of
a constant multiplied by e or e according as v is odd or terms in ux Z2n + corresponding to (60) differ from (60) only in and V l. Non-periodic terms arise from these terms in the
is
2JV
The equation analogous to (58), in so far as the as from (60). Z are t concerned, is the same as (61). This equation can be 2n+l
Af
n ~ l)
is
of the
same form
as (59).
Hence
in all cases
~ l)
A{
2n_1>
periodicity conditions.
With Af n
determined as in
is
The
= Ar +l) e'^'u
it
+Ar
+1)
e-'
^ u +V^T I
t
<-0
rfjf.
From
follows that
<Cff(o)=o,
(44)
is
imposed,
z2n+l
becomes
where
Af
is
Af
n+l)
This solution remains undetermined at this step. This the induction. completes purely imaginary.
+1)
is
real
if
340
III.
PERIODIC ORBITS.
PERIODIC ORBITS
The
FINITE.
shall now consider the question Differential Equations. of the existence of orbits which are periodic when m is finite, and which have the same period as those obtained in I. The question is one of deter-
173.
mining initial conditions for m ra2 and /x so that the motion of the system shall be periodic when it is finite, and shall have the same period as when
t
, ,
is
infinitesimal.
The
origin of coordinates will be taken at the center of mass of the The plane passing through the center of mass and perpendicular
1 ,
m,
and
,
/*
be &
',
to
>
f1
&
>
to
ft
and
|,
r?,
values of
t\,
f,
&, and
f2 be zero at
Further,
let
i(0
= -&(0 =
\,
initial
(4)--(4),
conditions
*U)--i(0.
&"
On making
1 1
&,
to" - to
fi"fi-
62 )
m +m
we have
+/z
= 0,
m +m
1 i? 1
i!
t? 2
+A"? = 0,
m^+m^+nt = 0,
(63)
e*o,
Hence /x always remains on the f-axis. With the units chosen as in 166, the
r,+Mr=o.
differential equations are
Sr
[^-{-(I+m)^
Mto
J
[,
2 372
]
'
"to- _!2i_ 8/
.s
+ (1+/i)
2
2 )
r .]i/i'
(64)
"
[r
(1+M)f
2
+(l+ -d+M) M rf f
where
r*
= t* -\-rj
l i
>
.
= rcost>,
Tj^rsin^
I.
= Vi/8(1+5)t,
(64)
(65)
where
Then equations
M(l
become
+ 8)
+ g)r
(66)
rti+2ft> =0,
_
f
(i+M)d+8).r
8[^+(i+M)rf
ISOSCELES-TRIANGLE SOLUTIONS.
341
For
/x
= vV8(1 + 8)t,
where ^
is
r=\,
is
f,
2-ir.
Now let
where
p, u, in (66), the
=VV8(i+)t+m, f*#+w, (67) and w vanish with = 0. When equations (67) are substituted differential equations for p, u, and w are found to be
r-f(l-fp),
/x
(i+p)ii+2p(Vi/8(l+)+M)=0,
,v,
_
[(l
(i+a)(i+/i)Ofr+ttO
+ p)
"
2 2
(i-h
+4(l4- M) (^+;)
of (68)
F "[l+4^r
- VV8(i + 8),
/x,
=
where d
is
(T^p
(69)
an arbitrary constant.
we
substitute
where
do
is
an undetermined constant.
On
substi-
we obtain
3
Pi~8(l+p)
(1
+ S)P _
U>
2doX/*+XV 3 (l+p)
(1
2
'
_ "
(70)
'
3/2
M=
174. Proof of Existence of Periodic Solutions of Equations (70). For = = = = admit the solutions 0. It will equations (70) periodic p p w
not zero, but sufficiently small, equations (70) admit solutions expansible as converging power series in /x, which vanish with and which are periodic in r with the period 2t. Let us take the initial conditions
that
if
|/x|
now be proved
fj.
is
p(0)
= aM
is
p(0)=0,
it
tc(0)
= 0,
w(0)
=a
(71)
With
can be shown from the properties of (70), by even in t and that w is odd in t. Therefore
p(7r)=p(-7r),
()(-),
(72)
and
if
the conditions
P(tt)=w(t)=0,
are satisfied, p
and
to will
342
PERIODIC ORBITS.
Equations (70)
in so far as the a,
will
now be
solutions in
enter linearly are denoted then the differential equations defining p t and w t are
o,
and n a, integrated as power series in a enter linearly in the solutions. If the terms of the
t , ,
by
p,
and
p +i(l+3)p
1
= fi = 2VV8(l + 8)X/x /2
1 (
fj
^
1
wMi+s)[i+.^(~]
-The
first
)(2j+i)(4^y]w
2
=w
(73)
f^ -(i+^( ^+ W) S
1
1
/,
("J
)iW^
The
equation of (73) is independent of the second equation. complementary function of the first equation is
power
series.
The
is
series unless
integer. the left side of the first equation of (73) has the
WsU +
V
an even
If
Wsd +
an even integer
When
same period as certain and the solution will contain non-periodic therefore side, 1 /8(l + 5) is an even integer, the period of the motion of
m^ and mt is an even integral multiple of the period of the oscillations of mThe mutual attractions of the three bodies will then have a cumulative effect and produce non-periodic motion. We therefore exclude from our consideration those values of a for which + S) is an even integer. With this restriction upon a, the solution of the pr equation satisfj'ing the
Wsd
initial
conditions (71)
is
p = [a
1
xC
(0)]
cosWsU-r-^T+M^r),
series,
(74)
and
it
contains X as an undeter-
When
known.
(74)
is
With the
substituted in the i^-equation, all the terms of x are = left side simplified and becomes l 0, the equation
iv 1
+[l+I18 ,a
t
v
~\w1
= 0,
4t,
(75)
where
0*
=~
f
2*
+ 9 cos2t,
is
et
g - 48 cos2t + ^ cos
sum
t,
the highest
ISOSCELES-TRIANGLE SOLUTIONS.
343
Equation
or
\{/\(t
t
one of the equations of variation, and the expression ^(t) )/V 1 /&(l + 5)\, obtained in (19), is the generating solution. Two
(75) is
arbitrary constants, viz., tQ and a, appear in its generating solution, and 32 and 33 the two fundamental solutions of (75) are obtained according to
derivatives of
\p
wii=|-*tt*-o)/Vi/8(i+8)}
and
it is
--;*(*)>
factor
This solution contains the periodic in r with the period 2tt. a, and since it is multiplied later by an undetermined constant
a
be absorbed by the undetermined constant. tion can then be expressed as (see page 330 for \f/)
the factor
may
This solu-
w n = <p= 2
where
t,
<pi]
a21 =cosT
+
(p 2J
a2 (cos3r
are
cosr)+
(76)
a<p
= d\p/dr.
Therefore the
2,7
sums
of cosines of
1.
The
initial
w n (0) = l,
The
t%(0)=0.
other solution of (75) is obtained by differentiating the generating solution with respect to the constant a; hence this solution is
w- =
Ya-\da) + ^Ya'
is
(78)
performed only
in so far as
Now
=
(H)
sinr+^a (sin3T-3sinr)+
=
Tf
dr = drdb Ta dlTa
l-2 a+
.
465
aJr
'
"I
(78) is
w
The
12
=[sinT+Ja (sin3r-3sinr)+
values of this solution are
.]+,-*[- a*+^a*
initial
w u (0)=0,
ti;
12
(0)=fi =
l-|a
+f a<+
344
Since
it is
PERIODIC ORBITS.
more convenient
for
wu
in
which
(79)
wu (0) = 0,
we take
as the second solution of (75)
w u (0) = l,
w^=x+At<p,
where x and
(80)
are found
by computation
2
to
be
._*,
M X= S Xija = sinr+^a
(5sinT+sin3r)+
i>Hk+f*+
From
the
way
in
viz.,
B\daJ
it
is
sum
sines of
odd multiples
of t
highest multiple
2j+l.
Further, since
<p
= --^>
adr
it
follows
of
\p
<p
v and
The
and
is
(80) constitute a
unity,
of (75) is
(81)
Wl
= nrcp+r>l
[x+AT<p],
where wf and n^ are constants of integration. becomes an odd power series in a When (74) is substituted in (73), with two types of terms in the coefficients. (1) There are terms not multiplied by cosV 1 /8(1 + 5)t which enter through p x and they form a triply odd power series. They have n as a factor and will be denoted by n (2) The remaining part of W, consists of terms which are multiplied by cos\/ 1 /8(l + 5) t. As we have already excluded those values of a for which W8(l + 5) is an even integer, and as we subsequently exclude those 1 do values of a for which /s(l + 5) is an odd integer, these terms in not have the period 2ir. In the direct construction of the solutions such terms do not appear in the right members of the wyequations. They appear as the complementary functions of the p r equations and, since they do not have the period 2w, they are excluded by assigning zero values to the constants of integration. which We can, therefore, disregard the terms in do not have the period 2x and consider to have the form fiM
ISOSCELES-TRIANGLE SOLUTIONS.
345
obtain
(82)
By
wf and
w"' in (81),
we
<V+n[x+^r^] = 0,
The determinant
^Htflx+^rH?)]^!,.
wf and
w"' in (82) is
of the coefficients of
a constant
[18], and from (77) and (79) it is seen that the value is unity. Equations (82) can therefore be solved for hf and h2 \ and the solutions are
tij*-pM \x+ATfl,
t
n? = ixM ltp
obtain
l ],
(83)
Upon
we
)
nt
= y t -n[aMTr+Mt +ArR
]
n^^^+ftR,.
(84)
The is a power iff are the constants of integration. a2 with constant coefficients. The is a power series in odd powers of a with sines of even multiples of r in the coefficients, the highest 2i+1 multiple of r in the coefficient of a being 2k+2. The i? has the same form as the except that it has cosines instead of sines. Since the w coefficients of a cos(2j l)r and a"sin(2j 1)t occurring in ip and x
r;"'
The
and
series in
M
t
so also
+ 2) r
and
is
Mf
and
When
is
(84)
found to be
(85)
w^^v+tfix+ATd+^S-aMTTv},
where S
1
r in the coefficients.
is
a power series in odd powers of a with sines of odd multiples of From the form of the <p, x, and Rt it follows that a triply odd power series. The expressions <S\ and carry X as an
is
M,
Mf
undetermined constant. When the constants of integration are chosen so that (71) shall be satisfied, the solution (85) becomes
w = [a l
j.\'S 1
(Q)-aM?\][x+AT<p]+
*[Sl -aMrT<p\.
(86)
Now
and
(86)
.
let
us impose the periodicity conditions (72) on the solutions (74) In so far as the linear terms in a! and a 2 are concerned, we get
p (tt) =
- - VV8(1 + S)
[o x
ft
+ 5) *
in a u
fta^,
+ terms in fia
w(ir)=0=
a 2 Aw
,
and
ft,
(87)
+ terms in a,
ft
in
ft,
at
and a
These equations are satisfied by a, = a., = ft = 0, and X arbitrary. minant of the coefficients of the linear terms in a^ and o2 is
The
deter-
tt.
346
PERIODIC ORBITS.
is an odd integer, exclude those values of a for which and as we have already excluded those values of a for which S) w is not zero and can vanish even is an S) integer, sinVVsU only when a is zero. In order to obtain solutions which are not identically zero,
We now
WsU+S)
WsU +
is
distinct
from
by the theory of implicit functions, equations (87) can be solved as power series in n, vanishing with /*. The coefficients of the various powers of n are power series in a and contain additional terms in l/a\ The X enters the coefficients of the solutions as an arbitrary, but the Hence periodic solutions of (70) exist solutions are unique if X is assigned. and are of the form
and a3
p=
where each
2p n',
t
w=
Sw,/*',
(88)
p,
and
is
2ir.
be at apsides of their
let
Periodic Orbits are Symmetrical. Let us no longer imposed that the equal bodies shall orbits when the third body crosses the ^-plane, and
is
us consider the question of the existence of periodic solutions of (70), with the period 2ir in t, when the initial conditions are
p(0)
= ai
p(0)
=a
w(0)-0J
w(0)
= a,.
2ir
(89)
p and
shall
are
p(2x)-p(0)=0,
p(2tt)-p(0)=0,
w(2t)-w(0)=0,
= 0.
(90)
These four conditions can not be satisfied by the three constants a, unless one condition is a consequence of the other three. We now show that the last condition can be suppressed when the first three have been imposed. The original differential equations (64) admit the integral
() +0h) +M(l+M)
(f')
^+
,
[r
+ (l ^ M)y]) +const.
and (67) are made and u is eliminated by means takes the form
(65)
,
PF
=
(91) )]
+7j|p5i+4|*(l+/)(*+w)
(1
+ 8)
+ const ]
ISOSCELES-TRIANGLE SOLUTIONS.
Let us make in (91) the usual substitutions
347
p = p(0)+p,
p=p(0)+p
at r
w = 0+w,
w = w(Q)+w,
(92)
denote the resulting equation by (91a). By putting r 0, (91a) an equation (91b) conin the terms of When this equanecting (91a) independent p, p, w, and w. tion (916) is substituted in (91a) there results an equation of the form
p, p,
where
w, and
w vanish
=
F(p,p,W,w)=0,
in
(93)
linear
which there are no terms independent of the arguments indicated. term in w enters (93) with the coefficient
The
8*(l+ M)&H-tK0)],
which,
is
show, is different from zero at t = 2ir. assumed to be finite, 8/z(l +m) is distinct from zero.
we
shall
t= 2x
unless w(0)
a.
Now
a/y/i/%(\ 8), determined that the orbits shall be periodic in t with the period
third
when assumed to be infinitesimal, has the initial speed and w(0)/V 1 /8(l + 5) is the additional initial speed to be so
finite.
speed speed is zero, and the third body remains at the center of gravity since there is then no force component normal to the ^77-plane. In order therefore to obtain solutions in which f is not identically zero, we must
body becomes
is
initial
Hence the coefficient of the linear term w in (93) is disconsider w(0) 7* a. tinct from zero at t = 2ir, and therefore (93) can be solved for w(2ir) as a
power series in p(2ir), p(2ir), and w(2t) which vanishes with p(2ir), p(2w), and w(2t) This power series is unique if X, which appears in the coefficients, Hence if the first three conditions of (90) are imposed, then is assigned. p(27r)=p(27r)=M;(27r)=0; and since w(2w) vanishes with p(2x), p(2ir), and
.
w(2ir),
follows that w(2tt) =0 or w(2t) w(0) = 0. Therefore the fourth equation of (90) is a consequence of the first three and may be suppressed. Equations (70) will be integrated as power series in n and a x {i=\, 2, 3),
it
but only in so far as the a, enter linearly in the solutions. The differential equations from which these linear terms in a, are obtained are the same as Their solutions satisfying (89), in so far as the linear terms are con(73).
cerned, are
a2
(94)
348
PERIODIC ORBITS.
the
first
When
p and w, we have as a
/8( I
consequence of (94)
0= a,
[COS
V/8(l +
2*
1]
+ Vl/8 (l + a) Sln V
/xa 3
,
'
+^ 2?r
(95)
+ terms in n,
+ terms
= 2cl At+ terms
3
[cosA/V8(l
+ 5)27r-l]
/za,
in a 1; a,,
/z,
The determinant
in (95)
is
+ 5)
2tt].
This determinant does not vanish when a is not zero, and consequently These (95) can be solved for o, as power series in /x, vanishing with n. solutions are therefore unique if X, which enters the coefficients, is assigned. Hence the periodic solutions of (70) for p and w, with the initial conditions (89), are of the same form as those obtained in (88) for the symmetrical The unrestricted and the symmetrical orbits are unique for a not orbits. zero and for any value of X, and therefore, since the unrestricted orbits
include the symmetrical orbits,
all
In order 176. Direct Construction of the Periodic Solutions of (70). to construct the periodic solutions of (70), we substitute (88) in (70) and equate the coefficients of the various powers of m- The arbitrary constants
of integration are to shall be periodic in r
The
differential
be so determined that w(0) =0 and that each p and w, with the period 2ir. equations for the terms in /x are
t
ft+i(l+*)Pi- ^i
=2<U- (T+^)
T/ij
The
is
p = ^ eosVV8(l+5)T+5 sinVi78TT:H)T+^iW,
1 1
where
even power series as in (74). Since the complementary function does not have the period 2t when \/i/8(l + 8) is not an integer, the arbitrary constants A and B must be zero in order that p, shall be periodic with the period 2ir Hence the desired solution of
l
(j) is
the
same
triply
the ^-equation
is
Pi-GM.
(97)
ISOSCELES-TRIANGLE SOLUTIONS.
349
When
(97)
is
substituted in
all
the terms of
are known.
The
general solution of the ^-equation is the particular integral does not carry the factor
>i
(98)
Since
ttf(O)
all
we impose
= l,
the condition
w,(0)=0
(t
...
oo).
(99)
As a consequence
of (99), the constant ijj is zero. In order that to, shall be periodic, the right member of (98) must contain no terms in r except those in which it occurs under the trigonometric symbols. The non-periodic
if
the constant y^
is
=aMf\
from which
it
follows that
^> =
where
ipV),
P a \a?)
}
values of
r?"
a power series in a2 with constant coefficients. When these and r?"' are substituted in (98), the solution for w t becomes
is
1 viJBflr* **
(100)
of r, the highest
where each
tiple
S^ +l)
is
sum
of sines of
odd multiples
terms in
2 2 /x
mul-
being 2/
1.
The
are
(a)
(6)
w +[l+ le
2j
a^w = W
2
(101)
equation is independent of the second, and the terms in P2 are completely known. The complementary function of (101a) does not have the proper period and is excluded from the solution by taking zero values for the constants of integration. Since P2 has the period 2-w the particular has the same and is the solution desired. The function Pt integral period 2 is a triply even power series multiplied by 1/a and since the particular integral has the same form as P2 it is denoted by
The
first
Vi
^2<rv,
even multiples of
r,
102 )
where each
is
sum
of cosines of
the highest
multiple being
2j.
That part 2 has been obtained, all the terms in 2 are known. which is w of independent of p 2 and x is a triply odd power series. The 2 term p2 is multiplied by a triply odd power series and yields a triply odd 2 The terms w and w\ are multiplied by triply power series multiplied by 1/a
When p
series respectively
and together
yield a triply
odd
350
PERIODIC ORBITS.
series multiplied
power
is
by
w\\}/
1/a*.
to
enters a*
where each
+1)
is
sum of sines
of
odd multiples
is
of
r,
being 2j+l.
of (1016)
Wt = n?<p+n?[ x +AT<p].
(103)
as (85)
The
same form
and
is
denoted by
(104)
i?<P+i?[x+AT<p]+St -\-M?T<p, a
and where j is a power series in a2 with constant coefficients. The jj{ and jjj are the constants of integration. The constant t]f must be zero to satisfy (99), and in order that w2 shall be periodic rjf must have the value
2
2>
0)
2)
C=^ = ^ V),
(2
(105)
where
(2)
(a
t\f
values of
2 a power series in a with constant coefficients. and t^2', the solution (104) becomes
is
With
these
^=^2J^ a
7=0
w+1,
w+l
,
(106)
where the S?1+1) have the same form as the 8?* in (100). The form of the w} is apparent from (100) and (106). The form of the P>0>2) is not apparent from (97) and (102), and, before the induction can be made, it is necessary to consider the term in // in so far as p 3 is concerned. The differential equation for p 3 is
R+!(1+*)A-1V,
where
Wj
is
(107)
all the terms of P3 are known. That part of a triply even power series multiplied by 1/a2
.
independent of the
Wj enter
The
multi-
by power series which are triply odd or triply even according as the considered These Wj, together, enter to odd or even degrees respectively. 4 terms form a triply even power series multiplied by 1/a Since Pz contains
plied
.
2 4 m^, the lowest power of a in a P3 is unity. The complementary function of (107) does not have the period 2x, and the solution desired is
the term
and
is
denoted by
V*
jrStcroP,
(108)
where the C
in (102).
ISOSCELES-TRIANGLE SOLUTIONS.
Let us suppose that the
351
j
pif w
{ ,
iff (*>-l,
n-1;
= l,
2),
have been
y=i
(109)
^2
*?!
Wj
^nri-r
(.a J,
have the same form as C?\ S? j+l) and From these assumptions and the differential equations (a ) respectively. from the coefficients of n" it will be shown that equations (109) hold arising
where the Pm 2
?'
,
C,
+1)
and
(0
(a
when i = n.
(a)
The
terms in
fj."
are
p.+|(l+*)p.-P
(b)
+\\+ljv a?>\w
= Wn
(110)
As
equation is independent of the second. Since the right member of the first equation in (70) carries n as a factor, the Pj and w} which enter Pn have j < n. Hence all the terms of P are comThe general term of Pn has the form pletely known.
K p
multiplied
Kw
tfi
(111 )
Mi+
by a triply odd or by a triply even power series according as are y.j, v'j +M^ is odd or even respectively. The X,, \[,
.
\K+
From
series multiplied
+\K+ihlH+
it
+MjM'^tt-l.
(112)
follows that
Pn
is
by
1/a',
where
i=(2\-2)K+
+/*,/#.
113 )
= \' = 0, i. e., This expression is even and has its highest value when X( = t in the terms of P in which only the Wj appear. Hence, from (112), the i of is 2n 2. P value the term wn _ ^, the lowest Since contains highest 2 tn ~ 2 Pn is found to be unity. Therefore the form of P is power of a in a
1
OO
Pn _ r
(21)
j n tn ~ i
'y pew n v
j
'
have the same form as the Cf. The only solution of where the P (110a) which has the period 2x is the particular integral, and it has the form
CO
lJ n
~
ritn-i
Zj ^n u
(2/)
where the
(2i)
are of the
same form
as the CJ
352
PERIODIC ORBITS.
has been determined, all the terms in n are known since has the from Pj(j^n) and wk (k<n). The general term of same form as (111), but it is multiplied by a triply odd or triply even The power series according as m>+ +/*J is even or odd respectively.
When p n
they arise
\,\[,
Hj,
n'j
\\[+
From
the form of the general term
it
+n
n' j
^n.
(114)
follows that
is
series multiplied
by
(1/a)',
where
l=(2\-2)\[+
+2Mj m;.
This expression is even and has its highest value, 2n, in the terms of n in The lowest power to which a enters ain n is which only the w} appear. obtained from the term in which ^ enters to the lowest power. This term in is WiW^ip, and therefore the lowest power of a in a n is found to be 3.
W W
of
is
*.where the
2 *!*?*.
.
have the same form as the SfJ+l) The complementary function of the second equation
J+1)
of (110)
is
B)
nJ
and n
we have
i
(115)
and where Mf is a power series in a 2 with constant coefficients. The i^"* and r^ are the constants of integration. The constant n) must be zero to satisfy (99), and in order that w n shall be periodic, itf* must have the value
where Sa has the same form as
Tjj
n,
_()
_
2
pd)/2\
a power series in a with constant coefficients. With these values of the constants of integration, the solution (115) becomes
)
where
P M (a
is
w,B
=^S^
+1,aW+1
'
where the
induction.
+1)
ISOSCELES-TRIANGLE SOLUTIONS.
177.
353
p
is
is
Periodic Solution of Equation (69). When the solution for substituted in (69), u can be obtained by a single integration. Since p
The
a power series in /x with coefficients which are triply even power series 2J multiplied by l/a the right side of (69) will contain terms independent of t. After the integration, u will therefore contain a term in t with X appearing in the coefficient. We shall now show that X can be so determined that
,
be zero.
When
periodic in r with the period 2-ir. The solution for p x obtained in (97), in so far as the term in X
con-
cerned,
is
p^vm+ry
When
substituted in (69) and d is replaced terms appearing on the right side of (69) are
(116)
is
(116)
X/x
and
fx.
(117)
we may
replace
X/x
by
a,
becomes
3o--f- higher
(118)
equated to zero, the resulting equation can be solved uniquely for for |ju| and \a\ /x, vanishing with n (and converging Let this series be denoted by sufficiently small)
If (118) is
a as a power series in
.
00
is
found to be
x=
1-0
xy,
(ii9)
the
<tj
and
X^
being constants.
With
and
is
since X
series in
the
is
also
denoted by
u= 2
where each
Ujijl',
\n\
sufficiently small.
The
solution
(120) converges for \a\ and |m| sufficiently small. In order to construct the Uj directly, we substitute in (69) the series (119) and (120), and the solution already obtained for p, in which X is to be
replaced
by
(119).
The various
X,
354
(69) shall contain
PERIODIC ORBITS.
no constant terms
ju'-
The
con-
-3[X
the X"" being
- 2 XTa"L
is
known
constants.
a
This term
zero
if
2j
<X
1=0
terms up to ju3 before the induction to the The constant term appearing in the coefficient
\ has
the value
\=
The constant term
can be written
2
i-0
Xf'V'.
is,
it
B
the
y=o
X^
being
known
constants,
and
this
term vanishes
if
is
so determined
that
00
\=
Suppose X way and that
,
~2j\ 1=0
\=
A
2 XfV',
X^-^jgxfV'
e = l,
-D,
it
the Xf being known constants. From the form of p t in (115) that the constant term in the coefficient of nn+l has the form
CO
follows
where the \
j> }
and X
if
(t
= l,
. ,
w-1), and
is
zero
2 " -2
^ &
00
\ aJ)
"
n 2> *
The same
tinued.
X,
ISOSCELES-TRIANGLE SOLUTIONS.
355
With X thus determined as a power series in n, the integration of (69) yields periodic terms only, and from the form of the j> s it follows that the u,
have the form
^pa^U?*
where the
being 2j.
(*
= 1, ...
>),
are
sums
of sines of
even multiples of
is
r,
The
therefore
u
where
-S^S^v+tf,
the constant of integration. Since the mass ra x is started from the point 0, 0, 1/2, at t = t or at r = 0, it follows that v = at r = and there= is zero. Hence the constant U is zero. fore, from (67), the value of u at t
is
178.
The Character
and w have been determined, the and f are obtained by means of the equations (67), (65), (63), and (62). These solutions are all periodic in t with the period P = 2 r vVsU + S) Three arbitrary constants appear in the solutions, viz., a, n, and i
tions for p, u,
, 4
. .
The
S) represents the initial speed of the third expression a/VtysU The mass /x is of the third body, and ta the epoch. in the mass body I, n increased restricted in magnitude but can be step by step by making the
analytic continuation of the solutions already obtained, provided the series do not pass through any singularities in the intervals. This can be done
by the process already developed. The parameter a is restricted in magAs already stated in nitude and so that VVsU S) is not an integer. 166, it can be shown from equation (2) that the motion of the infinitesimal
body
will
be periodic
if
is
negative.
With the
conditions chosen as in
if
(4),
the constant
+ 6)-l[. Now
infini-
body recedes to infinity with a velocity which is zero or greater than zero respectively, and therefore the motion will not be periodic. Hence a must be restricted so that 2a2/(l + 5)<l. The epoch t is arbitrary and may be chosen to be zero without loss of geometric generality. Hence for given values of \n\ and of \a\ sufficiently small and such that VV8(1 + 8) is not an integer, there exists one and only one set of periodic orbits which are geometrically distinct with the period P in t, and which reduce to those obtained in I for ju = 0. In proving the existence of the periodic solutions of (70), if the period were chosen to be 2vw in r, v an integer, it could be shown by the same process that periodic solutions would exist under the same restrictions on
356
a and
m-
PERIODIC ORBITS.
X could be determined so that the solution for u would have the period 2vir, and hence the periodic solutions of (64) would be unique as soon as v were chosen. Therefore, for given values of a and of /x
The constant
and such that Vrfsil+d) is not an integer, there exists one and but one set of orbits which re-enter after v synodic revolutions. Hence the result is unique for every v, and since the orbits reentering after v revolutions include those reentering after one revolution, there are no orbits which for m = Q reduce to those obtained in I, having the period 2vt in t which do not have the period 2ir also.
sufficiently small,
ti fs and f are odd in a and in t, the orbits are symmetrical with to the The two masses 1 i^-plane, both geometrically and in t. respect in the same orbit and always remain 180 apart. and 2 move
Since
>
Chapter XL
PERIODIC ORBITS OF INFINITESIMAL SATELLITES AND INFERIOR PLANETS.
179. Introduction.
This chapter
is
periodic orbits of the problem of three bodies in which two of the masses are The finite bodies are assumed finite, while that of the third is infinitesimal.
body to move in the plane of their other of the finite bodies. The periodic motion, relatively near one or the orbits which are obtained are those in which the periods of the solutions are
to revolve in circles,
and the
infinitesimal
The nearer the infinitesimal equal to the synodic periods of the bodies. the less its is to one of the finite bodies motion is disturbed by the more body
distant one.
The
zero, and they are therefore of the class de la Poincare "Solutions by premiere sorte."* The results of this chapter are of direct practical application, particuThey are coextensive in this domain with the larly in the Lunar Theory. Researches^ of Hill and the first memoir by Brown. $ When the masses of the two finite bodies have the ratio ten to one the problem reduces to that which Sir George Darwin treated by numerical processes, and the results obtained include the orbits called by him "Satellites A" and "Planets A." Darwin's "Satellites B" and "Satellites C" are imaginary for small values of the disturbing forces and belong to values of the parameter for which the series obtained in this chapter do not converge. It appears from the present that there are three families of satellites and of inferior planets investigations whose motion is direct. It follows that Darwin's search for them was exhaustive so far as the satellites are concerned; but he found only one family of
is
direct.
The work
an equal number of retrograde orbits; imaginary periodic orbits around each of the finite bodies. As compared with previous work on this subject, the methods
chapter are characterized
is
of this chapter shows that there that is, three families of real or of this
employed
the
proved.
by the fact that the validity of all the processes They have the merit of generality, not only in showing
but also in being applicable to any ratio of They have the disadvantage that the series do values of the parameter. The numerical processes
and
Hill's Collected
number
*Les Melhodes Nouvelles de la Mecanique Cileste, vol. I, p. 97. turner. Jour. Mathematics, vol. I (1878) pp. 5-26, 129-147, 245-260, %Amer. Jour, of Mathematics, vol. XIV (1892), pp. 141-150.
Works, vol.
I.
XXI
358
PERIODIC ORBITS.
easily applicable to the cases where the series fail. Instead of arranging the solutions as Fourier series, as was done by Hill and as has been customary among astronomers, power series are employed, and
is
power series. For example, every coefficient and is modified by no subsequent operations.
180.
Let m, and m 2 represent the masses and take the origin at m, for the determination of the motion of the infinitesimal body. Then the differential equations of motion in polar coordinates are
r"-r(*/)
+~^ = & m ^,
2
2
rv"+2r'v'
1
=k
i
m^,
(1)
tj
=
[R
rcos(vV)
i
-2rRcos(v-V)+r
are the polar coordinates of infinitesimal body, and are the polar coordinates of 2 In the present discussion those orbits are considered in which r
where
and
and
is
small
relatively to R.
In this case
(1)
U is
become
},
U=^[l
r "_ r(t
,
+ \^[Wcos2(v-V)]+l^[3cos{v-V)+5cos3(v-V)]+
T+
=
(2)
j,
+5cos3(y-F)]H
rv''+2r'v'
It
-^^{d
m(v-V) + ^[ S in(v-V)+5smZ(v-V)]+
of
is
was assumed in the beginning that the relative motion Hence we have, from the two-body problem, circular.
and
R = A = constant,
where
Af
is
V=
ky/n
^
,
nh
(t
-Q=N(t-
1 )
(3)
When
the angular velocity of the relative motion of the finite bodies. the right members of the second and third equations of (2) are
put equal to zero, that is, when the infinitesimal body is supposed to revolve about m, without being disturbed by m 2 they have the particular solution
r
= a = constant,
3^
(t
)=n(t
t ),
(4)
where n
is the angular velocity of the infinitesimal body with respect to m, In the periodic solutions of (2) when the right members are included, the mean angular velocity will be kept equal to n and a will be defined by the equation
.
n2 a' =
2
/i;
mj.
(5)
359
Since x is given, a has three determinations, two of which are complex. In the physical two-body problem there is interest only in the real value of a, and it is immaterial whether a or n is regarded as given. It will be observed, however, from the purely astronomical point of view, that n can be determined
by observation much more accurately than a, because the former is an angular variable and any error in its determination causes the theory to deviate secularly from the observations; while the latter is a linear variable and the discrepancies which arise from errors in its determination do not accumulate. But in the three-body problem all three values of a must be included in determining orbits whose period is defined by n, because all the orbits may become real for certain values of the parameters which occur in the right members of the differential equations. In fact, Darwin found three real
orbits for certain values of the Jacobian constant.*
In the Lunar Theory, as developed by de Pontecoulant, for example, the But solutions were forced into the trigonometric form by various artifices. it will be noticed that the method of procedure was the opposite of that adopted here, so far as comparison can be made, in that the mean motion was continually modified by de Pontecoulant in order to preserve the trigonometric form; while here the mean motion is regarded as being given arbitrarily in advance by the observations or otherwise, and it is kept fixed. New quantities p, w, t, and m will be introduced by the equations
r
= a(l+p),
n
= n{t-Q+w,
An ,. f = -{t-Q, r={n-N){t-Q
; , s ,
.
N
=^7n>
A
As a
p
=^>
N{l+m)
(6)
Kmi+rrhs
Vl
+m/
\mi+m /
2
>
For brevity a/ A
will
be written
a power
series.
two equations of
(2)
become
-_
(1+p)(1+m+
^ + ^^|!_^_( 1+p
-^^^-(l +
)j[
1+ 3cos2(x+;)]
.
+ !j(1+p)[3cos(t+w)+5cos3(t+w;)]+
(l+p)^+2p(l+m+tb) =
p)J3sin2(r+^)
1
(7)
I,
These indicate derivatives with respect to t. equations are valid for the determination of the motion of the infinitesimal body as long as a(l+p) is less than A.
and
XXI
360
It follows
PERIODIC ORBITS.
from equations (6) that ap and w are the deviations from uniform circular motion due to the right members of the differential equations. They are functions of m, and the initial conditions are to be determined so Since the right members that they shall be periodic in t with the period 2t. of (7) contain m* as a factor, the periodic expressions for p and w will contain ms as a factor.
= with respect to solution p = w = as the parameter. Since t enters explicitly in the right members of (7) in terms having the period 2x, it follows that the period of the solution must be 2w, or a multiple of 2ir. Equations (2) are not altered if we change the sign of v, V, and t. It
For m = = = equations (7) admit the periodic solution p w 0. It will now be proved that for m distinct from zero, but sufficiently small, equations (7) admit a periodic solution which has the period 2x in t, and which is expansible as a power series in m vanishing with m. It is the analytic continuation of the
181. Proof of the Existence of the Periodic
Solutions.
l/3
from this that, if v (t ) = V (t ) = p' (t ) = 0, the dependent variAn orbit in ables p and v are even and odd functions respectively of t ta which these conditions are satisfied is symmetrical with respect to a line Such an orbit will be called symmetrical. always passing through m and m 2 in the of variables Expressed (7), p and w are respectively even and odd functions of r in the case of symmetrical orbits; and w(0) =p(0) =0. The existence of symmetrical periodic orbits will be established, and then it will be shown, in connection with the construction of the solutions, that the
easily follows
.
condition that the solutions are periodic implies that they are also symIt will follow from this that all of the periodic orbits of the type under consideration are symmetrical.
metrical.
initial
conditions are
to
= a(l+p) = a(l+o)(l-e),
= ai> P=
>
= 0,
(l
r i
If
l+m+w= -r/rc-t-u>
r
+ m )^+e'
/s
(8)
,
(i-j-o)'
(l
e)
s/ *
the right
members
= a(l+p),
0=(l+?n)r+M>
made, then equations (7) reduce to the ordinary polar equations of the two-body problem for the motion of the infinitesimal body with respect to m In this two-body problem with the initial conditions (8), it is found that aa is the increment to the semi-axis a, and e is the eccentricity of
is
l
.
two-body problem known, we can at once write down the properties of the solution of (7) in terms of a and e, so far as they are independent of the right members of the equations. It was precisely for this reason that p and w were given the
peculiar initial values defined in (8).
361
e,
Equations (7) are now integrated as power The results have the form
P = Pi ( a p=p
2
>
series in a,
and
m}'3
e,
l/3
;
t),
(a,
fl,
ra 1/3 ; t),
w=p w=p
(a, e, (a, e,
m m
l/3
;
r),
1/3
;
t)
can be taken so small that the series converge while t runs through any finite range of values starting from zero. The period must be a multiple of 2w, say 2kir, and consequently it will be supposed that these parameters have such small moduli that (9) converge f or ^ t ^ kir. It follows from the symmetry of the orbits under the initial conditions = kir, the three bodies are in a line, and if the infinitesimal body (8) that if, at r is crossing perpendicularly the rotating line which joins m A and m 2 then the orbit necessarily re-enters at 2kir, and the motion is periodic with the period
of a,
, ,
The moduli
e, and m
1/3
2kir.
Conversely,
if,
at t
= 0,
,
line
which joins
and
.
the orbit crosses perpendicularly the rotating and if the motion is periodic with the period 2ir,
then the orbit at r = kir necessarily crosses perpendicularly the rotating line which joins m x and m2 Therefore, necessary and sufficient conditions for the existence of symmetrical periodic solutions having the period 2 kir are
p2 (a,
Since p2 = p3 =
1/3
e,
mV3
kw)=0,
p3 (a,
e,
l/3
;
kir)=0.
(10)
at t
fied by a = e = ra = 0.
in
1/3
,
The problem of solving them for a and e as power series vanishing for m = 0, will now be considered. All the terms of the solu-
from the solutions of the Since these terms belong to the twoleft members of (7) set equal to zero. when these terms are explicitly exhibited, body problem equations (9) become,
tions
p = p1
P = p2
V!
3
e^osi'T +|(cos2j/t
1)+
'~\+m
]
q1 (a,
e,
I/s
;T),
=+
'sin
vT+ev sin2^T
+
+
+ra2 qt (a,
2
e,
w}'3 r),
;
=p = = Pi=
\_-(l+m)T+vT + 2esmvT+
[
]+m
g3 (a,e,
2
(11)
1/3
;
r),
1/3
t'r
(l+m)+i>+2i>ecosvT
]+m
g4 (a,
e,m
;T),
/ 1oN (12)
where
1+ra
"-$+&'
The
series q l
,
. .
tions (7).
qi depend upon the right members of the differential equaAll terms in the [ ] which are not written contain e 2 as a factor.
.
The
become
Ptikir)
conditions, (10), for the existence of symmetrical periodic solutions as a consequence of (11)
ev\smvkir-{- esm2vkir+
[
+m}q
(a,e,m
i
l/3
;
kir)
=0,
l/3
p 3 (kir) =
(l-t-m)kw-t-vkir+2esm2vkT+
~^ j
+m q (a,e,m
3
;kir)
= 0,
(13)
362
PERIODIC ORBITS.
all
where
carry
as a factor
and are
linear
v
On
from
sin[l4-w
integer.
|o+
],/*=(
[ ]
1)* sin
[m
|a+
-~\jkir }
where j
is
an
contains either
m or a as a factor;
mora
is
therefore every term of the second equation of (13) contains either The coefficient of a to the first power is as a factor. 3/2 kit, which
distinct
from zero; therefore the second equation of (13) can be solved for The term of lowest a as a power series in m and e, vanishing with m = 0. its coefficient in m alone is the and depends upon the right second, degree
member
Therefore the solution of the of the differential equations (7). 2 second equation for a carries ra as a factor, and has the form
a = ra2p(ra1/3
,
e).
(14)
Suppose a
is
first of (13)
can be
degree,
divided out.
and
its coefficient is ( e
Therefore
vanishing
(15)
as a
power
series in ra 1/3 ,
= mq(m
1/3
).
As a matter
be shown.
as a factor, as can
1/3 Suppose equations (7) are integrated as power series in ra , easily = = = = and let the initial conditions be p P w w 0, in order to get the terms
e.
The
series will
+p
wi 8/3
''?*
is
w=w
m-)rWi
m -\-w mm
i i
-\-
The
m-t
1+2cost
cos2t],
p3 13
3
^r ^ L2 mi+Wj ^r
\r
cost +f 3
-cos2t
6
2rsin7-l, J'
w =
From
~ TsinT+^sin2T
4tcost],
2 these equations it follows that p2 (kr) has as a factor, but that it 3 does not have as a factor. Therefore the expression for e as a power
series in
mv
contains
m as a factor. a = w P (w
2 2
1/s
1
Then
e
(15)
2
and
i
),
= m P (m
'),
(16)
363
where
of
and
are
power
m sufficiently small.
equations (16) and (8) terms of the parameter m m which is defined, except for the cube root of unity, by the data of the problem and the third equation of (6)
.
which converge for the modulus Therefore the symmetrical periodic orbits exist, and give the initial values of the dependent variables in
series in
1/3
182. Properties of the Periodic Solutions. The periodic orbits whose existence has been proved re-enter after the period 2 for, where k is any
Those orbits for which k is greater than unity include those for which k equals unity. Since, according to the discussion which has just been made, the number of periodic orbits is the same for all values of k, it
integer.
follows that the period of the solutions is 2-k. When = the infinitesimal in can be taken so small that the orbit body makes a revolution 2t, and is as near this undisturbed orbit as may be desired. Therefore a synodic
in 2-k for all \m\ sufficiently small. If the expressions for a and e given in (16) are substituted in (9), the result becomes
is
revolution
made
= 2
i-e
pMm"
3
,
w=Sw,(T)m i=*
,/J
,
(17)
where the summation starts with j = 6, because the expressions for a and e have m2 as a factor, and p and w have no terms in m alone of degree less than
the second.
The
and
Therefore
3
;
1=6
whence
(t+2w)= Pj (t),
it
Wj(t+2t) = w,(t)
follows that
s
(18)
2
1-6
hi (0)
m" = 0, ^(0)=0
body
._
t
,
1-6
iwj (0)m' = 0;
/3
whence
p/0)
If
= 0,
(19)
retrograde,
is
negative and
ra=
._
-N/(n+N)
for a given sidereal period. Therefore, for a given numerical value of n, the is smaller in retrograde motion than it is in direct motion. For parameter
a given sidereal period the deviations from circular motion are less in the retrograde orbits than they are in the direct. The physical reason is that the disturbance of the motion of the infinitesimal body by m^ is greatest when the three bodies are in a line, as can be seen from (7) or by graphically
resolving the disturbing acceleration; and in retrograde motion this approximate condition lasts a shorter time than in direct motion.
364
PERIODIC ORBITS.
= j Mm,
(20)
is to be regarded in the analysis as a constant independent of m. where This amounts to a generalization of the m as it appears in certain places in
the last equation of (6). The particular transformation (20) is made in order that the right members of (7) shall be in integral powers of m. The
proof of the existence of the periodic solutions can be made precisely as before, because the transformation (20) affects only the higher terms which were not explicitly used. While there is nothing essential* in the transformation,
(7)
it will
become
p-(l+p)(l+m+^)
+^^ =
^7?(l+p){[l+3cos2(r+ti;)]
(
(21)
~Yv(l+p)hsm2(T+w)
j,
+ fMm(l+p)[sin(T+w) + 5sin3(T+w)]+
where
v
irk
mi+nk
In the right a
sum
of the first equation of (21) the coefficient of m' is of cosines of integral multiples of r+w, the highest multiple being j;
member
member
sum
is
of
sines of integral multiples of (t+w), the highest multiple being j. In a closed orbit around t there are two points at which
zero.
The
arbitrary
will
The
first
condition
of (8) will not be imposed in advance, and it will be shown that it is a consequence of the others. It will follow from this that all of the periodic solutions of the type under consideration are symmetrical. Equations (21) will therefore
be integrated
in the
form
CO
= 2 pjtn',
w=2 Wjtnf
American Mathematical
oo
(22)
365
2
.
p i -3p2
+2'p2 = -|r?sin2r,
is
P2
2 cos r + cf \n + cf + cf
(ij
sin r
- v cos 2r,
(23)
sin 2r,
w = cf 2
+ 3 cf t - 2 cfsin r + 2 cf cos r +
)
where
2)
By conditions
(18)
and
c?=-~v,
Therefore the solution (23) becomes
p2
cf
2c
(24)
-7j
+ cf cos r -f c
sin r
77
cos 2r,
'
cos r
+ ^ sin 2r,
far undetermined.
(25)
where
185. Coefficients of
3
.
The
differential equations
are
cos r +5 cos 3t]
2w
[3
ib 3
+2p =
3
(26)
2p2
|M?7[sinT+5sin3r]
=
(2cf
Mr?)
sin r
- 2 cf cos r - 4j sin 2r it is
f Mr; sin 3r
From
found that
(27)
which substituted
)
(28)
366
PERIODIC ORBITS.
of cost
In order that the solution of this equation shall be periodic the coefficients and sinr must be zero; whence
<f-gtf9
after
tf-0,
(29)
a)
(30)
where
c, c, and
The
result of substi-
w = 2r7-3c;
3
3,
3,
|iWr,cos3r.
(31)
In order that the solution of this equation shall be periodic the right must contain no constant term; whence
member
<f-ff.
With
this value of
(32)
cf
it is
second condition of
found, upon integrating (31) and imposing the computation has been made,
p2
= =
r,
cos 2t,
P3
+ \ v + cf cos t + c
sin r
cos 2r
|j
r,
cos 3t,
2 cf
+ 1 Mr, )
sin r
+ ^ q sin 2r +
where
c"'
integration will be carried one step further and then the induction to the general term of the solution will be
186. Coefficients
1
The
made.
The
differential equations
coefficients of
are
(34)
iv t
+ 2 p, =
pj
2 p3
2 p2 wt
a p
sin 2r
2
77
3m, cos 2r
_ -^M
16
[2sin2r+7sin4r].
367
developing the explicit values of the right members of (34) by means of (33), it is found that
P 4 -3p 4
Upon
-2w
]sinr
+ [-f^-^-iMV -fM^]sin2r
The
first
is
w, =
- 2 Pl + c< - 2 cf sin r 4)
[&
g Mtj
cos r
]
(36)
+ [^+^+l^Y+^]cos2r
where
of iv t
,
[f M*- ! M
cos3 '
+ [f *+S W| cos4 ^
Then, on substituting
this value
4)
is
p4
+p =[2cf-f v
4
+ [-2<?
+ [f"
2
+f M'-M]cosT + [-2i,'+f
(37)
+ ||MV+fM^]cos2r+[-f M^ + |M,]cos3r
+f W|cos4t.
C
f = fM^-|Mr,,
its
ef-0
(38)
must be
satisfied.
Then
Pl
= 2cr
-f Tf+f^-gMy + ^M^+^'cosr+^sinr -S MV -f M2, cos2r +[+! + [+I M -l^] cos3 - + [-|" -iH COs4T
,
r>
,?
']
"
'
where
4>
Cj
c, and
4)
368
PERIODIC ORBITS.
If (39) is substituted in (36), it is
3
-2crsinr+[-7,
+^ + ^|MV+j|M
fi
7,]cos2r
(40)
+ [-i^+S*']*"* + [f*+g*V|cos4r.
The
periodicity condition determines
c{
by the equation
(41)
cT
+^-r,+^My-\M\
is
Then the
(42)
The
p2
= =
-7j
D
+ Id Mtjcost
T7COs2r,
w =
t
Mrjsmr +
rism2T,
P*
=-[^M
sin3r,
+ [+ff-f,-iMV-|M^]cos2r
(43)
w = - 2<f
4
+2ejJ
2
cos r -
4)
[2c<
+ ^ ^V - ^ Mi,] sin r
+ [-^
+^ + li MV+lM^]sin2r
and where
4)
where
iy
'
c,
and c^ are so
far
undetermined.
369
is
and
the highest multiple being j. four arbitrary constants which arise at that step is determined by the periodicity condition on the wJ} and another by the initial condition on the w,.
sines respectively of integral multiples of r, At the j th step of the integration one of the
The
on
other two constants remain undetermined until the next step, but two which arose at the preceding step are determined by the periodicity condition
pj.
It will
187.
Induction to the
. .
p2
Pj
'
Wj=
Pn-i= +4'
sinT+i81 sin2T+
B - ,)
1)
0>
(j
= 2,
n-2), n-2),
= 2,
- 1)
1)
sinT+aJ
n - ,>
+c*- cosT+a<
- 1)
cos2T+
B " 1>
+a
t
(B
:l
1)
cos(n-l)r,| (44)
wn ^= -2cr- +2c<
cosT+[-2c;''- +6<
1)
]sinT+/3f- sin2T
-1 and cf " are where the a", &?, and 6j"~" are known constants, and Cj" undetermined constants. In writing the differential equations which define p and w all unknown The terms involving these undetermined quantities will be given explicitly. are the same at coefficients every step. It is found from equations (7) that
the coefficients of
p n -3pn
-2w n =+2c?Q
cosT+2c
sin
r
(
l)
sinT+Pn (pj,Wj,w
r),
iv;
w n + 2 p -+2cf
where
the
(45)
r-2df
x
,
r).J
and
known
n 2) and are polynomials in pJt p} ,Wj, and w, (j = 2, and w_i and where r enters in the parts of p_i p_i wn _
. ,
and
cosines.
It follows
from
from numerical
coefficients,
has terms
of the types
P" = ph Wj,
}
O'i
+j2 = n,
or n -
1)
P=p
Wh W h
'
(h+J2+h = n),
p)' r
PT = p)\ P^ = M'p)\
If
is
GU,+
+hjv =
,
?i,
n-l,orn-2),
&V&
+k^j+l;
+\p
if
li
= n-2).
\+
+Xiseven,
multiplied
by
costV;
and
\+
+XM
come
odd, the term is multiplied by sinir. The terms P"', Pf, and Pf from the left member of (7), and P? comes from the right member.
370
It follows at once
PERIODIC ORBITS.
P\
P?> and
are
from (44) and the conditions on the j and k that sums of cosines of integral multiples t, the highest
t t
If
it
\+
t.
follows
w% is a
is
sum
of
The
highest multiple of t
l
is
N=k j +
1
l
+kj,+\p +
+\
ll
p +i,
ll
which becomes, as a consequence of the relations to which the exponents and subscripts are subject,
N = n-2+i-jSn-2+j-{-2-j = n.
If
\+
+X
is
vftl
is Therefore, in this case integral multiples of t. of integral multiples of t; and it is shown, precisely as before, that the highest is a sum of cosines multiple is n. Hence the general conclusion is that
m* is also, Pf
of integral multiples of t, the highest multiple being n. By a similar discussion it can be proved that Qn is a
sum
of sines of
(45) can
t,
Hence equations
)
be
n)
]cosr
(46)
+^rcos2r+
ii>n+2pn = -2c<"- cosT+[2c<"l) 1)
+
where the
<B)
sin2r+
.
+^
known
n)
sinnr,
Af
and the
Bf
= 0,
n) are
constants.
is
The
first
Wn =
- 2 Pn +c[ - 2 g-
sin t
- [2 ef
+{">] cos r 1 B
cos 2r
(47)
_
where
n)
c{
is
an undetermined constant.
first of (46), it is
On
Pn
found that
+ Pa = - 2 cf-
sin
+A - 2 B?] cos r
(48)
4-[<>-|5r ]cos2r+
In order that the solution of this equation shall be periodic, the conditions
c?-"
- 0,
2c<"-"
(49)
n_1)
must be imposed.
371
is
of the
form
(50)
= c^ sin r +
C + cf
2
cos r
+ of cos 2t+
+a
cos nr,
<>=
[A <">
R""~l
r-_
uA.
2 B)"}
(51)
If
equations (49), (50), and (51) are substituted in (47), the result
is
[2o{"
+|^ ]cos2t
)
>
[2af
]cosir-
-^ar' + i^rJcosnT.
In order that the solution of this equation shall be periodic, must contain no constant term; whence
its
right
member
c?=-\A?.
Then
the integral of (52) satisfying the condition
) ,
(53)
w(0) =0
is
-r ]sinr
4[2ar+jBr]sinir
I[ 2a r + ^r]cosnr.
(54)
-|[2C + |Br]sin2r
The
2 <f
+ A - Bf
A/
+ af cosjr + <C COS flT, + sin r + & sin 2r + +/3l sin nr,
n>
-w_
.
[j
21?/]
,.
(55)
_ a?-2b? ~2
'
M'
fCf-D
n)
cr=o,
where cT and cf are as yet undetermined constants.
372
PERIODIC ORBITS.
Since the results expressed in the first two equations of this set are identical in properties with the equations (44), with which the discussion of
the general step was started, and since the properties of (44) were fulfilled The for the subscripts 2, 3, and 4, it follows that the induction is complete. process of integration can be carried as far as may be desired.
and that w(0) = 0. Solutions satisfying these properties and p(0) = were known to exist from the existence discussion, and therefore they could certainly be found because the assumed properties are included
solutions are periodic
It appears in the construction that in those of the symmetrical orbits. Therefore all periodic the hypotheses adopted imply also that p(0) = 0. the under discussion which are orbits of type expansible as power series in
It can be shown by direct consideration of the series are symmetrical orbits. no others there are that expansible in any fractional powers of m.
The
differential equations
it is
admit
integral not involving the independent variable. integral to the variables used in this chapter, it is
Upon
found without
that
its explicit
2
form
is
2
p*+(l+p) (l+w)
-m
(l+p)
2(1
+m)\2
(56)
l+P
-|(l+p)cos(r
^)}=-C,
where C is the constant of integration. It will be shown that this integral can be used as a searching test on the accuracy of the computations of the solutions, or to replace the second differential equation of (7).
Since the periodic solutions are developable as power series in m, the and written in the form integral can be expanded as a power series in
Fo+Fxipj, pJf
=C.
(57)
all
In the
Fn
is n.
Fn
separately
F(pj ,Pj,Wj,Wj;
It follows
t)
=C
(58)
(55) and (56) that n is a sum of cosines of of t. In F the sum of the products of the exponents and integral multiples subscripts of the factors of any term not involving cosjr or sinjr can not
exceed n; and in any term involving cos jr or sinjr the sum can not exceed ni. Therefore the highest multiple of r in F is n, and (58) can be written in the form
Fn = y?+y? cosT+
)
+y
C osjr+
+y?coanT = Cn
(59)
373
an identity
in t,
it
follows that
*
fulfillment iox
J?~Cmj
of
7r=0
. . .
(j
= l,...,n).
. .
(60)
a,
J*;
^\
j= 1,
U and on all
It will
of the computations.
If (56) is
f>j,
wj}
Wj] t),
(61)
where Gn is a polynomial in ph ph wJ} and ivj and involves r only in sines and cosines. Moreover, the greatest value of j in Gn is n 2. Suppose that p2 w_ 2 are entirely known, and that pB _ t and wB _! p_ 2 w2 " -1) are known except for the undetermined coefficients c2 it will be shown that equations (60) and the third and fourth equations of (55) define the ajB) and /3] n) uniquely. It follows from the properties of Fn and equation (61) that this function can be written in the form
,
. . .
Fn = [4
C + CT] +
[4c2
cos jr
+
Consequently equations
(60)
+[4a B +2w/3B
become
n>
")
+C
n>
]cosnr.
7r =4c2
,
n - i)
(62)
7r=4ar+2^r+c )= o
It follows
0-2,
>.
from equations
4c*
,
(55) that
- ,)
-2Aj" +4fl
,
w=
0,
I
4ar + 2i/3<''
+ | Br =
J
(63)
Ci-2,
,).
Upon comparing
it is
found that
..'.., n).
2B?~jC*
(n
= 2, ...
oo
j=l,
(64)
a? =
-\A?
Al df =
)
~ Cl
,
of'"
=0
(i=2,
.
...,),
(65)
, /l
^-
ur-cfi
(/-I)
flW
_ 4A,"-(iM-3)c r
'
2i(/-l)
-*.;
374
PERIODIC ORBITS.
These equations express the coefficients, which are determined at this step of the integration, uniquely in terms of constants which depend only upon the In practical computation it is first equation of (7) and upon the integral.
more convenient to make the determination of the coefficients depend upon and Cjn) than upon the A and Bf\ for the former have many the A terms in common, except for numerical multipliers, and both are coefficients of cosine series, which are easier to check than are the sine series on which n) the Bj depend. But the chances of error in lengthy computations are so
great that
safe
if
made
of equations (64). In order to illustrate the process the expression for Ft will be developed. It is found from equation (5G) that
method is to use both the second equation what is the same thing, to secure the fulfillment
i>\
(66)
.
member
by means
equations of (43),
found that
f r?
r,
S Mv - 1 MYl cos 2t +
4a "' + 6#
4>
+ f M^
2
77
2 t7
-|M
]cos4t.
(62),
C? = +c?-'gMr? = 2B?,
2Cj
-f -f u-i! Af V-i M^
2
= 2Bf,
(67)
3C=+fW- f^M-n
4C = - ^
o
rf 8
= 2Bf,
- ? M\ = 2Bf,
375
The
from equation
Solutions as Functions of the Jacobian Constant. It follows (57) that when the periodic solution is given, the constant C is
uniquely defined.
The
relation
between
C and
integral, expressed in terms of the variables in more ordinary If the origin is taken at the center of gravity of the use, will be found. system, the differential equations of motion in rectangular coordinates are
it is
when
X =-r
>
dx
1
-=
'
dy
1
u=
rt
>
n
(68)
V(x-x y+(y-y y,
These equations admit the integral
x'
2
= V(x-x,y+(y-y,y
+y' -2N(xy'-yx')=2U-C
(69)
.
where
is
is
defined in (3)
The
relation
between
C and C
The
r
variables x
and
v of (1),
by the equations
x = rcosv
AcosNt,
y = rs\nv
^AsinNt;
from which
x'
2
it
follows that
2
'2
v'
(70)
2
.
mi+rrii
Upon making
found that
(3)
and
(5), it is easily
k2 m, = n2 a 3 =A\+mya?,
/2 =
a2
Fm,= N A -n a
2 3
=N A
2
m 2p2
2
2
'
r v'
-2Nr v'=^{l+p)
2
2
2 2 2 2 {l+w) -a N {l+p)
2m
2
AN rcos(v-Nt)=-^-aAN
.
(l+ P )cos(T+w),
(71)
k mi
m _ N {l+m) a
2 2
rt
m (l +p)
2
Upon
AN
*
( 72)
376
It
in
is
if
PERIODIC ORBITS.
seen from (56) that when C is expanded as a power series in m, or a/A is eliminated by the last equation of (6), it starts off with a
is
mv
term which
independent of m.
Therefore
l/s the integral in the ordinary form, is expansible as a power series in = 0. The three periodic orbits, of which two are and is infinite for
complex for \m\ sufficiently small, corresponding to the three determinations = or C = oo Since the coordinates of a/ in (6), coincide and branch at
mVi
C
.
and
1/3
is
an analytic
is
function of
One branch-point
at
In the special problem treated by Darwin,* in which the ratio of is 10 to 1, he found by computation in the case of the orbits around the smaller finite mass that there is another branch-point for a certain value of C at which the complex orbits first become real and coincident, and then real and distinct.
.
C = oo
Lunar Theory. In the development of the have been so treated that the resulting expression for the distance, or its reciprocal, is a sum of terms which involve the time only under the cosine and sine functions. The longitude involves terms of the same type and the time multiplied by a constant factor. Considering the problem in the plane of the ecliptic, there are terms whose period is equal to one-half the synodic period of the moon. They are known as the variational terms. Now the period of the periodic orbits which have been found above is the synodic period of the revolution of the infinitesimal
190. Applications to the
differential equations
body, or twice that of the variational terms. The terms of the solutions which are of even degree in n have the period of half the synodical period of revolution. The variational terms in fact belong to the class of periodic
with the work of Delaunay was made in the Transactions of the American Mathematical Society, vol. VII (1906), p. 562, and perfect agreement was found except in the coefficients of the higher powers of m, where errors are almost unavoidable in Delaunay's complicated method. Hill wrote a remarkable series of papers on the Lunar Theory in the American Journal of Mathematics, vol. I (1878), in which he proposed to start from the variational orbit, instead of from an ellipse, as an intermediate orbit for the determination of the motion of the moon. The elliptic orbit as an intermediate orbit came down from Newton and his successors, and the inertia of the human mind is such that it was retained for over a century
orbits treated here.
The
detailed comparison,
up
to
in spite of the fact that it has little to recommend it. Hill has the great honor of initiating a new movement which, it seems certain, will be of the
highest importance.
*Acla Mathematica, vol.
XXI
377
The results obtained by Hill are coextensive with those given here if = and = 1 in the latter series. The method employed by Hill we put
77
was
from that of this chapter. It was convenient in pracbut its can not easily be established. The same method was tice, validity extended by Brown to include terms which contain as a factor* to the
entirely different
A comparison of the results obtained by first, second, and third degrees. the methods of this paper with those of several writers on the Lunar Theory, especially in the coefficient of a/A which converges most slowly, will be found in the Transactions of the American Mathematical Society, vol. VII
(1906), p. 569.
Darwin's Periodic Orbits. In Darwin's compuIt is tations,f the ratio of the masses of the finite bodies was ten to one. found from the definition of and the last equation of (6) that for the motion around the smaller of the finite bodies
191. Applications to
77
m^+m*
ii'
Kmt+mJ
\n)
of the
\l+m)
Darwin defined
his orbits
by the value
their periods were found from the detailed computations. In with his work it is simpler to take the periods which he obtained
the orbits from equation (43). The comparison will be made first with his "Satellite A" for the Jacobian constant in his notation equal to 40.5, loc. cit.,
p.
199.
to be 61
360.
Therefore
m= 6L|3 =0 17051>
1/3
(l)
(_^_)
.12449.
(74)
The m for this orbit is more than twice that occurring in the Lunar Theory. With these values of the constants substituted in the series of 186, it is
found that
r
= 0.12427+0.00652
-0.00006 cos4t+
(75)
The infinitesimal body is in a line with the finite bodies and between them when t = 0. The value of r at this time is found from (75) to be The corresponding value given by Darwin is 0.1265. The r(0) =0.12657. infinitesimal body is in opposition at t = t, and it is found from (75) that
r(ir)
=0.11345.
Darwin's value
is
0.1135.
the "Satellites
A"
*American Journal of Mathematics, vol. XIV (1891), pp. 140-160. ]Acta Mathemalica, vol. XXI (1887), pp. 99-242.
378
PERIODIC ORBITS.
In a retrograde orbit having the same sidereal period, the expression
for
is
m
In this case
n+N l+N/n
613.83
{7b)
N
n
therefore
61.383+360'
m= -0.12715.
The value
of
a/A
is
the
same as
before,
r
and the
which are seen to converge somewhat more rapidly than the series of (95). No retrograde orbits were computed by Darwin in his memoir in the Acta
Mathematica.
Comparison
this case
will also
In
*,-10,
.-!,
1-1,
154
t-^-fr
is
The
40.0.
The period
given
by Darwin
(loc. tit. p.
13'.
Therefore
m = ^||p = 0.42838,
With these values
r
= 0.43404.
-|
(78)
cos
3r+
(79)
From
found that
r(0)
= 0.42759,
r(x)
= 0.41415.
Darwin's results in the respective cases were r(0) = 0.423 and r(ir) =0.4140. The agreement of these results shows the identity of his "Planets A" and the orbits covered by the analysis of this chapter.
Chapter
The
XII.
preceding chapter was devoted to the considbody subject to the attraction of two
circles. The periodic orbits whose existence inclose only one of the finite bodies, and they are more nearly circular the smaller their dimensions and the shorter their periods. The present chapter also will be devoted to the consideration of the
motion of an infinitesimal body subject to the attraction of two finite bodies which revolve in circles; but the periodic orbits now under discussion inclose both of the finite bodies and are more nearly circular the larger their dimensions and the longer their periods. There are three families of orbits of this class in which the motion is direct, and three in which it is retrograde. For small values of the parameter in terms of which the solutions are developed, only one family each of the direct and of the retrograde orbits is real.
The mode of treatment of the problem of this chapter is similar to that of the preceding. A certain parameter naturally enters the problem. When n is zero, the problem reduces to that of two bodies, which admits a
fj.
The existence of the analytic continucircular orbit as a periodic solution. to the ation of this orbit with respect parameter n is proved, and direct
of constructing the solutions are developed. It is shown also how the integral can be used as a check on the computations, or as a substitute for one of the differential equations in the construction of the solutions.
methods
The
Lunar
results of the preceding chapter were directly applicable to the Theory; those of this chapter have no direct bearing on the practical
problems of the solar system, at least as they are at present treated. Their chief value at present is that they cover a part of the field of the problem of three bodies in which one is infinitesimal and in which the finite bodies
revolve in
193.
circles.
Differential Equations. Let the origin of coordinates be at the center of gravity of the finite bodies m,y and ra2 and take the x?/-plane as the plane of their motion. Suppose the infinitesimal body moves in the
,
The
Let the coordinates of m,, x x and (x, y) respectively. ( i> yd> ( i> 2/2) motion for the infinitesimal body are
z?/-plane.
i
and the
infinitesimal
body be
Then
d?x__dU
d?
rl
d?y_
= dU
dy'
l
TT
dx'
1
df
'
r,
(1)
= V{x-x y-\-{y-y Y,
379
r^Vix-xtf+iy-ytf
380
PERIODIC ORBITS.
(2)
Then,
in polar coordinates,
equations
'
(1)
become
dt
2
<Tr
ae
/dvV
\dtJ
dU
dr
<Fv
in ^ dt dt~
drdv
dU
(1)
r dv
w
and
(2) it
The
is
potential function
U will now
be developed.
From
found that
^^
d*
2
^lll [1+3cos2(v _
2
(4)
2;1)]
become
r
2
UJ +
"
m.+m, rU L 4
1+dC0S
1
yJ J
+ fil
r
2
~ fia
[3cos(>-t0+5cos3(>-t; )]+
}> (5)
<fc
+Z dtdt~
m +m
1
r*
1
\*
8mZ {V
l
Vl)
If
+ ~ -^~* [sm(v-v )+5smZ(v-v )]+ assumed to the orbits of m and m are circles, which
i
1
is
be the case,
equations
(1)
'
>-**
vx
w
(7)
mean motion of the finite bodies and that In polar coordinates the integral becomes
r
^n^tQ.
where the primes indicate derivatives with respect to t. When the right members of (5) are put equal to admit the particular solution
r
= a,
),
(8)
where n is the angular velocity of the infinitesimal body in its orbit and t is an It will be supposed that n is given by the observations, arbitrary constant. or that its value is assumed, and that a is determined by the second equation of (8). The constant a has three values, only one of which is real.
381
p.
and
will
be introduced
the equations
v
= a(l+p),
= n(t-Q+d,
(6), (8),
(n-n){t-Q = r,
(9)
-M
7~^h= M &
-
It follows
from
and
that
f
and equations
(5)
-m
2/3
;
(10)
become
i--<+')[T^+*]'+ft^?(W-
+SSfe)[
+
3cos(T
(1+p)s+2 ;[_JL_ + a] = _
+ s)+5cos3(r+8) ]+
<
>
_^
y<
ll
sin2(T+9)
^[-Dfr++8A,3(r+)]+
letters indicate derivatives
},
r.
with respect to
These
equations are valid for the determination of the motion of the infinitesimal body provided |m|<1. The right members of equations (ll) involve only
sines respectively of integral multiples of t+0. The parts in the 2/3 the coefficients of even powers brackets proceed according to powers of p
cosines
2/3
and
and second equations being cosines and sines respectively p of even multiples of t, and the coefficients of odd powers of p 2/3 being cosines and sines respectively of odd multiples of r.
of
in the first
= 0, 6 = 7
Suppose p = /3, p = 0,
at r
= 0, and
let
P=/(&t;t),
= <p(fi,y,T).
(12)
Now make
the transformation
P
= Pi,
6=~6 lt
T=-T^
(13)
The
resulting equations have precisely the form (11). Consequently their solutions with the initial conditions p l = (3, ^ = 0, 6 = 0, 6 x = y are
Pi=KP,y,T )=K(3,y-T)=p,
1
ei
= <p(P,y,T )=<p((3,y;-T) =
1
-d. (14)
Therefore, with these initial conditions, p is an even function of r, and 6 is an odd function of t. The orbit is symmetrical with respect to the p-axis
Such an
382
PERIODIC ORBITS.
consider the conditions for a closed symmetrical orbit. Since the involve sines and cosines of members of only (11) integral multiples right of t, sufficient conditions that in symmetrical orbits p and shall be periodic
Now
2jir
are
P-/(fc y;
h) =o,
(15)
and these conditions are necessary, provided they are distinct. In order to examine the solutions of (15), it is convenient to use parameters other than /3 and y. Suppose that, at r = 0,
r
= a(l+p)=a(l+a)(l
M
l-/i
e),
= a'p = 0,
(16)
+ a_
.
Vl+e
*~l-/*(l+a)*(l-e)*
It follows that
elliptic orbit
a(l+a) and e are the major semi-axis and eccentricity of the which would be obtained if the right members of equations
Because of the well-known properties of the solutions of the two-body problem in terms of these elements, the properties of the general solutions, so far as they do not depend upon the right members of These properties will be important in solving the conditions (11), are known.
(11)
were zero.
p = 0, p = 0, p = 0, = 0, = for all values of t. It follows that the moduli of a, e, and p I/3 can be taken so small that the solutions will be regular while t runs through any finite preassigned range of values. We shall choose as the interval for r the
in the vicinity of
1/3 :t = 2jt and integrate (11) as power series in a, e, and p range vanishing I/3 with a = e = p = 0. That is, the results will be the analytic continuation with respect to these parameters of the particular solution r = a, v = nt, which The results may be written in the form exists when p = 0.
:
P = p 1 (a,e,p P
1/3
;r),
1/3
= 2>
(a,e,p
;r),
=p 6=p
e,
(a,e,p
(a,
1/3
;T),
t/3
;
e,p
1/3
,
t),
where p u
pt are power
Pi (o,
series in a,
and p
with r in the
coefficients.
The
become
/3
l/3
;
jr)
= 0,
p, (a,
e,
;jt)
- 0.
(18)
It will
series in
be shown that these equations can be solved for a and e as power 1/3 ,/3 1/3 p vanishing with p = 0, which converge if the modulus of p is
,
sufficiently small.
Since the right members of (11) carry p10/3 as a factor, the part of the solution depending on the right members will be divisible by p 10/3 If
.
the right members of (11) were zero and if the solution were formed with the initial conditions (16), the mean angular motion of the infinitesimal
body
in its orbit
would be
383
follows that
PtW^
p3 (jir) =
sin fJTr+esin2vjir+
10/3
|+M
fc(a,
e, ii>
jv)
l/3
= 0,
(20)
-^j^jir-pJT-2esmvJT-
i-\-n
mq (a,e,
3
fx
;JTr)=0,
vjir,
and
observed that the first equation of (20) is divisible by m and the second by n. After dividing by these factors the = = still a are satisfied e n = 0; moreover, the determinant of equations by
Upon
referring to (19),
2
>
it is
their linear
terms in a and
e is
A=
>
=+
fiW0.
(21)
Therefore, besides the solution m = 0, equations (20) have a unique solution for a and e as power series in // /3 vanishing with ju ,/3 = 0, which converge for
,
the modulus of
factor,
ix'
and can be
These power
series carry
i/3
as a
a = M4/3 Pl (Mt/3 ),
= MV3
W
2
/3
)-
(22)
Upon
members
of (17),
which vanish
p=m
The
series
4/3
&(/x
1/3
;
r),
0=m Q
4/3
1/3
(m
t).
(23)
are periodic in t with the period 2j't because the conSince ditions that the solutions shall have this period have been satisfied.
x
Q and Q
sufficiently = 0, it small, and since the expressions for a and e given in (22) vanish for n 1/3 follows that the modulus of n can be taken so small that the series (23) conif
O^t^jV
the moduli of
a, e,
verge for all t in the interval; and since they are periodic with the period 2Jx, the convergence holds for all finite values of r. The integer j has so far been undetermined. When j is unity, the
When j is greater periodic solutions exist uniquely and their period is 2w. the periodic Since exist the solutions also than unity periodic uniquely.
orbits for j greater than unity include those for j equal to unity, and since in both cases there is precisely one periodic orbit for a given value of nVi it follows that all the symmetrical periodic orbits of the class under consideration
,
2ir
r.
384
It follows
PERIODIC ORBITS.
that t-\-B = Since in the periodic x is periodic with the period 2v, the period of the solution is the solution synodic period of the three bodies. Hence, if the motion of the infinitesimal
from
(6)
and
(9)
vv
is referred to a set of axes having their origin at the center of gravity of the system and rotating in the direction of motion of the finite bodies at the angular rate at which they move, and if the z-axis passes through the
body
then the periodic orbit of the infinitesimal body, which has been proved to exist, will be symmetrical with respect to the z-axis. Since, by hypothesis, a > R, it follows from (6) and (8) that n^ > n. Therefore, even if the motion of the infinitesimal body is forward with respect to fixed axes, it is retrograde with respect to the rotating axes.
finite bodies,
supposed that the period of the finite bodies, and therefore n is advance and remains fixed. The variation of the parameter /i1/a corresponds to a variation of the period of the infinitesimal body defined by n. If the motion with respect to fixed axes is forward, n has the same 1/3 sign as r^, and /x has three values, one of which is real and positive while the other two are complex. If the motion is retrograde, /tV3 has three different values, one of which is real and negative while the other two are complex. Therefore, for a given period, there are six symmetrical orbits, three direct and three retrograde; and for small fx one direct orbit is real and one retrograde orbit is real, while in the others the coordinates are complex. This means, of course, that the corresponding solutions do not exist in the
It is
given in
physical problem. The coordinates of the complex orbits are conjugate in For a certain value of /x 1/3 they may become equal, and therefore pairs.
real,
1/3 real and distinct. and, for larger values of /* Upon transforming the integral (7) by (9), it is found that
,
a-X+p)^
^^
}-(?
,
.
(24)
+5cos3(t+0)]+
2 2 It follows from this equation and (23) that Cj where C = n\(l m) o Cican be expanded as a power series in ///3 vanishing with ju 1/3 The term of 1/3 m Vi lowest degree in n after substituting (23), is n Therefore, n can be 3 For Ci = 0, the three branches of the expanded as a power series in C/ Vs function are the same. Since a = R/n the relation between C and (7, is
, 1
.
nun
.
(25)
From
this
it
follows that
,
C=oo
for
I/3
ju
= 0.
branch at
C = oo and
385
has been
proved that the symmetrical periodic solutions under discussion are expressible in the form
,/3
,
d=2e,n (=4
r. it
t/3
,
(26)
and
0,
|
are functions of
Since these series are periodic and follows that each p, and d, separately
periodic; that
is,
P( (r+27r)
= Pl (r),
0,(t+2tt)
m8&)
(27)
In every closed orbit there are points at which dp/dr = 0. Suppose t of (9) is so determined that this condition is satisfied at r = 0; it will follow from this and the convergence of (26) for all |m1/3 sufficiently small that
|
= Pf
(i=i,
...
co).
(28)
In the symmetrical periodic orbits the value of 6 is zero at t = 0. But this condition will not be imposed, because the general periodic orbits, whose initial conditions are not specialized, include those which are symmetrical;
and
in the construction it will appear that the conditions for symmetry are a consequence of those for periodicity. Hence all the periodic orbits of the
class
arranged as power series in p. the coefficients of the several powers of /x1/3 set equal to zero. The coefficients of nm set equal to zero give the equations
m and
,
(11),
P4 = 0,
= O. =&
(29)
The
(30)
The
coefficients of
5/3
,
.
are the
subscripts,
and
their
= aj,
6j
=b
(31)
where
all
the
a,
and
b s are so far
undetermined constants.
The
coefficients of
10/3
p I0
= 3a
-jM[l+3cos2r],
first of
=-
|Msin2r.
(32)
a,=
\M
4
must be imposed, which uniquely determines the constant a solution of (32) satisfying (27) and (28) is
Pio
Then
the
=a
10
+^Mcos2r,
lo
=6
lo
+|Msin2r,
(34)
where a 10 and
386
PERIODIC ORBITS.
The
coefficients of
11/3
/x
are defined
by the equations
(?,i
Pn = 3a 6
= 0,
flu
from which
it
follows that
a6 = 0,
fti-Ou,
= 6,1,
(35)
where a n and b n are as yet undetermined. The coefficients of m12/3 in the solutions satisfy the equations
Upon imposing
it is
found that
J'
a,
= 0, and
" ftt-Oag"
L 2(mj 2(,Wi+/as)
,
^ r3cosT+|cos3rl,
nh)
'9
,
- 3Af(mi
ms)r
L
(36)
mi+m,
sinr+ ? sm3rj,
where a I2 and 6 12 remain so far undetermined. In a similar way it is found from the coefficients of nli/* that
a 7 = 0,
p I3 = a 13 +|Mcos2T,
0i 3
=6
13
+|jMsin2T,
(37)
where a 13 and b n remain undetermined at this step. So far all the b} have remained arbitrary, and it is necessary to carry the The integration one step further in order to see how they are determined.
coefficients of
li/3
are defined
2
by
4
Pi4
cos2T],
(38)
fli4=
-^flio
3-M" [6 4 cos2t
2a 4 sin2-r].
1O
Upon
and
from
(33)
and
(34),
imposing the
conditions (27)
and
Pi4
integrating,
it is
found that o8 =
2
t^M*, and
fli4
|64-Wsin2T
^Af
2
cos2r,
(39)
sin 2r.
flu
= 6,4 + f&4Mcos2r - ^ M = 14
14
The
= 0,
in
Pl4
=a
=6
14
-|M sin2r.
15/8
(40)
It is
found
a similar
the coefficients of
ju
that b6 = 0, and
'-"'-^SftH'+l
9 ''= 6
83 *]'
(41)
'>+i
J,f
SfS[98inr+ S
sin3
where a 15 and
6, 6
387
cients of the 6j
coefficients of the p t are cosines of integral multiples of t, and that the coeffi, except for the undetermined additive constants, are sines of
t.
integral multiples of
In the computation of
6
,
pj
and the condition p = at t = has a,_ = required that &,_ 10 0. It will now be shown that these properties are have been computed and that general. Suppose p . . , , p; $4t . a n which enter additively the coefficients are all known except a_ 6 in p_ 4 b which enter additively in p respectively, and &_ 9
have uniquely determined
4
,
0_
of p n+1
and
The
determination
PB+l
= 3a n _
+|M6 _
B
sin2r+Fn+1 (r),
B+l
= Gn+1 (r)
(42)
where
FB+1 (t)
and
(jb+1 (t)
are entirely
.
that
+1 (t) is
sum
of sines
r. It follows from and the properties of p; sum of cosines of integral multiples of t, and Hence they may be of integral multiples of t.
known
4,
functions of
. . .
cos jr,
(?n+1 (r)
=2
first
3a n _ 6 +^r +I>
=
.
(43)
this condition
first
equation
P +1
= a n+1
-|M6_
at t
sin2r+S a^"
makes
it
cos jr,
a**
- -kA*+.
J
(44)
The
condition p =
necessary to take
9
&- = 0.
Then
and
p B+1
is
(45)
,
it is
The
sum
+i
ff+
-B?
+i)
.
(46)
a sum of sines of integral multiples of r, except for the undewhich must be put equal to zero in order to satisfy the condition on p +u These results lead, by induction, to the conclusion
Hence 0+i
is
termined constant 6 B+
that the p; and
0/
i ,
= 4,
oo) are
sums
of cosines
and
sines respectively
of integral multiples of t
whose
coefficients are
uniquely determined.
388
PERIODIC ORBITS.
the properties of the solutions which have just been established, Therefore these at t = 0, but also 0(0) = 0. it follows that not only is p = periodic orbits are the symmetrical orbits whose existence was established
From
in
194.
In the construction
it
orbits
were
symmetrical, and since this property is a necessary consequence of the periodicity conditions, it follows that all periodic solutions which are expansible as
power
series in
,/3
are symmetrical.
It is easily
shown, by direct
consideration of the construction of periodic solutions, that they can not be 1/3 expanded as power series in p except when,;' is a multiple of 3, and that then
The differential equations 196. Application of the Integral. integral (24), which, for brevity, can be written in the form
F(p,p,M,T,/z )=0.
It follows
,/3
admit the
of this equation
from the form of (24) and the expansions (26) that the left member can be developed as a power series in p} /3 giving
,
F = F,+F^+F^+
Since the
p,
+V+
1'
,
=0.
(47)
and 9, are sums of cosines and sines respectively of integral multiples of t, and since p enters in (24) only in the second degree and only in even degrees, it follows that the Fj are sums of cosines of integral multiples 3 whence of t. Equation (47) is an identity in p.
FB = SCfcos;V=0
Since these equations hold for
C<"'
(n-0, ...
follows that
;
oo).
all
values of
t, it
oo
(n-0, ...
. . .
j=o, ...
. . .
,).
n)
(48)
.
The Cf
Hence # af and #\ f, on the formulas can be used as check computation of the equations (48)
are functions of the
, ,
Equations (48) can be used in place of the second equation of (11) for B) the determination of the j8j the coefficients of the trigonometric terms in 6 n - have been the expression for 0. Suppose p4 p- and 4 It follows from determined except for additive constants in p_ 6 pn - t
, ,
. . .
.
(24) that
Fn
is
Fn = -26 + P n (P
n
j,
Pj ,
eh
Ot)
(i=4,
n-l),
where
Consequently
Cr--2j^+D(ar,/5T)-0
which uniquely determine the #"\
Chapter
XIII.
A CLASS OF
PERIODIC ORBITS OF
197. Introduction.
The
restricted
naturally the starting-point* for the consideration of the periodic orbits of an infinitesimal body, or particle, which is subject to the Newtonian
attraction of certain finite spheres whose motion is supposed to be known. The two finite bodies are supposed to revolve in circles about their common
center of mass, and the motion of the particle is restricted to the plane in which the finite bodies move. One class of orbits occurring in this problem is that in which the particle revolves about one of the finite bodies, and for
it is
particle to a plane rotating with the angular velocity of the finite bodies. All of the known periodic orbits of this type possess one and only one line of symmetry, namely, the line joining the finite bodies, and this property
of
and the
The purpose
For the analysis is, the motion of the spheres is prescribed. which follows the nature of the forces producing this motion is unimportant.
of the time, that
The
law.
spheres are supposed to attract the particle according to the Newtonian Besides involving additional terms in the disturbing function, this
generalization modifies the original problem by introducing cases where the periodic orbits have no line of symmetry, and cases where there are more lines of symmetry than one. This modification necessitates some
changes in the details of the analysis which must be worked out. In order to avoid cumbersome notation, the analysis will be developed for simple particular cases of the motion of spheres under their Newtonian attraction;
applicable to more general types of prescribed motion of the finite bodies, which are indicated in 207.
it
is
*See papers by
Hill,
American Journal of Mathematics, vol. 1 (1878), p. 245; Darwin, Ada Maihemalica, and Moulton, Transactions of the American Mathematical Society, vol. 7 (1906), p. 537.
389
390
PERIODIC ORBITS.
198. Existence of Periodic Orbits
Having no Line
of
Symmetry.
It
was shown by Lagrange* that an equilateral triangle is a possible configuration for three spheres revolving in circles about their common center of mass. This motion of three finite bodies will serve to illustrate the case
periodic orbits of the particle about one of the bodies possess no Let the masses of the three finite bodies moving accordline of symmetry.
when the
ing to the equilateral-triangle solution be denoted pose the particle P revolves about the mass
are unequal. f as origin and an axis having a fixed direction in With reference to and be respectively (Rx FJ, t space, let the polar coordinates of Af,, coordinates of the The bodies are expressed in terms and (r, v). (Rt Vt),
t
by
M M M
,
lt
and sup-
Suppose
of the time,
t,
as follows:
R^R^A,
where
F.-F.-f-M,
(1)
Here N denotes the angular velocity, A the length of a side and fc is a constant depending upon the units employed. The differential equations of motion of P are
2
of the triangle,
dt
\dt)
~dr'
d?
+ Z dt dt~
2
rdv'
(Z)
where
(3)
1 ),
i ).
Let us define
m and
a by the relations
mv = N,
where
v is
vW = k M,
2
(4)
a quantity to be assigned later. By the substitution v = w-\- Vx = w Nt the motion is referred to an axis rotating with the angular velocity of the finite bodies and passing and factors depending upon the units employed are t always through
vt
= r.
On making
these substitutions in
Prize memoir, Essai sur le Problbne des Trow Corps, 1772; Coll. Works, vol. 6, p. 229. the periodic orbit of P about lias a line of symmetry, namely, the median of the triangle from the vertex M, which is the line joining to the center of mass of the system. For the treatment of this special case it is convenient to make use of the property of symmetry and to employ analysis similar 202 and 203. to that developed in
tlf
M,=Mi
M M
391
can expand Q as a power series in ap/A which is convergent for all = ap is less than A and in all that w provided the distance follows this condition is supposed to be satisfied. The expansion has the
values of
We
MP
form
Qsb
VM;1
+ Kl p ) [l
+ 3cos2t4+i(i p ) {3cosw+5cos3wU
,
+^t+Kf#+3<--f)}
Let
+Klp)'{s~*(-f)+s<3(-f)j+
\
and X2 be defined by the
relations
M.^CAft+M,),
From
equations
(1)
Af.-X^+A/,).
(6)
we have
kXMt+M,) _ ~
M +M M+M +M, yt
1
Then, on setting
M +M M +M +M
l
jr A
t
'
it
members
of equations (5)
y2
(7)
v
2
ap dw
t
iiCm
p|
+|x sin2(;-|)+|x,Qp){sin(M;-|)+58in3;(w-|)}+
It is
(5)
by
the relations
m=p,
X,
= Xp,
T"**
8)
wherever the degree of a/A is higher than the first. The quantities X and r? are numerical constants. By relating X2 and p the existence proof is made to depend only upon general properties and certain terms of the differential equations which involve \ that is, upon terms in the disturbing function which are due to the body M\ We shall consider the solution of equations
;
.
parameter p. The differential equations, and consequently also the solution, do not represent the physical problem under consideration for any value of the parameter except the one satisfying the But if the solution is valid when this particular numerical relations (8).
(5) as
power
series in the
value of p
is
substituted, then
it is
392
PERIODIC ORBITS.
representing the physical problem and therefore has a physical interpreThe generalization of the parameter a/A is merely for convenience tation.
expressions in the equations which determine the coefficients at the various steps in the solution. On introducing the parameter p as indicated, equations (5) become
in
having
finite
where
f= Kp\\\^(l+3cos2w)+^^ p(3cosw+5cos3w)i+pF(p,p.,cosjw,smjw)\>
g=
where
are functions of the indicated arguments. Equations (9) are periodic in w with the period 2ir and do not involve t
and
explicitly.
Suppose that
P = &(t),
w = ^{t)
is
a solution.
the period
2pw (where p
an integer) are
fc(2px)-2p,r=fc(0),
tf(2p>)-tfty),
\f/ 2
where
^ and
\f/' t
When
orbit, is
p=
will
known, namely,
initial
= l,
= 0,
W = T, w = 0,
w'
and the
conditions are
'
p=l,
It will
l.
(12)
be shown by the process of analytic continuation that, for values of n different from zero, but sufficiently small, there exists a periodic solution which, for m = 0, reduces to equations (11). For this purpose we consider
the solution of equations (9) subject to the initial conditions
P-1+&,
where
ft
,
p'=&,
w-A,
tf-l+Aj
p,
(13)
/3,
/33
/3 4
vanishing with
It follows from the p, so that the conditions of periodicity (10) shall hold. differential equations that the solution is expressible as power series in ft ,
ft
that, for sufficiently small values of the parameters, to 2pir. the series are convergent for all values of t from suppose
>
ft
>
ft
>
an d p and
We
that this condition on the moduli of the parameters is satisfied. For the determination of those terms in the series which involve the
initial
it is
two-body
equations
(9)
PARTICLE ATTRACTED BY
ft.
SPHERES.
393
motion of a particle
P when
instead of the additive increments ft ft ft ft it new parameters a, e, 0, <p defined by the relations
is
M alone.
P=l+p
=(l + a)(l-ecosO),
I"
,_o _ ~ Pi ~
Vl+a(l-ecos0)'
"
a = w = r3 arc cos ft
,
COS0
Ll-ecos0j
2
C "| s
arc cos
cos(0 y) ti ecos(0
(14)
<p).
)3/2
By
of
introducing
2
body
ii
.
/x in the fourth of equations (14) it is possible to use the twofor problem determining all terms of the solution which are independent In terms of the parameters a, e, 0, <p the properties of the solution
P
W,
is
is
particle at
t=0.
W W
t
is
the longitude of
perihelion.
W- aacis
Fio.
8.
are well known, and the conditions of periodicity can be easily discussed. The geometric meaning of the angles 0, <p, and t which occur below, is
shown
in Fig. 8.
On making
tions (9),
the substitution
w=u
/jlt
and then
setting n
in equa-
we obtain the
is
two-body problem, of
C0Sjg
t
= (l + a)(l-ecostf),
cos(tt+ x
sin
W -W
x 1
ecosE
l
>
(15)
. >
where
ecosE
is
defined
by the
relation
V2
(l+a)
+0
esin0 =
'
esini?.
394
PERIODIC ORBITS.
s
.
To
and
pV
we
will write
p=l+p,p* + ppV +
W = T-TH+<p + W
n +WH*<p+
On
substituting these expressions in the differential equations, there results for the determination of p, and wt the following set of equations:
^-2^-3p
-57+2^
^(1+3cos2t)+^-|(3cost+5cos3t),
sm2r
=-
^(smr+SsinSr).
is
^ w =K\
i
201
\_(j+3j)T-(4:+
By
a similar computation
it is
shown that
>
V\
T21
288
(Q
333 Ol\
15
,11
45
of
2 p.
relation
w=u
are obtained from equations (15) pt; and the solution becomes
by Taylor's
p=l + a
ecosr
aefcosT
-Tsinr) +e0sinr
+ ae0(sinr
w=t
-Ta+2esinT+^
rp
r cos
r)
+p
+ppV +
'
'
'
t
(16)
3aerCOST
ed(l
s
2cosr)
+wpV+
'
'
'
be periodic, we have
O=-3pxa60+^ P T/*V+
= 3P
(b)
7rae+|-|pV+ "',
(17)
(c)
0=
3pira
2pjrp
6p7rae+
(d)
= 6piraefl-~p7rpV +
PARTICLE ATTRACTED BY
ft.
SPHERES.
a, e, 0,
<p,
395
if
The
and,
independent,
would determine them in terms of li. But the differential equations (9) do not involve r explicitly and hence admit the integral of Jacobi. This
furnishes a relation of the type
F(a,
e, 0, <p, /x)
= constant,
It follows that
if (a), (6),
and equations
In this problem the dynamical interpretation is simple. Since the finite bodies move in circles the origin of time is arbitrary.! The most convenient choice is t = when w = 0, which is equivalent to choosing <p = 0. Consider the solution of equations (a), (6), and (c) for a, e, and 0. The equations have the following properties: This follows from the (I) There are no terms independent of a and n. fact that, in the two-body problem, the period does not depend upon e and 0. (II) There are no terms involving n to the first degree except the one term 2 pirn, which occurs in (c). (III) There are no terms in independent of e, since does not enter
conditions independently of e. It follows from these properties and the particular form of the first terms of the equations that a, e, and
are solved for the three quantities a, e, and in terms of ju and <p results substituted in (d), the equation is satisfied identically in <p.
the
initial
by the
following steps
m[
3+
+ function
(fx, e,
i)\
divided out.
This value of a when substituted in (6) permits a factor /z to be We can then solve the result for e as a power series in /x and which contains /tasa factor, and obtain
(2)
+ function (m,0)]. [i6Z~r(3) When the values of a and e are substituted in (a) a factor n* can be divided out and obtained as a power series in n alone, vanishing with fi. thus found in the expressions (4) By the substitution of the value of for e and a, we obtain finally
e==/x
= mPi(m),
= np
2 (n),
= np
(n).
to be convergent for all values of Hence, for a given value of n a, e, 0, and n which are sufficiently small. is to determine it the initial conditions (14) as possible sufficiently small,
are
known
power
shall
series in /x such that the solution of the differential equations be periodic in r with the period 2pir.
(9)
*See Poincarfi, loc. cil., p. 87. the finite bodies do not form a fixed configuration in the rotating plane the integral of Jacobi does not exist and the origin of time is not arbitrary. In this ease it is necessary to determine the four parameters from the conditions of periodicity. The case of the triangular solution when the finite bodies move in ellipses has been treated by Longley in a paper in the Transactions of the American Mathematical Society, vol. 8 (1907), pp. 159-188.
fWhen
396
PERIODIC ORBITS.
When
in t,
the values of
a, e,
is
in
terms of
/*
obtained.
The
where p is an integer, and from the conditions of periodicity (10) it is apparent that the particle makes p revolutions in the rotating plane during a period. The process by which the periodic solution was obtained yields a unique result; therefore, for an assigned value of n, there exists one, and only one, orbit having the period 2pir. Since the orbits having the
include those having the period 2ir, it follows that all the orbits of this analytic type are closed after one synodic revolution* Since t = vt the period of the solution in t is 2v/v, and the quantity v, which is so far arbitrary, can be determined by assigning the period of the
period 2pir,
p> 1,
solution.
then determined by the relation n = m=N/v, that is, the numerical value of /z is the ratio of the mean motions of the If the direction of revolution of the parfinite bodies and of the particle.
The parameter n
is
the same as that of the finite bodies that is, if the orbit is direct and N have the same sign and m is positive; if the orbit is retrograde, v and N have opposite signs and n is negative. Since for an assigned value of n there exists one, and only one, periodic orbit, and since values of ju which are numerically equal, but opposite in sign, give orbits having the same period in t, it follows that for a given period there exist two, and only two, real orbits of the type under consideration. In one the motion is direct, and in the other it is retrograde. We may now state the result as follows The period 2-k/v of the solution
ticle is
v
:
may be assigned arbitrarily in advance, subject only to the condition that the ratio N/v is sufficiently small, where 2ir/N is the period of the motion of the
finite bodies.
Then
particle having the required period. All the orbits of this type are closed after one synodic other it is retrograde.
and only two, real periodic orbits of In one the motion is direct, and in
the the
revolution.
In deriving this conclusion no use was made of the explicit values of The those terms in the disturbing function which are due to the body 2 proof depends entirely upon the form of certain terms of the solution which
involve \. Hence the analysis and conclusions are applicable without change to the case where n finite bodies revolve in circles in such a way as
to
form
Having no Line
of
Symmetry.
It
possible to construct the periodic solutions of the differential equations (9) by the method indicated in the existence proof, but the process is labo-
rious.
will now be given by which the solution to any desired can be conveniently constructed. It is not necessary to determine the initial conditions explicitly in advance, and the computation involves only algebraic processes.
method
number
of terms
Since no
new
by taking
p> 1 we
will
397
has been proved that the periodic solutions are expressible in the
form
(18)
The
series (18) satisfy the differential equations (9) uniformly over a finite interval in p, and hence, when the series are substituted in the differential
equations, the coefficient of each power of p must vanish. Furthermore, the series are periodic with period 2w in r; and, because the periodicity holds for a continuous range of values of p, each coefficient p, and w sepaIt has been shown also that we rately is periodic with the period 2x in r.
t
can choose
that
x
w = when r = 0, and
p, it
follows
for every i. Let the solution (18) be substituted in the differential equations (9) and arrange the results as power series in p. The terms of the first members have the following forms, where the accents indicate derivatives with
w (0) =
respect to r:
Q>
2 ^2 =PiP + P M +P
it
i
"
2
"
3
+
i
+p
I
H+
*
I
(^+m)
= 1+
[ Pl +2(;;+l)]
2
2
p
2
+ [p +2w +2 Pl K+l) + K+l) ]p +[p +2^ +2(w' + l)w' +2p w' +2p (w' + l)+p (w' +iy]
2 3
1
ti.
+
+ 2p
P
1
+[
wJ_,+
Pt
+ 2 Pi _
(w' 1
+ l)
]p
2 2
+
3
p 2 -4p )M
(19)
+
p|j
+(2^-6*.^+
)m'+
3
=<p+
+
+pX) m'+K'+^K+^Om
+(w:+ Pl w'u+
2
+p 1
<v+
^(^+m)=p;m+[p;+p;k+d]m
+[p;+p (^+i)+pX] m
2 3
+
<
+t^+pU(5+r)+
+p>;_J M
398
PERIODIC ORBITS.
first
power
^1+2^=0. dr dr
2
(20)
v
It follows
cos r,
are constants of integration. Since p, and w, are This condition determines periodic, the coefficient of t in w^ must vanish. = Since when t = 0, c"'= 2c"\ the constant cf, namely, c{ = 4/3. l) The constants c and cj are so far undetermined.
where
equating to zero the coefficients of the second power of lowing set of equations is obtained
:
On
p,
the
fol-
Pl
(w1 +l)-3p
2
l
+f
^
tions are explicitly
(22)
+ 2^=- Pl <-2p>;+l)+0
where / and g are obtained from / and g respectively by writing p = 0, w = t, p = l. The second members are known functions of t and the equa-
^-2^-3p =
J
yl
+ (^"+^ )cosr+fc
)
>
sinr
+A?co82t+A? cos3r,
(23)
^? + 2^
On
sin3r.
^+2p,
On
= cr
2)
(24)
first of
^+
Pi
= A?+2c?
(25)
-|i)r)cos3T.
PARTICLE ATTRACTED BY
tt
SPHERES.
399
In order that the solution of equations (25) shall contain no non-periodic term, the coefficients of cost and sinr must vanish; hence c and c"' are determined by the conditions
2c^=AT-2Df,
With
these values of c
2)
{
ef-O.
becomes
2>
"
)
and
2)
p2
= ^; +2cf
cos2T+^
cos3r,
where
substituting this value of p2 in equation (24) and integrating, for w2 a solution of the form
On
we obtain
w = cf - (2 A? +3 ef ) r - 2 c<
t
2)
sin t
+2
2)
c<
cos r
2)
5,
sin 2 r
+ If sin 3 r,
= 2,3).
where
hT=-UDT+2jaT)
Since
is
periodic, c
is
2A? +3c? = 0.
)
Since w, =
when
= 0,
is
expressible in terms of
c,
namely,
cf+2c = 0.
Of the eight constants of integration which have been introduced in the first two steps, five (c, c, c, c, c) have been determined uniquely; c has been expressed uniquely in terms of c; while the remaining two
(c,
c)
By
are
still
arbitrary.
equating to zero the coefficients of the third power of p the following set of equations is obtained:
^-2^-3
?
Pl
= 2( Pl +w[+l)w' +2 Pi (wl+V
t
(26)
^ +2^=-p <-pX-2p;k+d-2p>;+^
1
coefficients of
t,
dw
|^-2^-3p,
^ +(^
)
C;
+A? cosjr+BfsmJT
(27)
^ +2^=(fc+i>r)
?
sinr
+(-f c + c
"
coST
+D?mnJT+C?9t*jr.
400
PERIODIC ORBITS.
The treatment
employed
of equations (27) proceeds by steps similar to those Four new constants of intein the solution of equations (23).
gration are introduced, namely, cj", c\ cf, cf, while cf , cf, uniquely determined by the conditions
and cf are
2c? = A?-2D?,
2c? = B?+2C?,
3ef-
2ilf.
The
ft
+
From
sible
2 (^smJT + yfcosjr)it
w = when r = 0,
3
follows that cf
is
expres-
uniquely in
cf+2cf+2 T f = 0.
The two constants cf and cf are determined in the next step. It can be established by complete induction that the preceding process can be carried as far as is desired. Suppose p l} w pl} wt p^ w _ have the The have been determined by this process. following expressions
t
; ;
. . .
form:
p,
w = y?+ 2
t
(1=1,
2,
i-2),
p1 _ 1 =
1)
oi'-
+c<'- cosT-r-c<
- 1)
,,
i,
i)
i)
+fj5r-
+|cr-
cosr)
+
The constants
l)
2 (^y-^sinir + Tr^osir)-
of integration
l>
cj'~
and
<-1)
cj
in
terms of c""
expressible
- ,) -
ci
r-2cr
i,
2
/si
1)
7;'-
= 0.
401
The equations
p,
and
w,
|*
-2^-3
Pt
= A?+
(^
<<-
+ ^>)cosr + (^cr'+ Br
J
)sin
+ s(^
Hi
cosir
+^
sinir)
\
^ +2^'
The
coefficients
(28)
(^
+ D
S (Dj"sinyr
+ C<
,,
cosir)-
solved
are known constants and equations (28) are A, B, C, the by steps employed in the solution of equations (23) During the process four constants of integration are introduced, namely c"\ c, cf, c,
.
= A? - 2 Z){
2 c*'-" 3 <f
= JBJ" + 2 C,
""
ci
= - 24
(29)
>.
The
p.
u i = c;-f-6
,
sinT
+ 7icosr+
2 (Jsin./T+7j
i)
cos /T),
<
where the
by the formulas
(31)
^^Or+yC),
The formulas
(31) together
,,
Ti
=2 Cr +|5r
,
+|c,
accuracy; the computation is constants of integration entering in the last (i th) step, it is necessary to ,+1) 1+n ,+l) of the next following step. C[ D[ compute the coefficients Af** 5{
t
, ,
with the conditions (29) are sufficient to any desired degree of In order to determine the entirely algebraic.
402
200. Numerical
PERIODIC ORBITS.
1 For the purpose of illustration, we assign in the preceding analysis and construct involved numbers to the constants an orbit. In this and the other numerical examples which occur later, it has
Example
not been shown that the processes are valid for the numerical values which are employed and which have been selected for convenience in graphical It is probable that the series are convergent, although it representation. ha3 not been found possible to determine the true radii of convergence.
The
differential equations of
(5).
m = n and
members
explicitly as far as
On
we have
/>
3COSM,+5COS3M;
+^(^-)VJ9+20cos2w+35cos4w|+
}
(32)
n dp/dw
JfMipfs
+ f jp[sin;+5sin3wj+^( j)V{2sin2^+7sin4w|+
+H(i)V{*! (-|)+^*(y-|)}+";J
We select M for the unit of mass and suppose M = 10,
x
5. For the unit t the distance between the finite bodies, that is, A = i; = 1. The period of the solution is and the unit of time is selected so that = assigned so that v 5, whence
M=
of distance
we take
N
v
403
The constant k
is
whence
&2 = 0.06250,
k 2 M,= 0.62500, k2
M = 1.56250m.
2
defined
by
ai = k2 M, whence a = 0.67860m-
==
With
become
+ (0.78125-1. 17188 cos 2w+ 2.02977 sin 2w)pm + (0.47714 cos w+ 0.79523 cos Sw) pV + (0.59643 cos w + 1 .03308 sin w - 1 .98808 cos 3w)pV
3
+ (0.16188+0.35974cos2w;+0.62954cos4w)pV+
(33)
= p^f +2^(^ + M)
-(0.93750sin2M;)pM
+ (1.17188sin2w+2.02977cos2w)pp
- (0.15905 sinw+0.79523 sin 3w)pV
+ (-0.19881sinw+0.34436costf>+1.98808sin3w)py
- (0.17987 sin 2m; +0.62954 sin 4 w)pVH
The
p
periodic solution of equations (33)
is
=l-| M +(0.45139+0.39762cosr- 0.62500 cos2t)m + - 0.47647 + 1 .04542 cos r +0.86090 sin r + 0.46875 cos 2r -1.35318sin2r-0.16567cos3r)/x +
(
'
(34)
w = r+(-0.79524sinT+0.85938sin2T)M
+ (0.13882-3.25720sinr+1.72180cosr+0.27995sin2r
-1.86062cos2T+0.19881sin3T)p H
3
404
PERIODIC ORBITS.
= Substituting the numerical value m 0.2, equations (34), if convergent, The orbit is shown in Fig. 9. are the equations of motion of the particle P. In this and the figures of the following numerical examples the comparison circles are not the circular orbits which have been called the undisturbed
The undisturbed orbits are referred to fixed axes while the drawThe comparison circles ings are made with reference to rotating axes. represent orbits in which the particle would make a complete revolution
orbits.
-Mt
with respect to the rotating axes during the period. The points which are numbered 1, 2, 8 represent positions of the particle in the periodic The corresponding positions in the comparison orbit at intervals of t = x/4.
.
by the numbers
1', 2',
8'.
201.
Some
Problem
of
n Bodies.
The
existence of symmetrical periodic orbits of the particle depends upon the masses and motion of the finite bodies. So far as the analysis is concerned,
be arbitrarily periodic, without reference to the nature of the forces producing it. It is required only that the motion of the finite bodies shall be known and that they shall attract the particle according to the Newtonian law. It will be interesting, however, in developing the analysis, to prescribe motion for the finite bodies, which is possible under the law of the inverse square of the distance. For three finite bodies the two solutions of Lagrange are well known. In the case of the equilateraltriangle solution the periodic orbits of the particle about one of the bodies
this
motion
may
PARTICLE ATTRACTED BY
71
SPHERES.
405
have a
th^*
if the other two masses are equal. In the'case of solution the orbits of the about straight-line periodic particle any one of the bodies are symmetrical with respect to the line joining the finite bodies. The particular solutions which Lagrange has given for three bodies have
line of
symmetry
been extended to some cases of more than three bodies,* and we shall consider two examples: (1) in which there are five bodies, and (2) in which
there are nine bodies.
Let the masses of n finite bodies be represented by t s n that the bodies lie in the same and that their coordiSuppose always plane, nates with respect to their common center of mass as origin and a system of rectangular axes which rotate with the uniform angular velocity are,
,
M M
respectively, (xx yj, {x2 ,y^, attract each other according to the
,
.
Supposing that the bodies (xn ,yn ). Newtonian law, the differential equations
,
of
motion are
>=i
(35)
df
,
dt
yi
jJ
t
rfj
t u = V(x -x y+(y -y )
t
(t-1,
2,
n; j9*i).
If we assume that each body is revolving in a circle about the common center of mass of the system with the uniform angular velocity N, its coordinates with respect to the rotating axes are constants and the derivatives of
the coordinates with respect to the time are zero. Equations (35) therefore reduce to the following system of algebraic equations
-ivit
+y2; ^^
i=i
'
%
'
Xi)
=
=0.
-*.
***>
(36)
t.i
-N%+IIt follows
M^r
yi)
M x +M x +
l
l
+M x
n
= 0,
M +M y +
lVl
2 2
+M y
n
= 0,
(37)
which express the fact that the origin of coordinates These equations may be used instead of two of (36)
is
.
*See Hoppe, "Erweiterung der bekannten Special ltisungen des Dreikorperproblems;" Archiv. der Math. und Phys., vol. 64, pp. 218-223. Andoyer, "Sur l'equilibre relatif de n corps;" Bull, astron., vol. 23 (1906), pp. 50-59. Longley, "Some particular solutions in the problem of n bodies;" Bull. Amer. Math. Soc, vol.
400
PERIODIC ORBITS.
This system of 2n simultaneous algebraic equations involves the square 1 ratios of the masses, and the 2w 1 of the angular velocity, N, the n n 1 of ratios of the distances x, and y these Accordingly quantities may be chosen arbitrarily and, if the resulting equations are independent, the In order to remaining 2n quantities are determined by the relations (36). 2 the admit physical interpretation, and the masses must be real quantity and positive, while the coordinates must be real. With these restrictions it is not easy to discuss the general solutions of equations (36), but some interesting results can be obtained by a study of special cases. If, in the
t
.
problem of three bodies, the assumption is made that the triangle formed by the three bodies is isosceles, it can be shown that equations (36) can be
satisfied
On
the triangle is also equilateral. supposing the number of bodies to be five, the system of equations
only
if
to be satisfied
is
(a)
M
M _ N\ X 1~
-i.
1L2
Z,
+ M x + M x + M x + M x = 0,
t
(x t
'
-x
(b)
">
t)
"1
Mfa-x,)"~ Af fa -s ) ~T M fa -z,) _ n W, ~S ~3 Tj
.
1.1
'
1.3
'
1.4
'
1,5
(c)
_#! r
">
-\.
(x t
-x )
1
Mzjxs-Xs)
'2,3
Mfa-xJ
'2,4
(xt
-x _ n
b) b)
'2.1
'2,5
(d)
_ Nl r 2*3
_i_ ~
I
M {x -xj M Qr -s
x
2)
(s 3 -a; 4 )
_,_
M (x -x
b 3
1,1
'3,2
3 r '3,5
_ V, n _n U,
(38)
0,
(e)
t)
M
1
(x t
^j
-x
3)
K,
1.3
1, B
(/)
M,y
-ft*,,
"'
+M y +M +M y +M y =
1
I j/ 1 4
(g)
-1-
M*(yi-y*)
'1.*
(y t
-yJ
M^y.-y,)
'1,4
...
M.fa-y,) _ A
M.5
'1.3
(h)
M,(yt -y
t)
"'
'2.1
M<(yt -y<)
Mfa-yJ =U, n
^ *2,5
t
'2,3
'2,4
(0
- jg
*
4. ^1(1/1
-yi) 4
{y 3 -y,)
'3,2
M (y,-y
t
t)
M (y -y
t
3,5
t)
_n
'.l
'3.4
C?)
_ jg y 4 M (y< - y,) r
x
2 tf *
M (y 2
4
y,)
3 (i/ 4
-y
3)
M (y - yj
b t
Jc
r*
1.3
1.5
Let us suppose that b lies at the origin of coordinates and that the other four masses, which are equal in pairs, lie on the coordinate axes at
PARTICLE ATTRACTED BY
71
SPHERES.
407
the vertices of a rhombus with the equal masses opposite each other (see Fig. 10\ This is equivalent to the relations
Xi
J\j
X2
U,
Ay
X^
Uj
x6 = 0,
M = M, = M
1
,'
= 0, fc
th-KA,
y = 0,
t
y,=
-KA,
y,-0,
M^M^M".
(39)
On substituting
we find that
Fia. 10.
()>
(c), (e),
(/),
(</),
and
(t)
are satisfied.
'
Equations
(6)
and
(d)
become
identical, yielding
2M
k*
(2A)
(j)
(40)
and equations
(h)
and
become
identical, yielding
W
k
2 3
2M
A (Vl+i^)
3
2M"
(2KA)
3
(KAY
(41)
X, and
2
.
eliminating /k between equations (40) and (41), relation between the masses,
On
we obtain
for the
M
The
W-K'iVi+K'y i^
8K -(Vl + ICy
3
J
(4(i-jp)(vr+g)'i
8K - ( VT+K
3
(42)
choice of the constant K, which is the ratio of the diagonals of the rhombus, is limited by the condition that the resulting ratio of the masses
must be
regarding
positive.
To
we
set
M=
b
l.
Then,
K as
408
PERIODIC ORBITS.
plane.
M'M"
point in
Only those pairs of values (M',M") which represent a the first quadrant are admissible. This condition will certainly be
is, if
the coefficient of
M'
is
positive.
This condition
is
easily
found to be
-^<K<V3.
If
if
(43)
the slope of the line is negative, it may still lie partly in the first quadrant the intercept on the 3f"-axis is positive that is, if the coefficient of h It is easily verified, however, that values of in equation (42) is positive.
Fig. 11.
K which" make
;
the slope negative, also make the intercept on the M"-axis Hence the choice of the ratio of the diagonals is limited by the
is
arbitrary.
fifth
body at the
may be summarized as follows Suppose a value of satisfying condition (43) is assigned; then two of the masses M', M", and be chosen arbitrarily* and the third is determined by can 6 The length, A, of one semi-diagonal can be selected at pleasure equation (42).
center
and
the
angular velocity is then determined by equation (40) or equation (41). In the following discussions this configuration will be referred to as
configuration, (B), will consist of nine bodies, arranged as
configuration (A).
The second
M M M M
1}
t
,
st
lie
on the coordinate
circle of radius
K=1
M M
6
,
t ,
409
M M
7
,
on a
that
circle of radius
all
of the angles between the coordinate axes and while the ninth body is at the center. KA, Supposing the masses on the same circle are equal, we have the following conlie
on the bisectors
ditions:
x,
xa
= A, = 0,
2/i
= 0,
y,
2/3
x3 =-A,
x 4 = 0,
=A = 0,
yk =-A, yb = \V2KA,
y 6 = \V2KA,
.
= \V2KA, x,
x6
=-V2KA,
(44)
y7 =
- i V2KA,
-lV2KA,
y,=
= o, y* = ilf = ilf =
6
M = M".
8
Corresponding to equations (38) of the five-body problem, there of equations, which, upon the assumptions (44), reduce to
1
is
a set
N\A*
ki
m,( M
'
J-
4-
l\
45)
FA
k?
2 Eliminating JVMVA; from equations (45), there results the following condition on the masses M' M", and the ratio, K, of the radii of the 9
,
circles,
'
KV^2
)
2
~
(46)
2K-V2
2K+V2
j
IN
The
K and two of
the masses
is
limited
by the condi-
tions that the third mass, as determined by equation (46) and the square 2 The limits within of the angular velocity, iV from (45), shall be positive.
,
which the choice can be made have not been determined, but for the purpose of application in numerical example 3, the following set of values satisfying the conditions has been computed
:
K = 2,
M' = l,
A = \,
M =l,
9
M" = 8.2526,
A/
= 1.6399
A;
410
PERIODIC ORBITS.
For the development 202. Existence of Symmetrical Periodic Orbits. of the type of analysis applicable to symmetrical orbits we shall use configuration (A) of the preceding section, the notation being unchanged except
reference to
now be denoted by M instead of M With M as origin and an axis having a fixed direction in space, M M and P be, respectively, (#i,F the polar coordinates of M M
that the mass at the center will
t
.
let
t ),
l}
ti
and
(r,v).
The
:
R^R
where
= A,
R = R =KA,
2
t
F.-F.-f
= V -* = V3
^ = Nt,
,,
K
is
The
differential equations of
^T_
dt*
/^Y_i_ r
\dt)
JIM' 9
i*
dr'
7x *' }
where
n= k
m
L
r,
A _M 1 + Mi + Mi +M A* r n
r,
s
{v K
_ V) V)3 2
~
r1
r1
j^rcosiv-VJ
2
f|rcos(t;r3
^r os(vC
i
V<)]>
t
ri
the relations
v
2
mv = N,
where, as in the preceding case,
v
a3 = k2 M,
denotes the
mean angular
velocity of P.
referred to an axis rotating with the angular velocity 2V and x passing always through by the substitution v = w-\-V1 w+Nt, and factors depending on the units employed are eliminated by the substitution
The motion
is
M
1
= ap,
d*p
vt
= t.
.
We
d?w
P
d? +2 drKd7 +m )
,
dpfdw,
3Q 7a^d^1
rAQ ^
(48)
These equations have the same form as the set (9) and the analysis and results of that problem are applicable in this case. We know, then, that there exists one, and only one, orbit in which the particle P moves with direct motion and with a preassigned period. We shall see that there exists one, and only one, such orbit which is symmetrical to the line joining and x (also to the line joining and hence it will follow that there 2) are no unsymmetrical orbits of this type. Furthermore, all the periodic orbits which are given by this analysis are closed after one revolution in the rotating plane; hence in case symmetrical orbits exist, they are also closed after one revolution.
411
ap/A which is convergent for all values of w so long as the distance, ap, of the particle from remains less than the distance from to the nearest finite body; and in all that follows this condition is supposed to be satisfied. The expansion has the form
ft
We
can expand
as a
power
series in
a " _k
2
l
+Kl'
) {
+ 3cos2w
+ Vjp)
J3cosw+5cos3w4+
+ 8^(2 p )
3cos ( w
+ i(-jp)
2
.
.1
{3cos(w-7r)+5cos3(w-a-)j+
we have
Let
X,
relations
M'-xfW, ~ Al M +
2M"
<V1+jK*)V'
From
equations (40)
we have
l~ + (Vi+g)V (M' A
3
2M"
[MVd+DM-SAfjF
(M'+4M)v (l+X +8M'
2 3 / /
'
On
2 setting the coefficient of 2V in this equation equal to the second members of equations (48) have the form
k, it
follows that
= Km
v^a d(ap)
r)Q
2
ap
dw
<cm p[3X!sin2t(;
3X2 sin2w;
.
412
PERIODIC ORBITS.
introducing the parameter of integration p by the relations and a/A = r;p, where rj is a numerical constant, equations (48) become
On
m=p
g -,(+)+ ,W/,
where
,+2$( +)-*,
],
(50)
/=Kp[X,(H-3cos2w)+X 2 (l-3cos2w) +
0=
Suppose
kp[3(X,
X,)sin2w+
p=
^( T ),
W = fc( T )
(51)
a solution of equations (50) such that p' = w = at r = 0. Then it follows from the form of the differential equations that , is an even function, and that ^, is an odd function of t. Hence, if the particle crosses the w-axis = orthogonally, the orbit is symmetrical with respect to the line w 0, and of this line.* when r = v the with respect to the time crossing Suppose that
is
= t) again particle crosses this line (or, what is the same thing, the line w orthogonally; the orbit will be symmetrical with respect to this line and the
time of crossing, and the particle will have again its initial position and Hence relative components of velocity at the end of the period t = 2tt. sufficient conditions that the solution (51) shall be periodic are
p'(tt)=0,
w(t)-tt = 0.
(52)
the equations have the form which occurs in the problem of two bodies, and a symmetrical solution having the required period is known,
For p
namely,
p=l,
W = T.
w = 0,
w' =
The
initial
p=l,
p'
= 0,
\.
Consider the solution for values of p different from zero but sufficiently = be small, and let the initial conditions for t
p=l+ft = (l + a)(l-e),
p'
= 0,
w = 0,
(53)
/=!+& =
*It
vr= (l + a)"'(l-e)
I
may
be remarked
conditions, p'
= 0, w = r/2,
The
=t
sufficient
an even function, and i, t/2 is an odd function of t t, tc = x/2 orthogonally, the orbit is symmetrical to this line and conditions of periodicity (52) might be replaced by the conditions
),
example, that
if
is
is
>'(I)=-
<!)-i-*
For the purpose symmetry.
of covering
since the orbit would then have two lines of symmetry. is better to base the existence proof on only one line of
more general
cases
it
413
symmetrical with respect to the epoch t = 0, and can be expressed as power series in a, e, and n, which are convergent for an interval in r including the interval to ir, if the parameters are sufficiently small. If a and e can be determined in terms of n, vanishing with /*, so that the conditions (52) are satisfied, then the solution will be periodic with the 2 period 2-k. All terms of the solution which are independent of m can be obtained from the two-body problem by making the substitution w = ufj.T. These terms are given in finite form by the expressions
The
solution
is
p=(l-\-a)(l
ecosE), n
w = arccos(-
s \iecosE/
=arcsin( \
^-)> J ecosE
where
E is defined by
the relation
jp^yr
On
series
=E-esmE.
returning to the variable w, writing the terms in a and e as power by Taylor's expansion, and applying the conditions (52), we obtain
the equations
0=-|7rae+
It follows
0=-|7ra-7rM +
(54)
from the known properties of the series that there are no terms in e alone, and there are no terms involving n to the first degree except the The equations are satisfied by a = e = n = 0, and in the second term irn. the coefficient of the first power of a is not zero hence the second equation can be solved uniquely for a as a power series in e and fi, which contains n as a factor. The result has the form
;
;(-f+
When
this value of a
is
)
first of
'
(55)
equations (54), a factor /x This equation is satisfied \ = by e=n 0, and since the coefficient of the first power of e is not zero, it When furnishes a unique determination of e in terms of n, vanishing with fi.
substituted in the
ire
equation (55), we have a expressed uniquely with ix. n, vanishing Hence for a given value of n sufficiently small it is possible to determine the initial conditions (53) as power series in n such that the solution in fx is = = symmetrical with respect to the line w t and the epoch t t. Since it is symthis value of e is substituted in
in
terms of
to the line
w = and
the
epoch t = 0,
it is
periodic in t
with period
2ir.
is
symmetrical with respect to the line joining the bodies and it and M,. If we take for the initial line the line joining follows from the same analysis that the orbit is symmetrical with respect to
orbit
The
414
this line
PERIODIC ORBITS.
and the time of crossing it. Since for a given value of p there is one only periodic orbit of this type, it follows that it is symmetrical with with For the conx and with t respect to both the lines joining of the finite bodies the orbits of the periodic particle have two figuration (A) lines of symmetry; there are four apses and the apsidal angle is ir/2. For the configuration (B) (see numerical example 3) the periodic orbits have
four lines of symmetry, there are eight apses, and the apsidal angle is 7r/4 In establishing the uniqueness of the periodic solution of equations (54)
it is to be noted that no use was made of the explicit values of the terms from the second members of equations (50). Hence, if the second members have forms which permit symmetrical solutions, the preceding analysis is applicable without change to show the existence of symmetrical periodic orbits. // the masses and motions of the finite bodies are such that there can exist orbits of the particle about one of the bodies having a line of symmetry, we
have established the existence of periodic orbits having this line of symmetry. As in the problem treated in 198, the period of the solution in t may be assigned arbitrarily (that is, when the finite bodies form a fixed configuration
in the rotating plane). There exist then two, and only two, symmetrical closed orbits having the required period; in one the motion is direct, and in the other it is retrograde.
203. Construction of Symmetrical Periodic Orbits. The method of solutions is similar to that constructing symmetrical periodic explained in There is a slight difference in the conditions which determine the 199.
constants of integration, and the calculation is simpler because p contains It has been only cosines of multiples of t, while w contains only sines.
proved that symmetrical periodic solutions of equations (50) exist, and that they are expressible in the form (18). Since the solution is periodic for a continuous range of values of p, each coefficient p, and w, is periodic with period 2w in r. Also, since the initial
conditions p'(0)
=0, w(0)
i.
=0
hold identically in
p.,
it
follows that
p' t (0)
=0,
and
(0)
=0
left
for every
The
members
and
of equations (9),
of equations (50) are the same as the left members therefore, when the solution (18) is substituted in
left
(19).
The
members have no terms independent of p and the equations for the determination of the coefficients of the first power of p are
<fpi
dwi
_ ~Z
d?Wi
'
~dr
{) '
of
415
c{"
is
4/3.
The
ft '(0)=0,
Wl (0)=0.
1'
Therefore
1'
c,
= c"' = 0.
The constant
c,
is
step of the integration [see equations (23)]. This process is applicable to all the succeeding steps.
The
differential
equations (50) have a particular form which admits a symmetrical solution, and it can be established by complete induction that the equations for the determination of p, and w have the form
t
f|*
-2^ _3 A = iC+(ifc$-
+ 4)cosr
+4J"cos2t+
+jifcdBtr;
(56)
,
^ +2^ = (fc<
1,
+D>ntY,
(29)]
where
c"
-1 '
is
2cf*-ir-2Df:
The
solution of equations (56)
)
is
<
+ a"
>
costr,
>
(57)
+if mn%T,
where
(58)
r-
nb (
a t- 2J
f)>
r=
2.
-A (i>r+2i.f)
As a
first
o= 2
.,
204. Numerical
Example
example
of a
symmetrical
periodic orbit we consider three finite bodies revolving in circles according to the straight-line solution of Lagrange. suppose that the mass M,
We
is
we
select
M=
x
10,
M=
2
between
5.
and
Choosing
is
The
unit of distance
M for MM
X
;
and
follows from the solution of the quintic equation of Lagrange* that = 0.77172 .... The unit of time is selected so is the distance 2 2
it
MM
is
assigned so that
= 5; whence
416
PERIODIC ORBITS.
The
motion
+|apj3cosuH-5cos3w]+-aV J9+20cos2uH-35cos4w]h
+^
2 1
p[^l+3cos2(^- r)}+^p|3cos(;-7r)+5cos3(w-7r)J
],
(59)
+^(|-)V{9+20cos2(m>-7t)+35cos4(w -*)!+
= p ^F +2 3?(^ +M )
-fc2iW
^p[fsin2w+faJsinw+5sin3t4
]
+^aVJ2sin2w; + 7sin4M;]+
J
^3 ^[fsin2(w;-7r)
+ |^-pjsin(M;-7r)+5sin3(w-7r)[2
k? is
determined by
M+M +M,R v
t
^ (1+Rl))
'
whence
fc*
= 0.23763,
fclM> 2.37630,
^=2.58518, ft
a = 1.05914/i,
-=1.37222 p.
The
differential equations of
motion become
cos2u;)pM
2
gjS
-P
^ (^ +m/+ -
= (2.48074+7.44222
(1.15938
^ +2
=1-
5?
+M ) =-( 7
44222sin2 ^PM
The
p
|M
+ (0.62584-19.13740cost-2.47963cos2t+0.40256cos3t)m +
3
(61)
3
+ (41.10885 sinT+10.74793sin2T-0.48307sin3r)p H
PARTICLE ATTRACTED BY
11
SPHERES.
417
the orbit represented by equations 24 points which are numbered 1, 2, of the in the periodic orbit at intervals of represent positions particle
p.
On
is
= 0.2,
(61)
shown
The
The corresponding positions in the comparison circle are indicated by the numbers 1', 2', 24'.
r
.
.
= 7r/12.
21 18 19
19
20
Fia. 12.
Example 3. For a second example of a symmetrical we use the configuration (B), 201, of nine finite bodies, the numerical values being those given. The unit of time is selected so that
205. Numerical
periodic orbit
->(
+<)*+ ?
p
% 5?'"'[if
t i
+ 3<*
2<
*>}
+ 1|
J3
cos
]>
j|p{sin(u;-W,)
2 2
+ 5 sin
(
3(w-F,)j
(|-)
[2sin2(tt-lf
+ 7sin4(u;-lV )j+---](
418
PERIODIC ORBITS.
so that
On taking account of the relations (44) and choosing the unit of distance A = l, the preceding set of equations takes the form
,
^-
.],
^+
tXtr + )= -^Vp[?aVsin4. +
'
'
+ k*M Kf-Vp[!V^+
From 201 we have
the following values:
}
A*.
K = 2,
Since iV =
l,
M' = \,
M" = 8.2526,
8
A;
JV= 1.6399
= 0.60994.
,
It follows that
*
^|P
On
=0.62920,
a2 = 2.10300 M2
(J?)
= 0.52575 M
*.
become
fi ar
p
+[3.63039+8.32914cos4m;]pV+
^ +2 ^(^ +M
The
4
W]pV+ ) =[-8.32914sin4
is
On
= 0.2,
is
found to be
,
= 0.86403-0.00133cos4t+ w = T+0.00150sin4r+
The orbit has four axes of symmetry, namely, the lines connecting the central body with the others. It differs from the orbits of the other numerone respect that is, it lies entirely inside the comparison In terms of p the radius of the comparison circle is about 200). 0.8855. 13 is not drawn to scale, but the characteristic properties of Fig. the orbit, which are readily seen from the numerical values of p and w, are
ical
examples
in
circle (see
exaggerated to
is
in a small drawing.
The
inner circle
drawn merely
circle.
from
PARTICLE ATTRACTED BY
71
SPHERES.
419
206. The Undisturbed Orbit Must be Circular. In the proofs of the existence of periodic orbits ( 198-201) it was assumed that the undisturbed orbit is circular. It remains to be shown that this assumption is necessary. The proof will be made for the case of symmetrical orbits [equations (50)]
198. The undisturbed orbit is given is applicable also to the orbits of = 0. of when For p = 0, the equations are the solution p. by equations (50) the equations of motion of a particle subject to the attraction of a central force varying inversely as the square of the distance. The undisturbed orbit
and
is
must be an ellipse.
Since
unity
(in p).
The
eccentricity,
e, is,
which
will
be denoted by
admit an
infinite
number*
of
solutions.
ellipse for
it
will
be
the
FlGURE 13
e is
-
eccentricity
must
be zero in order to
fulfill
conditions of periodicity.
p=
where
r=
ecosE,
7,7
M
;
e"sinii\
ecosE'
fpn \ (o2)
E is defined by
P
the relation T = E
esinE.
The
initial
conditions for
are
e,
p'
= 0,
w = 0,
w'
=
(1-i)
Consider the solution for values of n different from zero, but sufficiently small, and let the initial conditions be
P
ic
=l-e+^ = (l+a)[l-(c+e)],
Vl-e
(l-e)=
p'
= 0,
w = 0,
+ft =
VT-Xe+e?
(l+a) [l-(i+e)]
in
3/s
If a
and
can be determined
terms of
n,
vanishing with
p,
so that the
conditions (52) are satisfied, then the solution will be periodic with the
*The general case when the differential equations (for >i=0) admit a periodic solution containing an arbitrary parameter has been mentioned by Poincar6, loc. cit., vol. 1, p. 84.
420
period 2x.
PERIODIC ORBITS.
All terms of the solution
obtained from the two-body problem by making the substitution These terms are given in finite form by the expressions
p
may be = w u ht.
n*
= (l + a)[l-(e+e)cos],
c os
W = arccos( 1 where
E is defined by
the relation
(l+o) 3/2
On
series
returning to the variable w, writing the terms in a and e as power by Taylor's expansion, and applying the conditions (52), we obtain
the equations
0= n
l+e
2^(131)^-7^(1-=^+
(63)
e alone,
from the known properties of the series that there are no terms in and there are no terms involving n to the first degree except the term Hence the second of equations (63) can be solved for a as a pw/x. in e and ju in which /i is contained as a factor; the result is series power
It follows
(l-e)'
When
a factor n can be
This equation can be solved for e as a power series in n, which vanishes with n, = 0. Since only those solutions are under consideration which if and only ife are the analytic continuations with respect to // of those for m = 0, the condition e
orbit
= must be
imposed.
The condition
= means
must
207.
be circular.
This section contains some remarks upon possible extensions of the analysis which will permit applications to practical problems of celestial mechanics, and is followed by an illustrative example. The particular problems treated in the preceding articles have no application in nature because the configurations assumed for the finite bodies do not exist. But a glance at the details shows that these configurations are not essential to the proofs. The possible generalizations of the motion of the finite bodies can be made in three ways:
of
Motion
421
198 the existence proof depends only upon certain terms of (1) In the disturbing function which are due to the body If t x retains the motion there prescribed, we may add other bodies to the fixed configuration in the rotating plane provided the operations with the power series are valid. This merely increases the number of terms in the second members of the equa-
tions of motion ; the existence proof and method of construction are unchanged
(2)
In the examples treated the finite bodies form a fixed configuration in a plane rotating with constant angular velocity. This is not necessary If for the type of analysis used. x moves in a circle with uniform angular bodies can have the other velocity, any periodic motion, provided always that the convergence conditions hold. In this case the differential equations of motion of the particle involve t explicitly and are periodic in r. Two
points of difference occur in the analysis: (a) Suppose the period in r of the differential equations is T; then the assigned period of the motion of the particle must be a multiple of T. (6) The differential equations do
not admit the integral of Jacobi, and hence no use can be made of this in the existence proof. This is equivalent to saying that at t = we can not assume w = 0, but must determine the initial longitude of the particle by
the conditions of periodicity. The method of determining the constants of integration in the construction of the solutions is explained in a paper in the Transactions of the American Mathematical Society, vol. 8 (1907), pp. 177-181. further generalization of the motion of the finite bodies is pos(3)
sible
to
by permitting x to move in a path which is not circular. It is possible is subject only to show that the analysis can be used if the motion of
the mild restrictions that the expression for the radius vector shall contain only cosines of multiples of t while that for the longitude shall contain x is an ellipse is treated in the only sines. The case when the orbit of
For
this generalized
motion of the
finite
bodies
may symmetrical orbits of the particle. In equations (50) the first contains only cosines of multiples of w, and the second only sines of
there
exist
multiples of w. The periodic orbit of the particle may be symmetrical if the first equation contains also sines of multiples of w multiplied by odd
r, and cosines of multiples of w multiplied by even functions and the second contains cosines of multiples of w multiplied by odd functions of t, and sines of multiples of w multiplied by even functions of t.
functions of
of t,
apparent that the treatment can be made sufficiently general to permit applications in the problems presented by the motions of the solar system. For example, suppose P is a satellite of one of and that x is the sun. This implies that the disturbing effects the planets
these remarks
it
From
is
upon the other bodies are neglected, since we assume that its The conditions upon the motion of mass is infinitesimal. x are fulfilled if we neglect the perturbations of the other planets upon M; that is, if we
of the satellite
an ellipse. If we neglect the inclithe other planets, and suppose that their motion is
relative to
is
422
periodic (that
it is
is,
PERIODIC ORBITS.
periodic motion which is approximately correct), the methods given to treat the periodic motion of the satel-
we assign a
possible
by
the plane of the planetary orbit, when subject to the attraction of the sun and all the planets. The following numerical example is a simple illustration of the general idea.
lite in
is taken as the unit of Example 4. The mass of mass 10, is supposed to revolve about in a circle of unit = = 1, is radius with uniform angular velocity N. A third mass, t
208. Numerical
x
,
mass and M
of
supposed to revolve
about
M M
in a circle of radius
is
The
unit of time
is
M = m =^=0.2.
M as origin and an axis passing always through M With and coordinates of M M and P respectively (R W
reference to
l}
t
,
the
are,
(1, 0),
t ),
(r,
w).
The
motion of the
particle [corresponding
^-
= (^+M) +^
A-
M M p[i|l-f3cos2^|+|apJ3cosw+5cos3w;|
2
1
.
+ ^aVJ9+20cos2w+35cos4w|+
2
.1
3-V P
)+5cos3(w;-Pr2
]i
)j
~(J-)VJ9+20cos2(w-TF,)+35cos4(w-TFs )JH
2 2
(64)
+ ^ay-|2sin2w+7sin4w;j+
-
]
t)
Vp[| sin2(io<
W + f |- P W {sin(ti>t) i 2
+5sin3 (w-Wt )
+ ^6 (f
2
)V{2sin2(t
2
;-Hg+7sin4( W
is
] ;-^)j+ = N k (M+M^,
t
whence
From
the
M, = 0.90909,
follows that
h?M = 0.09091.
it
a = 0.76910p.
The angular
Hence
velocity of
about
M MM
X
will
is
be selected so that its period with one-half the period assigned for P.
or
N -N = 2v,
t
The
radius,
#,= 11.
t
with respect to
is
determined
that
{M
On assuming
423
,
the finite bodies are in conjunction in the order l} t the coordinates (Rt are given by the expressions t with respect to t ) of
,
W
.
M
t
MM M
,
R = V\+A\+2A
t
cos2vt =
ttt cosW =
^l+A\+2A
7214-1
'
<iosi2T,
sin
W=
T
t
4,sin2T
'
A\
2i^
,
in
2 substituting the values of the constants and the coordinates Rt of for the numerical differential we obtain equations equations (64),
On
relative
motion
2
+ (0.86523 cosu>+1.44205cos3w)pV + (0.33273+0.73940 cos 2w+ 1.29395 cos 4w)pV + (0.276008^278^2^-0.13800008210.4140000821-0082^/*' + (0.10474+0.17456 cos4t+0.31422cos2w+0.10471cos4tcos2m>
-0.55864sin4rsin2w)pM
4
+ (0.07960sin2Tsin^-0.31840cos2rcosw-0.53068cos2TCOs3w;
+0.39801 sin2rsin3w;)pV+
,
(65)
p~ +2^(^+ M ) = -(1.50000sin2w)pM
- (0.28841 sinw+1.44205sin3u;)pV - (0.36970 sin 2w + 1 .29395 sin 4w)pV
+ (0.27600 sin2rcos2w;+0.41400cos2rsin2w))pM
4
"'
- (0.3 1422 sin 2w+0. 10741 cos4r sin 2;+0.55864 sin 4r cos 2w) Pt
The
The The
and
then unchanged if we replace w by t. w, and r by right member of the second equation contains only, (1) sines of multiples of w, (2) sines of multiples of w multiplied by cosines of multiples of r,
first
equation
is
(3) cosines of multiples of w multiplied by sines of multiples of t. the second equation is also unchanged if we replace w by w and r Now let us suppose that
Hence t. by
a solution of equations (65) satisfying the conditions p'(0) =w(fl) =0. It is an even function, follows from the form of the differential equations that and ypi is an odd function of r. When r = the finite bodies are in conjunction
is
\f/ l
in the order
M,
M M
lt
Therefore, if the
-particle
P crosses
the line
MM
orthogonally when
M,
to this line
and
424
PERIODIC ORBITS.
= 1-.66667m+(0.38889+0.72102cost-co82tV
(66)
*
The
by equations
(66) is
shown
in
Fig. 14.
The
by the
.
",
8'.
M,
to
the
tion that, at t = 0, the finite bodies are in conjunction in the order lt t Without this assumption the expressions for t and t contain a parameter its orbit at the origin of time. With % in indicating the position of
,
.
MM M
reference to the physical problem, therefore, we can not affirm in this case that, for a preassigned period, there exists one and only one direct periodic
orbit.
It is necessary here to
of the configu-
and
For example, particle at the origin of time. orbit the periodic having preassigned
the particle crosses the line the finite bodies are in x when in the order and we conjunction M, t M^, might have still other orbits with the preassigned period if P crosses this line when the finite bodies are
M M
,
not in conjunction.
Chapter XIV.
CERTAIN PERIODIC ORBITS OF k FINITE BODIES REVOLVING ABOUT A RELATIVELY LARGE
CENTRAL MASS.
By Frank Loxley
Griffin.
For a given system of k finite bodies, moving in a relative to another given plane given body, there is a 4A;-fold infinitude of possible orbits the variations which the configuration of the system under209.
The Problem.
goes and
at
orientation in the plane being determined jointly by the mutual attractions of the bodies according to the Newtonian law, and by the values
its
any instant
The
differential equations
form transcendental integrals,* energy and areas, even when the masses
and their first derivatives with admit no algebraic or uniaside from the two fundamental integrals of
of all the bodies except one are very small; nevertheless, by restricting the initial values of the coordinates and their derivatives, in a manner to be shown below, it is possible to find
an extensive
In fact, for arbitrary values of the masses (save that one of them, M, shall be large in comparison with the others, x t t )rf there
M M
,
exists
a k-fold infinitude of
an
In these orbits (which, for small finite arbitrarily ^reassigned period T. values of ... lf t , , t} depart but little from a set of concentric
M M
circles
rical
about the planet) the k satellites come periodically into a "symmetconjunction," that is, they are all momentarily in one straight line with the planet and moving at right angles to that line. These conjunctions may,
or
may
the motion
not, always occur at the same absolute longitude; in the latter case is periodic with reference to a uniformly rotating line.
Besides the demonstration of the existence of such periodic orbits, this method of constructing the solutions without integration, chapter contains:
a single application of that process having provided formulas which reduce the problem to one of algebraic computation; a numerical application to the case of Jupiter's satellites I, II, and III; a proof of the non-existence of certain other types of orbits;
See memoirs by Brims and Poincard, in Acta Mathematica, vols. 1 1 and 13. tin other words, the distribution of masses is such as is presented by the sun and any number of planets, or by a planet and any number of satellites. For convenience, in what follows, a single expression, planet and satellites, will be used with the understanding that it covers also sun and planets.
425
426
PERIODIC ORBITS.
/3,,
&
be
(1)
by
/S's is
Mfrv =
(i
= l,
.
,*),
number
qt ^0
let
below, and
Let a system of positive = l, k divisor, p (t 1), and a be selected arbitrarily, save for the restriction mentioned v n, and a be defined by
to be selected arbitrarily.
common
(*-i,...,*-i),
p,
(i
(2)
nJaJ^M
where
k*
= l, ...,*),
(3)
denotes the gravitational constant, and where, of the three values Also let the of a, satisfying (3), that one is to be selected which is real. at are in ascending order of magnitude, notation be so selected that al}
. .
the
being so selected that no two of the a, are equal and now, vanishes. that is, if the satellites were "infinitesimal" possible If m were zero ak as radii; from (3) orbits would be circles about the planet with alt
Pi
.
. .
it
It is quite follows that the angular velocities would be n^ n,. immaterial whether any of the n are negative; the results obtained hold The configuration irrespective of retrograde motion of some of the bodies.
. . .
of the infinitesimal
period T;
for, it
follows
system would undergo periodic variations with the from (2) that
2t
m i.
-rit
or each synodic period is a sub-multiple of T. This condition being satisfied,* the motion of the infinitesimal system would be periodic with respect to a
rotating with uniform angular velocity that of t or, indeed, that of any other bodyf Mj though whether or not the system ever returns to the same position in space depends upon whether qt is rational
line
through
M,
or irrational.
In describing the orbits mentioned, the infinitesimal satellites would be subject to certain initial conditions, the 2k coordinates and their derivatives with respect to the time having at the instant t = t certain values, say c w (t=l, k] j=l, ,4); but if the k finite satellites are subjected
.
.
to these
same
mutual disturbances
in general destroy
then, to determine what, if any, increments Ac{J can be given to the former initial values b, to preserve the periodicity when all the satellites are finite.
periodicity.
first
The
problem
is,
*Poincar6, treating three satellites (Mtthodes Nouvelles de la Micaniqve Cileste, vol. 1, pp. 154-6), states the condition thus: Integers o, 0, y, mutually prime, exist such that a+^+-y=0and on,+/3n,+7n, = 0. Evidently, in the case of three satellites, this condition is equivalent to (2), since (n, n,)//3 = (n, n,)/( a); but for a greater number it is not so. Thus, if n, = 7, n, = 5, n,=3 vXn, = V27 the integers 5,-7, 1, 3 For the general case a re-formulation satisfy a condition similar to Poincar6's, but periodicity is impossible. such as (2) is necessary.
1,
fThe commensurability of n t -n t (i = l, fc-1) evidently involves that of ty-nj (t' = l, For from n< nt = p v and nj ni = pjv it follows that n, rij**{pt pj)p. k). 7+1
. . . ,
PROBLEM OP k SATELLITES.
210.
427
motion
and
MS and MH being rectangular axes which rotate in the plane with the
Differential Equations. Let the common plane of relative of the k bodies be selected as the .ztf-plane, the origin being at ,
The
uniform angular velocity N. Let the coordinates of t referred to these axes be & and t?,; then the differential equations of motion are
(a)
g - 2N^ - N%+>(M+M
^k
t) jjj
+ Z'*M,(*=& +
(*-*
^)=>
(4)
(b)
+ 2N
& - N\+K\M+M
r)
t)
% + 2' tM,
J) -0,
where
= Z+-n),
Except
-(,-,)+(%-*).
in proving a certain
symmetry theorem,
these coordinates are less convenient than polar coordinates referred to rotatBesides (1), (2), and (3), let the following definitions ing reference lines.
be made
a tj
= a /a
t
vq t
= n, = (.Pj-Pi)T+(\-\),
(i
= 1,
k),
vt
= T,
<t>ji
where the
the
M,
are arbitrary constants, later to be taken as the longitudes of at the origin of time for n = 0.
X,
= r,cosw,,
ij,
=r
sinu
r,
=a x
i
u =w
i
+p
T-{-\ i
t
(6)
and
line
referred to a be taken equal to nt so that w, is the longitude of rotating with uniform speed n,. The differential equations become
,
(a)
x w"+2x' (w' +q
t
f
'
<T t j/
=0,
(7)
(6)
x';-x (w:+q
t
y+ q2 {1 tf x\
<>
tfi)
+m2' 8
where
a)a-\ J
[^
( l
+Xj0O8(il>j
+w,-w,)
(-,
^-J J
=0,
= a]x\-\-a)x)
2a aJ x xj cos(<t>j
-t-Wj
Symmetry Theorem.
ta
and
proof will
,+^o-
428
PERIODIC ORBITS.
The
<,
m=-vi,
l=~t.
(8) into
(8)
Consequently, every solution of (4) is transformed by of (4). Moreover, the initial conditions
some solution
*-*'
* =0
>
S =0
>
m=
(i
b"
(9,)
Therefore, that solution of (4) which satisfies the initial conditions (9) transformed by (8) into itself. Hence, if that solution is
is
6-*,(0.
then
#,(0 -*,(D -*(-<),
*-*<(<)
= l,
...,fc),
(10)
*<(0 --*(?)
--*(-*),
)Vt)-
whence
also
&(vi,
,
*?i)
<(
lit
noted that the proof holds, whatever the value of N. It is also geometrically evident that the symmetry, if present at all, is independent
It will be of the rate of rotation of the reference line.
replaced
by
X,
is
= Xi(t),
W,
= W,(t)
(l
= l, .,.,*)
(11)
a solution, then so
is
x,
= x (T+2mr),
t
w = w (T+2mr).
t
(12)
These two will be the same solution have the same values at t = t that
;
if
is, if
x
l
(T
+2nw)=x (T ),
t
w; (r
1
+2n7r)=w;
(T
),
x' (T If
+2nT)=x' ( To ),
l
w'(T
all
+2nir)=w' (T
t
).
values of
r,
x
that
is,
(T+2mr)=x
(T),
w,(r+2n7r) =w,(t);
(13) are sufficient conditions for the periodicity of the solutions. is obvious, if the period is to be 2mr.
That
PROBLEM OF k SATELLITES.
429
In the case of a symmetrical conjunction at t = 0, other Special case. sufficient conditions can be formulated. For, if x'(0) =w,(0) =0 (t = l, ...,*),
and
if
every \
is
a multiple of
x,(ir)
i
ir,
it
w,(ir)=-iv (-ir),
w' (Tr)
l
(Tr)
= +w'(-ir).
if
is
put equal to
and
w,(t) are
X (7r)=x (-7r),
( i
(c)
(ir)=W
ir),
(15)
x'M=x' (-T),
i
(d)
iv'M=w' {-ir).
t
Of these conditions
(c)
(a)
if
and
(d)
t
are satisfied
It
by
= 0. x'^ir) =w (Tr)
t
may
(
(with period in
equal to T) are
1
a<(0)-0,
w,(0)
= 0,
\-0ott,
xV)=0,
(
w,()-0.
(6)
are
the differential
x
t ,
= \,
t t
Wi
=Q
('=1, ..'.,*),
( t t t
giving the circular orbits r =a, u =\+p T, in which at t=0, x =1, x' =w =w' =0. If these initial values are given increments Ac ;(i=l, k; j=l, 2, 3, 4), then the solutions of the differential equations (7) for /xt^O are developable
(
.
n and the Ac which converge throughout a preassigned interval of t for sufficiently small values of those parameters.* Such solutions are in general non-periodic; in fact, the periodicity conditions (13) or (16) impose the condition that 4A; power series in these 4&+1 parameters shall In the cases to be considered these Ak equations will determine vanish.
as
power
series in
as unique functions of p, holomorphic in the vicinity of m = and vanishing with n; so that, for sufficiently small values of n, there exist initial
the
Actj
conditions (depending upon T, gt , the p n, and the ft) such that the orbits described are periodic with the required period. Evidently for smaller and smaller values of ;u, smaller and smaller devit
ations from the initial conditions of undisturbed motion are sufficient in order to get periodic orbits. These orbits for 5^0 may be said to "grow out Of course, for of" the undisturbed circular orbits as n grows from zero. any given masses, n and the ft being fixed, the possible orbits of this sort can
fj.
See
14-16.
430
vary only with T, qk or the a class of orbits.
,
PERIODIC ORBITS.
p,
;
be inquired whether the conditions for periodicity can be satisfied by such values of the Ac,, as to prove the existence of a class of periodic orbits of each of the following types Type I. The finite system has a symmetrical conjunction. Type II. The infinitesimal system has a symmetrical conjunction, but the finite system has none. Type III. Neither system has a symmetrical conjunction.
it
In what follows
will
214. Integration of the Differential Equations as Power Series in Parameters. It will be necessary to obtain the first few terms of the developments mentioned in the preceding article. Instead of increments Ac,, to the initial undisturbed values of the coordinates it will be more convenient,
in finding the properties of the solutions, to
e,
co,
t<
defined as follows.
At
=
i
let
co S 6
),
^=Ac=n,(l+An,)* 1
,.
,
^^bO
,.
,'
(*-!,...,*),
-<Z,(l+An,)r,
(17)
^; = n,Ac =
<4
t
n,(
l*_
^ly
e]
~n
= 0,-c,sin0,,
the v being equal to n,+gt T, the true longitudes from a fixed reference line. It is evident that the Ac,, are holomorphic functions of the An, e, a>,
, ,
,
and
for sufficiently small values of the latter quantities. Consequently, solutions of (7) exist also as power series in the new parameters. Further,
t,
since the real positive values of the radicals and the smallest values of the inverse cosines are to be taken in (17), the Ac,, are given uniquely in terms
of the An,,
e,, a>,,
and
t,.
From
it
follows that,
if
the latter
quantities can be determined as unique power series in p., satisfying the conditions for periodicity, then also there exist for the Ac,, unique power series in n, satisfying the conditions. Conversely, while the Jacobian of the Ac,, with respect to the new parameters is zero for An, = e, = w, = r, = 0, yet, in the only case where discussion will be necessary (viz., for Ac, 2 =Ac,, = 0, whence a>, = t, = 0) the solution for the An, and e in terms of the Ac, x and Ac, 4 is unique; for the Jacobian of the Ac (1 and Ac, 4 with respect to An, and = e, = 0. Hence, in this case, if the Acn and e, is distinct from zero for An,
,
Ac, 4 exist as
unique
e,
exist as
such
series.
In the developments of the coordinates as power series in i* and the new parameters all those terms independent of n may be obtained, together with a knowledge of their properties, in the following simple manner: The terms
PROBLEM OF k SATELLITES.
in question are those
fj.
431
is put equal to zero, and are thereremaining when k of infinitesimal satellites when the initial problem in are other the solutions of k two-body problems. conditions words, (17) The dynamical meaning of the new parameters is then evident. The
orbits of the infinitesimal system, subjected to the initial conditions (17), are
which the mean angular motions, major semi-axes, eccentricities, longitudes of pericenter, and times of pericenter passage are respectively
ellipses in
n,(l+An,),
If in
Ckil+An,)-*,
X,+o<,
^.
the development of xt the coefficient of Anf ej u>* t\ be denoted by xiJ(lh i then, by applying Taylor's theorem to the well-known developments of the coordinates in elliptic motion as power series in the eccentricity,* it is found that the following coefficients of first and second degree terms do not
vanish
^i.oooo
1>
Ef.iooo
o>
Pi.oioo
COS^T,
i ,
#(,jooo
n 9
>
Z(.
o*oo
= sin
g (T
x UBm
= -q
s\nq,T,
w.iooo
w.
= ?iT,
5.
io <iW0 ,
= 2sing,T
wumt = l,
w
(18)
w,, mi
=q
t t
i,aioo=jsm2q T,
t
t,
mn =-2q,coaq,T,
t,
im
= 2q TCOsq T.
From
upon
simple dynamical considerations the following important properties can be established. Let x uim be written x]lt to indicate its dependence
r.
Then
*S?-CL-o
, , , ,
-i, ...,*),
(19)
where m is any integer; for, the coefficients x' etc., are those of the t>Bltt terms which do not involve An, w, and t, these terms being obtained by
putting Arij = a> = r, = in the developments. But for these parameters equal to zero the initial positions are apses and the periods (in t) are 2w/n Hence, at T = mv/q M, is at an apse and x' = w = 0, whatever the value
( { { ,
t
.
true for a range of values of e it follows that the coefficient of each power of e, in x\ and in w is zero at r mir/q
of et
.
Since this
is
t ,
t .
terms independent of n only those parameters the same as that of the coordinate developed; the terms involving n introduce, however, the other 4(/c 1) parameters. Terms involving /x. The only terms involving n whose coefficients are
It is evident that in the
is
needed
H
(t'
ix
{
and
tie,
(j=l,
t),
in the
development of x be
a;,(0;
coefficients of the
same
.
.
quantities in
.
Let the coefficient of and that of ne, be x,(j; t); let the be respectively w (0; r) and w (j; t)
. . . ,
k).
= l,
,k;j = l,
,k).
The
p. 171.
432
(a)
PERIODIC OKBITS.
Since the solutions
satisfy the differential equations identically in the parameters, the equating of coefficients of corresponding powers on both sides furnishes sets of differential equations for the successive coefficients in
must
the solutions.
The arbitrary constants which the successive coefficients carry are (6) determined by the conditions that the solutions shall reduce identically to
equations (17) at r = 0. For each pair "of coefficients x,(f; r) and w (f; r) (/=0, k), two simultaneous differential of the second order. equations (7) give equations
t . .
.
(7a) can be integrated once immediately, and its integral combined with the equation from (76) renders the latter a well-known type,
oo
z<"C/+o?z,(/;t)+ 2
+ aT = 0.
(20)
Its solution, x,(f; t), when substituted into the first integral, permits the final integration for The initial conditions are t (f; t).
x (f;0)=x' (f;0)
t
= w (f;0)=w' (f;0)=0
i
i
(t
= l,
. ,
ft;/-0
ft),
(21)
do not involve n at
all.
Now
or sinOiT
is
the form of the solution varies greatly according as a term cosg<t or is not present in (20). In the former case the solutions con-
tain a so-called Poisson term, rcosqa or Tsing,r, and in the latter case they do not. In all the z<(/;r) and w,(/;t) (/ = 1, k), a Poisson term is
.
.
present; they are present in the x,(0;t) if, and only n say n, and n there exists an integer J such that
( ,
,
if,
for
some pair
of the
J
The meaning and consequences
{n t
-n )=n
t
(22)
219.
- 2e
tJ
cos
<f>it
+ el) "1
(s
= 3,
5)
as a cosine series, where, for the sake of a uniform notation, the following definitions are made:
=a
Then
and
&&=o,
if
/<t;
e,j
=a
(JO
and
%1,
if j>i.
(23)
,/ (l-2ecos<k,+e? ,)- *= 2
j
F^t^cosm^,
r;n (e,/)cosw4>J(
,
n-2ecos</>+?,)where the
6/s
= 2
Fm and Gm
beginning with
e".
PROBLEM OF k SATELLITES.
Finally, the desired coefficients of the functions x
t
433
and
are,
for
i-\,
k and for/=0,
1,
+E
tf
TCOsq
t/
TCOii2qtT
00
+K
w,(f
;
lf
Tsm2q T+ 2 (a^'cosmT+c^sinmr),
(
r)
-L
lf
+N
if
-q D
t
if
r*
-2
t
|X + ^1 sin q
iT
(25)
+2\C
if
if
T 2H if TCOsqiT J jcosq +
<1
-2E Tsmq T
t
l
^J
i
tf
Tsin2q
where
D =
to
and J l/
=K
if
for/^t, while
(/
if
every
o
A,
or x, then
J = 0, ci/ =A / =^
but
if
=A
=c 7 =
(
,
C=
lf
(/=0,
...,*);
(26)
no relation
E = H = J = K = 0,
io
io
tt
D =
in the proofs are
(/
(),
...,*).
+ g o /'7?/Cri+4'7?/(l+a?
(
/ )(y,
ga^ij^Crj
+ X,,],
(27)
C-2%
1
m=l
ze(
M )rinm(xJ
-*^,
where
2?
6-(J-
2 q t (m q l
-q -q
i
2
i
Za "' -
*--+*-+>
(28)
where the
A' (/
and
among
the
q,.
434
PERIODIC ORBITS.
215. Existence of Periodic Orbits of
Type
I.
When
finite satellites
are subjected to the initial conditions (17), the solutions of the differential equations are expressible as power series whose first coefficients have been
tabulated in equations (18) and (25) in /x and the An,, e,, ,, and t,, converging throughout an arbitrarily preassigned time interval for sufficiently small values of these 4&+1 quantities. And, although the orbits are in
general non-periodic (as
it
in the tabulated
that the conditions for periodicity can coefficients), nevertheless be satisfied, provided that every X, = or tt, by assigning to the co, and t, the value zero (identically as to n) and to the An, and e, certain values dependent upon n, that is, at t = there is to be a symmetrical conwill
now be proved
junction in which the velocities and distances must be properly chosen with reference to the masses if periodic motion is to result.
(a)
In this case the conditions for periodicity are (16), of which equations The necessary and sufficient conditions are then are already satisfied.
(a)
(b)
7r)+M^,(0;7r)-|x',
i i
0=e
(q s\nq,ir)+fi(x' (0;ir))+An e
i
f
Am (ir)
(t=l,.
.
.
(29)
,*).
+ i-2e,x'(f]Tr)+
f
The problem
Here minant
of solving these 2k equations has been treated in Chapter I. the functional determinant, taken at An, = e, = ju = 0, is merely the deterof the linear terms,
whose value
is
4=
so that its vanishing depends
71-'
(30)
upon the
q,
Since,
from
(2)
.
and
.
(5),
?+P
it
(i=l,
if,
fc-1),
follows that
A vanishes
t
if,
and only
Since
the arbitrary qk
the An, and
is
selected as an
integer.
Case I: qt
is
not
an
integer.
A^O,
e exist*
t
as unique
vanishing with n, converging for n sufficiently small, and satisfying (29). Thus, for all sufficiently small masses of the k finite satellites there exist initial conditions for which the resulting orbits are periodic and the presence of qt and of the arbitrary integers p shows that for given masses there is a A--fold infinitude of orbits of this type. A family of such
power
series in
ju>
"growing out of" any set of circles for which the infinitesimal would have commensurable synodic periods, whose least common system In Case I consecumultiple is not divisible by the sidereal period of satellite k. tive symmetrical conjunctions do not occur at the same absolute longitude; nor will any later ones occur at the same longitude if qk is irrational. t
orbits exists
See
51, 2.
fSee J219.
PROBLEM OF k SATELLITES.
435
Case II: qt is an integer. Here Aj = 0. Nevertheless the Jacobian of the ivJtt) with respect to the An,, taken at An, = e, = /x = 0, is
A 2 = tt*
II
o,^0.
t
(31)
qt which requires n = 0, and such a Hence (29a) can be solved for the An*
converging for sufficiently small values of the latter quantities. Now, by (19), every term in (29a, b) has either n or some as a hence the solutions have the form An, factor;
fx,
and
An =/P(cy ,M)
J
(*-l
fc;
i-i
Jb).
(32)
If the
power
converge
may
(This the whatever satisfy periodicity conditions, If n is divided out,f relations are obtained
and
es
and contain n as a
factor.
among
Q
and
ll
of the
form
= x[(0;w)-e
this point
x'
t
,
im U)
w (0;r)+ JJ^O'JlO+MJ
t
!<
(33)
two questions of importance arise, viz., as to the vanishing of the x',(0; ir) and as to the vanishing of the determinant of the coefficients of the linear terms in e,. By (25) and (27) every x' (0; ir) is zero unless the relation (22) holds for some pair of the fit. When such a relation does
t
At
by
e,
= = 0,
it
the
vanishing with ll, do not exist. Hence, periodic orbits of Type I, "growing out of the circular orbits," do not exist, if, for any n, and n,
e )}
(n,
n) = n, where J
is
an
the equations (33) are satisfied Li determinant A, of the first degree terms in the
of its elements,
t
power series in the e,or a /aj, a,/a and it is unknown whether there are any sets of values of the ui for which A 3 = 0. It will, however, be shown that there is an infinite number of values for which the determinant is distinct from zero. Let P(f be any element of A 3 the first subscript indicating the row and the second the column; then, by equations (33) and (25),
,
f/
(/*),
(
The
e tJ
depend upon
v,
qt
the
It will
now be shown
that for
which were arbitrarily chosen. a fixed selection of the fi, v, and qt there is an
/3,,
and the
p,
less
For although Ai has no minors of order This case is not essentially different from that treated in 4. than k distinct from zero, yet, so far as the linear terms alone are concerned, the equations (29) may be regarded as k independent pairs. fit is precisely this step which makes the selection of the parameters An, c, especially advantageous. {That qt an integer does not involve the existence of such a relation is shown in 219.
,
436
infinite
PERIODIC ORBITS.
number
which the
let all
e 4/
are "suffi-
For convenience, by
the
be expressed
o,-6Ja*- ai
t
,
0-1,..., *-l),
(35)
where the b v, and qt (hence also a t and nk) are constants independent of a. and (27) Every element of A 3 is, then, a power series in a; for by (35), (34),
PROBLEM OF k SATELLITES.
437
main diagonal element than elsewhere in the same row, except in the first ls_M as does Pu And even this exception row, where P a may now carry a is immaterial, since, when the rows have been factored as before, the constant term in the element Q a is to be multiplied by a minor whose first column contains a8 as a factor, and hence can not destroy the constant term in the main diagonal product. Therefore, it is true without exception
,
.
that
for all values of a sufficiently small. Therefore, equations (33) can be solved for the et in terms of
3
5*
/*,
vanishing
t ,
with
jut ,
as the e
substitution of these solutions in (32) gives the as holomorphic functions of n, for /x sufficiently small.
is
The
An
as well
Hence there
is
exist initial conditions giving periodic orbits of Type I integer, provided that no relation (22) holds and that a
even when qt
an
The
sufficiently small. In drawing this last conclusion, however, a point of delicacy arises. p are functions of a in the foregoing argument, and obviously the p
t
are not integers (as the formulation of the problem requires them to be) The question arises as to whether there for all values of a on any interval.
are, indeed,
From
(36)
integers.
follows that
ft-ft(ffj =3-l)
(i
= l,..., k-1).
(37)
t
discussion will be confined to exhibiting a selection of the b such that there is an infinite number of values of a less than any assigned quantity, for each of which the p{ are integers without common divisor.
The present
let
the
&<
be defined by
and consider (37) for a = a^y/l/n, where n is an qt is an integer, and since t?<=(?*+ /c-*> '-?t
the p are integers.
t
integer.
Evidently, since
(39)
If
(i~i,:..,h-i),
Consider the possibility of a common factor. factor, their difference has the same factor.
to (qt +l)n
pk _
Thus,
also a
,
there
is
a factor
common
qt
and
(gt
+2)n
qk
it
is
n2 +n(w l)(qt -\-l). Hence if n is prime and greater than qt such a factor must divide n+(n l)(gt +l) and also the difference between this number and p t _ 1; or (n 1). But, as n and n 1 are mutually prime, 1 which divides n+(n there is no factor of n l)(gt +l), and hence no factor common to p t _j and p t _, if n is chosen a prime number greater than qt There is an infinite number of primes; hence the p have the stated property. The periodic solutions exist, then, and might be obtained as series in ix alone (convergent for sufficiently small values) by substituting in the original series in n and the Aw, and e the values of the latter 2k parameters, A far more advanas obtained in terms of n from the periodicity conditions. available. tageous method is, however,
factor of
.
438
216.
that, for
PERIODIC ORBITS.
Method
of Construction of Solutions.
Type
"
I.
It
x,(t)
+n.l 2^,
(t)m",
w,(t) r
to,
=2w ,,(r) M
(
(i'=1
k),
-l
(40)
which
(7),
(a)
converge for
x\ (0)
^
0,
2ir,
(b)
satisfy
(c)
satisfy
s
(0)
identically
series (40)
n,
and
(rf)
satisfy
-x x, (t + 27r)
(t)
=w
(t+2jt)
w,(t) =0
identically in
and
(a).
,M=w .M=0,
l
,
(41)
^, B (r+2x)-x (
(T)=u\ n (T + 27r)-w,,(T) = 0.
(42)
These equations
(41)
and
Since
satisfy
(40)
must
satisfy
(7)
identically
in
m,
the
and u\
(t)
must
t
(a)
iv" l
,
+2q
(l-rfJ
l
-2
F^ejcosmfa)
=0,
(43)
(6)
x; u A -'2q w'
t
-Hx^+<t^
( (
+2
y
Since every $
is
equations
(43)
are of the
type
()
(/>)
w"A + 2q
x",
l t .
x' tA
Z)
sin Wt
= 0,
(44)
where the DJ and ^J^i arc linearly related to the Fm and can be expressed in terms of the latter as soon as the p, are chosen. The solutions are
xul (r)
= - d2q
Hi
i
(45)
where the
c|;j (,/
= 1,
{r -q])A
term
4;
. . .
m*B*J = D*J
titi
- 2m$ .4 |j.
(46)
Poisson terms do not appear in (45); for, since no relation (22) holds, no in cosq,r or ednftr is present in (44). Now, by (41) and (42),
i.i
c (.i
c,
c i.\
3o
''
so
that
the k constants
And
here
arise
two
PROBLEM OF k SATELLITES.
(
439
Case I. qt is not an integer. Here, g, not being an integer, cosg r does not have the period 2-nr; consequently, by (42), c = (i=l, k). Case II. qt is an integer. Here cos^r has the period 2x, and (42) is satisfied for the arbitrary c^'^O. These k constants remain undetermined until the second-order terms are found, when the cf} are uniquely determined in destroying Poisson terms. Terms of any order. Case I. Assume that for n=l, h 1, the x (t) and w n (r) have been found, the constants being determined, and
.
.
.
(,
t,
= 2 AcosniT, m=0
w un (r)= 2 B^s&tmit
m=0
x, ih (j)
(-!,...,
t).
(47)
An
and
iih
(r)
(47); that,
moreover, the differential equations for xtih and w are of the type (44) ; and that the constants of integration are determined just as in the preceding case for w = 1. The differential equations are [see (7)]
(a)
2
t+t=A
x,, t
w",+2 2=
t
+2'5 y
1
2
k+l+m=h-\
xJik (sm<j> Jf
+w -w
1
3
t
) l [x J
-a7j m
}
= 0,
(48)
(b)
x"h -
w', t
)-q
i
l
x ,+q
t, >
i
l
(x7')
3
l,
)i
+
sum
+ Wj - w^lxj - aZ
3
1 /j.
.
3
]
m\
=0,
where an expression in parenthesis, having a subscript I outside, denotes the It is to be shown of all those terms in the expression which involve (a) that the variables in (48) whose second subscript is h enter in the same form as the x and w enter (44), and (/3) that the remaining terms of (48) (a) and (6) reduce respectively to a sine series and a cosine series in
(il
liX
multiples of
r.
iih
.
Evidently in (48a) the only terms involving the xiih and w (i= 1, x' and in (486), aside from {x~ 2 ) n they are x"h 2q w' ult Uh Now it can be shown easily by induction that
l
h)
.
q]x Uh
where the
v,
Z.;-*,
2 fr,~h.
k
(49)
Now
vh
(d
x /dix
i
)"h,
and
for this
2
term
N=2,
1,
= l, and
2?, W(,
= 0(f=
,
coefficient "
),,, x un appears with the 1, 2. Therefore, in (48b), the terms involving x Uh and w Uh are %q\xtA and this establishes statement (a) above.
. .
.
440
PERIODIC ORBITS.
On
series
and a
using the notation F'(j) and F'(r) to designate respectively a cosine sine series in multiples of t, it is evident that
F[-F\ = F
C
e
,
Fl-Fl-P,
i*
F
(F
F' = F\
(F
)"
= F%
(F')
.
= F%
.
r WJrl
)
= F'-
h Un (n=l, Hence, as every x Un and w' sums of products of these quantities, and also
. ,
1)
all
(x,~
)t
and parts of
are F'(t).
(a~/),.
Similarly, the
sums
of
all
x' ut Ui w'
There remain in (48) only the terms (sin mu+wj w), where m tJ = <p in = +ir/2 in (486). Let, for the moment, z lJ = m iJ +wi w (48a) and m u r Then, since [d zij /dix'] ^ = w} f w un it can be shown by a simple induction
Jt
<f>j l
t
ll
that
where the
[^"]...= ?*.n sm(m-K!) (,,-,.,)"'. N are numbers and the v, satisfy (49) after h is replaced
t
Now
R't-\(
wuf
j>
= 1, (/
by
/.
1)
is
a ^'( T ) or F"(t) according as 2'zlvf is odd or even. But when this sum is odd, is odd; and when even, v is even. The entire = )t or a F"(t) if tJ = il -\-T/2. is then always a F\t) if product j The differential equations (48) are, therefore, of the form
i
w).r~ w i.tY'
no
<t>
(j>
(a)
,
w' [h
(
u1t
(50)
(b) x'
i
where no term in cos q,r or sin q r occurs under the summation sign, since g, is not an integer. Obviously the integration of (50) is the same problem as
t
x,M=
B^mmr
(51)
(i
= l, ...,*),
(
whore
(m*
Also,
<ll)
A IS = E
(42),
f m Ag
wBj- = D -2mq A
(52)
by
(41)
and
K=-~E?:,
<-<-<&-<)
(i
= l,...,k),
(53)
PROBLEM OF
so that the induction
is
A;
SATELLITES.
441
completely established.
Thus the
successive xtM
and
wUn
reduce to
Xt.n(r)=^?
E+
2 ^'cosmr,
xo Un {r)=
(51')
be obtained without integration by applying (52). It is merely and E from equations (48) at each step. necessary to compute the D Terms of any order. Case II. Assume, as in Case I, that for n = 1, (h 1) the x Jj) and w Un (r) have the form (47), all the constants of The differential integration having been determined except the c*_x equations for the x u1t and w uh are again (50) (a) and (b), where, however, since q, is an integer, cos q, r and sin q r may occur under the summation sign. These terms arise from two sources from the terms cJ5_, cos q r and c2_, sin q r in the x Uh ^ and w,,^ and from similar terms in the earlier xtM and w Un as well as (usually) from combinations of the n in the coefficients. When (50a) are integrated and combined with (506), equations for the x lih are obtained; to avoid Poisson terms in the xUh the coefficients of the terms in cos a, r in these last equations must be made to vanish. These co-
and
may
<f>
efficients,
E^
1,
.
the e
(-
2D\^, involve the cj"-! and various known constants; e.g., will usually h-2); and the vanishing of the
.
.
E%-2D%
determine the c^_, In the first place, the only terms of (48) in which the x uh ^ and
.
w uh -i
appear are*
(O)
X uh .
w'll
+ Xu M[n-x + ^\.H-iWu^2x\ A W uh ^
x
+2
,
1
-U7, _ cos0i
1
,|w,.,_
- 2.T,,_
sin</>
6
fJ
sin <,
(aX-i
(&)
- 2o x
1
(iA
_ 1 m;;. 1
- 2q x
l
l ,
+ Gtfx,^^ - 2gfr3
a;,, ft
_1
(54)
/l4
+j!i'iiIli(iir , ;i*-i-*M- 1
- 3i?J/r
f/
i(o
-|-to
;,
For h = 2 the
first
(b)
are to be divided
by
2.
442
PERIODIC ORBITS.
171
<r
where
Whence it is seen that the cf\^ enter =(l2eu cos^+e*,) E% 2D% linearly; and, denoting their coefficients by R (i=l,
iJA
.
the
,
t/
k;
/=
1,
k), it is
found that
>
R{-l)"i
where the
(i-l,...,*;/l, ...,*),
(55)
are the elements of the determinant A,, discussed in the when A, 5^0, the determinant of the coefficients of
E$ - 2D?j is zero.
The constants
(51)
cj?_, (/=
1,
k)
are then uniquely determined. The values of the xiJt {r) and
(53) are, however, replaced by
and
(52)
equations
C= the c
E,
0-1, ...
lb),
(56)
The induction is thus estabremaining undetermined at this step. In this case, too, the successive xUn and wUn have the form (51'), the lished. c "n appearing as Ajjf, and are obtained without integration. But, besides
computing the Dj' and E^ and applying (52), it is necessary to obtain also the !n+i and E% +1 and to determine the cj" by solving the equations
(1) In constructing the solutions it has been tacitly assumed the Fourier series representation of the x Un and w Un and the attendant that This is justified by the consideramanipulations of these series are valid.
Remarks:
well-known functions encountered in the first step are representable, together with their derivatives, by uniformly convergent Fourier But the products of two such series is another of the same type, and series.
tion that the
also the integral of a uniformly convergent Fourier series
is
uniformly con-
vergent.
(2)
At every step the convergence remains uniform. The question naturally arises as to whether, in any
is
orbits of
Type
I,
a multiple of the period of the infinitesimal system, namely, mT. An examination of the method of constructing the solutions furnishes the answer. If mqt is not an integer, the constants are determined
Case I; if mqk is an integer, then by the method used in Case II. In any event, for a given value of n and a given set of p, and qk there is a unique solution satisfying (41) and (42) (with 2mir replacing 27r). The solutions found above, however, satisfy these conditions; hence there are no orbits of this type whose smallest period is a multiple of the period of the
precisely as in
,
infinitesimal system.
217. Concerning Orbits of Type II. For Type II, as for Type I, all the X, are multiples of t; but it is proposed to ascertain whether the co, and t, exist as functions of /x (not identically zero, but vanishing with n) so as to
satisfy the periodicity conditions (13).
easily
by
examining the
method
PROBLEM OF k SATELLITES.
Let the origin of time be selected as the instant when
satellite k
443
has an
apbidal passage, and the origin of longitude at that satellite's apsidal position so that co t = Tt = 0; and let it be assumed for the moment that the w and t,
t
n satisfying (13); then there exist solutions of the type (40) satisfying (42), but not satisfying (41), for all values of n except for i = k. Because of the absence of conditions (41) some of the constants in each step are left undetermined until terms of higher orders are found.
(i=l,
. .
.
1) exist as functions of
terms the differential equations are again (44), and the solutions are (45), the coefficients being determined by (46). Evidently " the c are determined as in Type I, but for the other c"J two cases arise. Case I. qt is not an integer. Here, by (42), c, = c* = (i=l, k), but the c"' are at present undetermined, except c"J = 0. Case II. qt is an integer. Here the terms cos^t and sing,T have the required period 2ir, and hence the cf\ and c*, together with the c, remain
first-order
(
{ . . .
For the
0. undetermined, except c^ = c Terms of any order. Case I. Consider next the terms of order two. The differential equations are still (48), using h = 2; however, because of the ejj these equations now reduce not to (50), but to the form
(a)
w'*+2q
x' u2
+H +
+E +
2 (D$MLmr+H%co8mT)**0,
(57)
(b)
2 (E^cosmT+J^slnmr) =0,
integration apparently introduces non-periodic terms Hf^r; but computation shows that these coefficients vanish. Thus
J,'
c.
The
first
//^ = 2'(^-^)5
^[fa,^(2G
the
-r;2
)-|F
(7(e
.
f
],
and by Le
Verrier's relations
among
F(e u ), and
,),* it is
found that
solutions
the bracketed expression vanishes identically in the of (57) have the form
e tJ
Hence the
o'/i
.+ 2 (A^cosmr+P^smmr),
m=
l
(58)
- cJS
2 (B% sin mr
1
+ QJJ cos mr
vanish with the c *}, which (together with the c%) and Q where the P remain at present undetermined, the other constants arising in the x u% and
Tisserand:
Mtcaniqw
444
the
PERIODIC ORBITS.
w IA having
first-
order terms.
For the terms of order three, the differential equations are of the form do not vanish identically. For, while the c% drop out just (57), and the H in the did the as preceding step, the c"J are now present (linearly only), c% and Q% and directly from the wm (t) The entering both through the P destruction of the non-periodic terms H t, by setting the #Jj = 0, will be found to require the vanishing of these c"J which have been previously unde.
termined.
The
which
consist of
two
appear because of relations /( the latter sort can not be collected, unless the numerical values of the p, have been chosen; but their possible presence will be shown to be immaterial
may
which arise identically by and its multiples, and those Terms of among the various
tf>
.
In equations (48a), using h = 3, the terms which reduce identically to constants may be selected by fixing upon some one Jt and expressing the Incidentcoefficients of cos w4> ti in x ul and w ui and of sin rrupJt in w ul and x lA
<f>
,
.
each x tM and w Un as given by (47) or may ally in a of several Fourier series in the various reality rearrangement (58), is The resulting complicated constant, involving several series in the F n (ev ) G and n (e tJ ), can be treated advantageously by expressing all the a n e u q and5in terms of a single parameter a, just as in (35) and (36). In the the coefficient of each cjjj, say becomes then a power series in a; tJ = 2' and it is found that u w and
it
series in
<f> ti
t ,
M = *+-. N
tl
t3
(j>i),
Mu = a"e
tJ
.N
,
ti
(j<i),
(59)
where the
tJ
stant terms.
Now
requires that the c% satisfy k linear be homogeneous, since the vanishing of all of the un-
to their values in
Type
I, viz.,
zero.
That
is,
y-i
2Jf*-0
=i, ...,*).
(60)
above, it is evident that the determinant of the is identically zero, so that any equation is a consequence of the others tJ and may be suppressed. Let the first equation be the one which is dropped. Then the remaining k 1 equations (i = 2, k) will determine the
M
c
lt
(i
A;
1>
>
order
1 in
the
l)
tJ ,
terms of c"J if their determinant A,, of uniquely is distinct from zero. From (59) the lowest powers of
in
PROBLEM OF k SATELLITES.
445
o in the various elements of any row of A B may be ascertained. Evidently the exponent i+2j6k takes its smallest value 4i+2 6k when j=\, and
its
laigest value 6i
value
6i+6
. .
6k
whenj = i-\-l.
it is
.
6j6k
takes
its
smallest
As
. .
power of a in any of the is4i+2 ,il, i+1, ,k l) u (j=l, 6k, which occurs in n th and also in A Thus, in the r row of 6 (where t = r+l), the lowest & 2 in two columns, power is 4r+6 6k; and this appears for r=l, the first and (r-\-l) th in the r = k last where row, But, 1, the lowest power can appear only in the first column. Hence if the factor a4r+6_w is removed from the elements of the rth row (r= 1, k 1), a new determinant A 6 is obtained in whose first column the series of each element
M.
begins with a constant term, as does also the series of one other element in each row except the last row.
developed by the minors of its last row, it is clear that a constant term can not be lacking when the development is rearranged as a For the only element of the last row which can consingle power series in a. tribute to the constant term is that in the first column; and in the minor of this element constant terms are present in all the elements of the main
Now,
if
A,
is
diagonal,
likewise
A are
6
distinct
from zero
Therefore the vanishing of the Hfl = k 1) uniquely and homogeneously in terms determines the c"} (7 1, of c"J But this latter constant is to be put equal to zero by reason of the choice of the origin of time, as noted above in discussing the first-order terms.
for all values of a sufficiently small.
. ,
.
Thus the w ui (t) reduce to the values which they would have for orbits of Type I. Then also every PJJ = Q = 0; and the x ui (t) and wt.t( T ) reduce to the values in Type I, except for the c% which remain as yet k 1. undetermined for j=l,
Hence every
c"J
.
= 0.
In the next step these c% will enter the (by identity) in precisely the same way as the c'fl entered the Hfl, and must likewise vanish. Similarly,
+i
are set equal to zero, when they must vanish together with the +1 an d in to which will the meantime have rise. Thus the final they Q]"Vi> given
determination of the constants arising at any step reduces the terms of that order to the values which they would have for Type I. If various relations among the pt give rise to other terms in the
than those which are present identically in some one n even if these terms introduce into various elements lower powers of a than have been treated in A 6 these new terms can not affect the argument in general; if introduce into the development lower powers of a than were they for, previously present, with non-vanishing coefficients, then this new development is equally as useful as the former value of A, while, if they do not furnish
<fr
,
,
446
PERIODIC ORBITS.
terms of lower order, neither can they in general destroy the terms treated A6 since they must involve new @'s in their coefficients. Any cancellation could occur, then, only for a few special relations among the masses. The conclusion is not yet warranted that no orbits of Type II exist in Case I; but there can be none when a is below some ''sufficiently small" finite value, unless possibly for a few very special relations among the masses. Terms of any order. Case II. The differential equations for the secondin
,
are again (48); but because of the c and c the equations reduce to (57), as in Case I above, where, however, the //*' and ./' now involve them (linearly) and vanish with these two sets of constants.
order terms
x,,,(t), w,;,(t)
In order that x ta and w u2 shall be periodic, it is necessary, just as for 2D\', the coefficient of cosqj in the final differential equaType I, that E
$
i
It is equally necessary that J?-\-2H? the same equation, shall vanish. Further, the "secular must vanish. The H are free from the c%, just as in Case I terms" but they now contain the cjl and ej3, vanishing with the latter set. above; Now, from (54) and the fact that no relation (22) holds, it is evident
,
Ht
cj; also the and vanish with the c involve neither the c nor the c Further, since the x ui (t) and w>,.,(t) may be written
that the
,
.
f+2H$
x {T)=c
l . l
sin((] T
l
+ ?-)+c?^mq T+
i l
A^cosmr,
(61)
l
I Bfifanmr,
it is
are
J$+2H$, which are the coefficients of same way as the cf\ enter the E$2D$, which the coefficients of sin(</,T +*, 2). In the treatment of Case II for Type I it was shown that, when A?^0, the
found that the c% enter the
equations
E^ 2D^ =
conclusion
is
The same
admit a unique solution for the cj\ (.7 = 1, k). it valid and also follows at once that here; evidently
. . . ,
the equations
J%+2H $ =
( ,
(i=i,...,k)
(62)
= 1, admit a unique solution for the c In k), namely, (62), c, does not appear, being already zero; but this does not affect the 1 of the conclusion, since there are obviously no more solutions for k unknowns, after the last one has been determined, than there are for all k.
The
c0.
H now
The
c"J
,
vanish identically.
cfl
,
are
now determined
step,
as for
Type
but the
remain
PROBLEM OF k SATELLITES.
Case I of Type remain undetermined. c%,
(58), as in
II,
447
the
In getting the third-order terms certain constants will be determined: and J*g cfl and c% by the vanishing of the % respectively; and the c^ by the vanishing of the Hfl, which reduce to their values in Case I when the c%=0.
E^-2D^
+2H
c of any order are determined in the next order, Likewise, the cf n Each set is obtained from linear (n+1), and the cf n in the order (?i+2).
,
equations, whose determinant remains the same for each successive order. Thus, with the same possible exceptions as in Case I, it is impossible in Case II to determine the constants otherwise than as for Type I.
Remarks:
value of
a,
(1)
If either of the
still
there
may
determinants A3 or A 5 vanishes for some be no new values for the constants of integration
which would satisfy later conditions in subsequent differential equations and render the solutions periodic in form. Whether the series converge for this value of a would be unknown; so that the mere vanishing of A 3 or A 6 would not warrant the conclusion that orbits of Type II exist.
(2) The foregoing conclusions extend beyond a denial of the possibility of obtaining equations of periodic orbits by a certain method of analysis, or solving the differential equations in a certain way: the non-existence of a class of physical orbits of a certain type is asserted, though possibly there
numerous individual orbits of the type. arising from any set of Ac,, and ju [see (17) and
exist
can be represented by
parameters converging through the interval 5 r 5! 2t, provided that the parameters are sufficiently small. And from the existence of a class of periodic orbits of the type sought would follow the existence of a range of values of the Ac i} and n (including zero) satisfying the periodicity conditions. These equations obviously could not be satisfied for a range of
power
series in these
y. by arbitrary values of the Ac,,, and therefore they would define the Ac,, as functions of n, holomorphic for /x sufficiently small and vanThe substitution of these values of the Ac,, into the original ishing with fi.
values of
developments of the coordinates would render the latter series in /x alone, having properties (40a, b, d). If these series are impossible save for Type I, then there does not exist a class of physical orbits of Type II growing out of the circles named.
It may be inquired whether there out of circular orbits of an infinigrowing tesimal system which has no "grand conjunctions"; that is, whether there are periodic solutions of (7) when some of the X, have other values than are
Type
III.
possible in
Type
I.
Let the initial conditions be (17), let some instant when t is at an selected of let the be as the and of time, apse origin origin longitude be the
148
PERIODIC ORBITS.
u> = t=0
apsidal position of
both for n
and
for
n^O* Then
= 0, and
and /xj and the conditions for periodicity are (13). admit two integrals, an examination of which shows that But equations (7) = x' are a consetwo equations of (13), namely zt (27r) =xk (0) and x' t (2T) t (0)
identically in k
,
quence
(a)
of the other 4A
2 equations.
Hence equations
T
(
(13)
become
= e,(l
cos2<7,7r)
+juX
(i0
-!-e
- sin2^ 7r)
(
+^Un,X?>+eX,Z+o>,XZ+T / X
(&)
<
}+.
= e,(g,sin
2q ir)+iJ.X'j
i
+c T
1
i
(
(q
\-cos2q
ir)
+^\^hxy+e,x:+u>,x:y+r,x' y}+.. J
t
r-i
(c)
cos2g,ir)
(63)
(d)
= e ( - 2q T:- cos2q
t i
tir )
+nW' +e T
i<lt
(2q sin2g,7r)
i
+2\to,W$+e,W'
i
+t*/
W%+T,W%}+
(-l
*-i),,
where
X, =X,(0;2t)-Z
(0;0),
pr;,
2
=T7;(o;27r)-i7;(o;o),
;
X^ = X,(/;2tt)-Z,(/;0),
and the
here.
Tf,'!
;=^;(/;27r)-TF;(/
o),
(j=l,
3, 4), etc.,
*In treating orbits of Typo I, the instant of a symmetrical conjunction was regarded as the beginning of a period and was taken as the origin of time; but this was merely for simplicity, since in periodic motion any other instant could be so regarded. By taking as t = the instant when Mt of the infinitesimal system is at arbitrarily selected longitude, and choosing that longitude as a new origin of longitude (so that X* = 0), it is clear that the longitudes which the other infinitesimal satellites have at that instant constitute a set Each family of orbits of Type I may thus be said to arise from of X's, not all of which are multiples of ir. any one of an infinitude of sets of the X, other than multiples of ir, though all the sets have X* = 0. By reason of the u>i (t), which are in general distinct from zero, the absolute longitude v ( of any finite satellite at the new t = would vary with n) (but, since families of orbits of Type I exist for every position of the line of conjunction, one may obtain a family in which the initial longitude of Mt is zero, identically as to it, by selecting orbits for different it a from different families, taking the conjunction line as needed for the it used.) For such an origin of time Mt would not in general be at an apse for r = 0. Hence if the origins of time and longitude be chosen at an apsidal position of Mt for all values of n, the sets of X 4 other than multiples of x, which can give rise to orbits of Type I, are largely excluded. (Whether any such sets remain, depends upon whether, in Type I, Mt has any apses other than at the symmetrical conjunctions of the system; and this has not been ascertained.)
PROBLEM OF k SATELLITES.
(
449
,
While the determinant of the linear terms of the An u e t) u> and r in is zero, yet, when q t is not an integer, solutions of (63) (c) and (d) exist for the Aw, and e as power series in n and the w, and t\. Properties of the series (63), similar to (19), are easily established, which show that the solutions for the An and e and also the series obtained by sub(
(63)
stitution of these solutions into (63) (a) and (b), carry n as a factor. this substitution and a division by ju, (63) (a) and (6) become
After
+r/[+^r
(6)
ii/
]}+..'.
(64)
=^+2(l-co^)^+|{^[^
sin 2q l7r
2(l-cos27
7r)
m w "'' J
,
-i
',
sin 2g (7r
2(l-C0S2q T)
i
Mi \ W " *' J
,
'l-cos2<7,7r
in (64a) vanishes, since for q t not an integer no relation If any of the (22) holds; the constant term in (646) reduces to 2q Cu0 = = = 0; hence w T distinct from not satisfied are are /i by ( zero, (64) C,,
. t (
holomorphic functions of m vanishing with n, or w, t, orbits of the periodic type sought do not exist. The necessary condition for periodicity, namely, that the C (i0 vanish, is, by (28),
exist as
and
do not
2'
(/
2 e m (,>imn(X,-X,)=0
(-l,
k
ifc).
(65)
Of these ^-equations
in the quantities
\ = 1,
is
.
. .
1),
one
is
evidently
a consequence of the others; for, before X t of the C,, with respect to the X,(t'=l,
zero.
i = 1 be
put
k;
j=l,
k) is identically
Let the equation for suppressed. values be assigned to all, save one, of the X, the last X can still be given an infinitude of values for which any one C,, is distinct from zero Whether hence equations (65) impose very special conditions upon the X, there are any sets of X's, other than those of Type I which satisfy (65), is unknown.* It will, however, be shown that there are no others "in the
If particular
,
;
is
sufficiently small.
fully as severe here as in
sets
seem
Type
I.
450
If
PERIODIC ORBITS.
the
are developed as power series in the \ t of the linear terms are simply [dC^/dX,] for coefficients
CtA
from
(28),
[i&'L
=* l/ Swe.(, /)cosm(,/
(
-;>
(
(./Vt),
(66)
)7r
(jVi).
Denoting by ducing a by
the coefficient of X, in the I th equation of (65), and introIt is found (35), the S u are obtained as power series in a. that the lowest exponent of a present in Sif is (fi) if f<i and is 5(f i) if
Sf/
f>i.
the
(1
fifth.
Hence, of
th
;
all
the
S(f
a
(f<i),
Sn
power
of a,
and
of all the
if
S tf (f>i), S ul+l
. .
be removed as a factor from all the S{/ k), a determinant A, is obtained (equal to k1; (/=1, the determinant of the coefficients S t/ multiplied by a power of a), in whose A A row all elements save those of the first and r* (r+l)' columns begin with Therefore A7 is of precisely the same type as A, and the disa power of a. cussion of the latter shows also that A7 (and hence the determinant of the S(f) is distinct from zero for all values of a sufficiently small. Therefore = l, k 1) exist such that equations (65) are not quantities A,(/ Jf ir\ <A, except X7 =0 or ir. satisfied for any values of the X, for which |X r Moreover, the existence of a set of X r satisfying (65) would not prove the
Consequently,
.
u_,)
i = 2,
existence of w,
and
t,
as functions of
the
moment
that orbits of
Type
If it is assumed for satisfying (64). III exist, and the method of construction is
/x
examined, equations related to (65) are encountered. Thus, in finding the second-order terms, the must be made to vanish, and each of these involves
X,
When
qk
is
an
integer,
somewhat
is,
differ-
same
difficulty
however,
(22) holds.
219. Concerning Lacunary Spaces. The relation (22) may be expressed in the form J(p f p l )=p ,+qk hence this relation can hold only if qt is an
l ;
But the converse holds true only when k = 2. For example, integer. the following selections qk is an integer, but no relation (22) holds:
ft-2,
qt
P*-.
in
= 3,
= 3,
Pt-i-5,
p,
= 2A:-l,
etc.
= 2,
p*-i
P-t-7,
...,
p = 4fc-l,
1
PROBLEM OF k SATELLITES.
k
451
When k = 2, (22) always holds* if q is an integer; for then p, = l (otherwise T would not be the smallest synodic period of the infinitesimal system), and hence Jj) = q is satisfied by giving J the integral value q Since n T = 2q ir, the case q an integer is the one where the consecutive
i
conjunctions of the infinitesimal system occur at the same absolute longitude; and, denoting the synodic period of the two infinitesimal satellites M, and M,
by Tfl
since
'
na
or the consecutive conjuncoccur all in the same absolute
when
(22) holds;
and
longitude.
all
Moreover, since all the w, vanish at the beginning and end of each period, the "grand conjunctions" of the finite system occur at the same longi-
tudes as those of the infinitesimal system, and intermediate conjunctions of any finite pair occur at very nearly the same longitudes as those of the
corresponding infinitesimal pair. These facts suggest a physical reason for the non-existence of periodic orbits under certain circumstances. The greater part of the mutual disturbances of two bodies occur while they are near conjunction; and, if the consecutive grand conjunctions occur at exactly the same longitude, the perturbations of the elements would tend to be cumulative. Nevertheless,
if
bodies, the mutual disturbances may so balance to each other as yield periodic orbits, especially if the bodies are far apart But if two bodies have con(i. e., a sufficiently small) unless (22) holds.
there are
junctions between the grand conjunctions, all occurring very near the same longitude, the other bodies can not counterbalance the large perturbations of
the two.
exactly, there exists a range of values of the masses and the e including zero, for which periodic orbits are impossible; so that, unless the orbits for n = are eccentric rather than circular, there are for small values of
More
H no periodic plane orbits of k satellites when (22) holds. Ho far as this result extends, it would indicate that no asteroids having nearly circular orbits would be found, whose periods compared to Jupiter's are in the ratios
ratio should
\, f,
etc.
in
any such
be found subject to very great perturbations. known that lacunary spaces of the sort just mentioned do occur among the asteroids. That there are such spaces also when the ratio of the periods is f and other such values, is not surprising, as in any case the slightest deviation from the correct initial values destroys periodicity, there being Poisson terms in the solutions.
It is well
,
*In Poincarti's discussion of the problem of three bodies, therefore, the case where q k
is
an integer
arise.
452
PERIODIC ORBITS.
Of Jupiter's longer known satel220. Jupiter's Satellites I, II, and III. almost exactly in a plane, apparently in lites, the innermost three move and their masses with periodic orbits having symmetrical conjunctions; for orbits of Type I the respect to that of the planet are very small. Since
increase in the longitude of M, during a period is independent of m, being equal to n T, the average angular velocity of each finite satellite for a period
x
may be
taken as the corresponding n. The unit of time being the sidereal day, and the unit of mass being the mass of Jupiter, the observational data are :*
= M, = 0.000017,
Xj
7T,
X,
= 0, = 0.000023, = 1.769322711,
.
0,
.
M=
3
000088,
(67)
,= 3. 551552261,
w,= .878207937.
Since (w,-n 1 )/3 = .891114775, and n t -n3 = 891 114774, then, of (67) satisfy, far beyond observational accuracy, the equations (2), where v = .891114774.
is
= 3,
= T + 2t,
= l, =
7T
q3
= .985516077, =r
>
+ 3t,
<As3
so that satellite III advances 354?785788 during each period, while satelIf n lites I and II advance 1080 and 360 respectively more than this.
is
= .17,
ft
= .23,
ft
= .88.
seems desirable, however, to retain the fts in the computations, inasmuch as a new determination of the masses may render it necessary to use other values than those given above.
It
The
form
and
by writing equations
(43) in the
'#2 m
=l
Csin>/^, =0,
(69)
= "<r'2 C'cos//^, 0,
r,
= 2a u F
-I'\,
K =^-+F -2F,.
)
V"=\+2a
n,
IJ
F -(F +2F
l
3 ),
^^rO,
(m>
Tisserand, Traiti de Mtcanique Celeste, vol. 4, p. 2. The and 50?3 1760833) are here reduced to circular measure.
PROBLEM OF k SATELLITES.
Evidently the
etc.
453
From
(23), (5),
enter respectively the Dff, and Dg, relation {ajatf = (%/nO*, the e and S are found
D*
to be
m - . - .6284333,
= .3939606,
/S x
,
(n
^a
Suv\ 3
= 3-1366 ^
m
5
- e = .6268932,
32
= .19132 ft
.30443 ft
(70)
.
= j 2326 ft
and
= .77464 ft
^=
2
etc., encountered in the higher orders are the tables of coefficients given by LeVerrier.* readily computed by using In obtaining the successive A and Bjf from (52), the smallest divisors
Xj ),
The Fm {t
also the
Om(eJ),
-q\, or .1156616, .05772591, and .02875806 These divisors decrease materially the effectiveness of the respectively. small value of /x; nevertheless the terms above those of the second order
4-^, and
seem
of
/x
unimportant and will not be computed. and w,(t) are found to be:f
The
coefficients
^i,iW
= (-3ft-.lft)-.8ftcos2r-.lftcos3r-203.7ftcos4r
-K.7ft-.2ft)cos6r-.2ftcos8r-hlftcosl0r+
.
Wu(r)
+.3ftsin2r+.4ftsin3r+406.3ftsin4r
u\Jt)
ar,. t
(t)
(.4/3,
+ .lftcos4T
w
3,1
(t)= -55.2ftsinr-.8ftsin2T-(.4ft
+ .2ft)sin3r-.lftsin4r
E
are found from (54), where it 5 is convenient first to rearrange the coefficients of o^? and <r~ according to The x ul and w ut involve all multiples of r; but, as they multiples of r.
*
vol. 2.
Supplement.
regrets to state that the values formerly published {Transactions of the American Mathematical Society, vol. 9, pp. 29-33) are practically all erroneous, the factor q t of the last terms of (46) having been overlooked in making the calculation.
fThe writer
454
carry the factor ju racy in the x x and
, t
,
PERIODIC ORBITS.
1
save the following terms fall below the limit of accuTo facilitate comparison, the second-order w,,, above.
all
.
They
are
-.6^+(.1|8 j9;+.1^+.1AA)cos2t
1
= .30,0 2
sin r
- 18.3/3,0,) sin4T+.l&jS
.0001.r,,,(r)
sin6T+5.1$sin8T
(.20
+.60,03 +.50
2
3)
+3.10 )cos2t
,
-(.20I+.10,0,+60,03 +.50Dcos4t+
(.20,03+.20,0s +.40Dsinr-(12.505+25.O0,0,+25.10,03
.0001 x m {t)
= - (2.30,0,+2.50
-.10,0,cos7t+
+3.90,03 ) cost,
.10, cos2r
.OOOlw3.,(r)
=
<
(4.70,0,+4.90l+7.60,0,)sinr+.20, sin2r+2
Since
?!
= a (l+z M M+z,.jM
t
and
v
t
ii
the radius vector and absolute longitude of each satellite are obtained by computing the a from (3) and using the coefficients above; the deviations from their values in the undisturbed circular orbits are given to five significant
figures, so far as the
terms of the first two orders are concerned, How much by terms of higher orders is unknown; in fact no proof has been given that the series converge for /x = .0001, although they have been proved convergent for all m sufficiently small. To show the general shape of these orbits, the values of the i\ and r, a will be given to four decimals, using the values of the 0, tabulated above:
these would be affected
{
r,/a,
= l-.0044cos4T,
= ir+3.9855T+.0089sin4r,
r,/a,= l-.0093cos2T,
vt
0185 sin 2 t
0006 cost,
vt
these orbits are thought of as ellipses rotating in the plane, the major semi-axes would be the respective a the several eccentricities would be .0044, .0093, .0006, and the axes would rotate forward at rates
if
Hence
The three satellites are in line with Jupiter whose average values are the n the and middle of each at beginning period, II and III being on the same side = = w. Whenever of the planet at t 0, and I and III on the same side at r
.
PROBLEM OP k SATELLITES.
II is in conjunction with I or III, the inner of the pair
is
455
near a perijove and
the outer
is
its
X, +q r, respectively) largest departures are for satellite II, as r2 /at reaches a minimum of .9907 at t = and a maximum of 1.0093 at about t = 7t/2 and every half-period thereafter, v t meanwhile ranging from 64' more to 64' less than the mean longitude of II.
. l
The
Similarly, satellites I and III get 30' and 4' respectively ahead of and behind their mean positions, and the r /a, at such instants closely approximate their mean value, unity. For satellite I the maxima of r/a occur at intervals of
{
a quarter-period, and for satellite III they occur at intervals of a period. Finally, it may be noted that, for this system of bodies, the increments Ac tj (see 209) which have been given to the initial values of the coordinates
and
are finite
are approximately
r
t
456
PERIODIC ORBITS.
This substitution requires the flattening to vanish with the masses so that the central body becomes spherical if the others t become infinitesimal; but the amount of flattening corresponding to any given set of finite masses remains quite arbitrary, even if the p and q k are specified; for the values of the a merely determine the ratios of the 7,, and one 7 may be taken at pleasure. In the solutions of (71) satisfying initial conditions (17), the terms independent of n are the same as formerly. Hence in Case I where qt is not integral, the An, and e, still exist as convergent
M ,,..., M
n satisfying (29 a, b), though of course their values in terms Thus periodic orbits exist. of n are now different because of the 7, In case q k is an integer the 7, enter the x,(0; t) and x,(i; t), but do not Thus the argument in Case II is likewise appear in the Pt) or P (/ of A 3 and orbits of unaltered, periodic Type I exist under the same conditions
power
series in
as
when the bodies are all spherical. The numerical results for Jupiter's
first
satellites
The
corrections to be
to x u (t) to .0001 to .0001 to x2fl (r)
xlt2
4t
1.8
&7, cos 4r
7, sin
wM add
add
-3.6
At
.l7 2 cos It
(.2/3 1
to.0001a-2i2 add
to .0001 u?
+.4/33 ) T2 cos 2r
add
add add
-(.5/3,
8/^)7* sin It
tox3A (r)
to .OOOLr,,, to .0001 ir
.17, cos r
.l/3 2 .l/3 2
7 3 cos
M add
3
7s
sin r
And
since 7,
t>,
a,
{
At,
to r2 /a2
to v2
2t,
then
r,
a,
r2 /a 2
^ = 7t+3.99855t+.0059
et
sin At,
+ .0001
r, a,
+ .0006
cos
t,
v3
Chapter XV.
CLOSED ORBITS OF EJECTION AND RELATED
PERIODIC ORBITS.
In the problem of two bodies there is in no sense between circular orbits revolving in the forward and retrograde continuity directions, except where their dimensions shrink to zero or become infinitely But in the restricted problem of three bodies the deviations from great. the circular forms of the orbits are such that there is geometrical continuity in some classes between those which revolve in the forward direction and those which are retrograde; and the limit between the two types is an orbit If the infinitesimal body leaves passing through one of the finite bodies. one of the finite bodies, its orbit is called an orbit of ejection; and if it strikes a finite mass, it is called an orbit of collision.
222. Introduction.
In certain cases orbits of ejection are also orbits of collision, or closed When the direction of collision is exactly opposite to orbits of ejection. that of ejection, they are the limits of two classes of periodic orbits, in one
of
is direct and in the other of which it is retrograde. of closed orbits The ejection are not themselves periodic orbits, even if the physical impossibility be disregarded and the problem considered purely
of view; for, if the expressions for the coordinates are followed, in the sense of analytic continuity, beyond the values of t for which a collision occurs they become complex, and never become real
collision are
again for increasing real values of t. Those orbits in which the ejection and not in opposite directions are not the limits of periodic orbits,
or at least of orbits which re-enter after a single revolution. The object of the investigations of this chapter is to determine the
limiting types of certain classes of periodic orbits, and thus partially to prepare the way for the discussion of the evolution of the various classes of
periodic orbits with varying values of the parameters on which they depend, and to show the relations among these various classes. The existence of
the closed orbits of ejection will be established, some of their properties will be derived, and it will be proved that each one in which the direction of ejection
and
two
223. Ejectional Orbits in the Two-Body Problem. As preliminary to the general problem, the special case in which there is only one finite mass
be treated. Let the mass of the finite body be be so chosen that the gravitational constant is unity.
will first
n and
let
the units
of
458
PERIODIC ORBITS.
the infinitesimal body projected along the fixed -axis satisfies the differential
equation
d?
de
= -\~n +
,,
-T'
is
(1)
or + according as the motion where the sign is direction from the origin. at t = t Suppose f = f and d/dt = $' = equation (1) becomes
.
in the positive or
negative
Then the
first
integral of
(|)
'
= r=
has a
||
2(l^)
finite
2(l^)
+^ =
which
2(lp)
'
+Ci>
if c, is
'
(2)
If c, is negative, ||
maximum
for
vanishes;
zero,
approaches zero as
is finite positive, is negative in order to get orbits without loss of generality, it can be
large;
if c, is
is the greatest value of for projection in the positive direcsupposed that Then tion, or the least for projection in the negative direction.
-0,
*-+'*&=**.
6b
(3)
With the
initial
values
(3),
vb=H*-i(-i+D
where j
(1)
it
^^
(t
('-'.)
G+>H
is
an integer.
consider
as a function of
Now
and
values of
follows that
a regular function of t for all values of t except those for These values of t are easily determined from (4), and are t tx \ /
The
character of
(t
1 )
in the vicinity of
tj
is
easily
determined from
V+ =
i7,
expansible as a power series in and the equation can be written in the form
left
is
The
member
It is
found that
#-**,
g-*
. 1
-=o,
,
^).-4(1
Therefore r? is expansible as a power series in (t tj) '3 , starting with a term Vi = *7 2 it follows that is expansible of the first degree in (t Since t) as a power series in (t ti) '*, starting with a term of the second degree in
l/i
(t
tj)
t]
It
fore
is
an
easily seen from (4) that F(ri) is odd series in (t tj) l/3 , and is an
is
of
?j.
There.
l/i
tj)
Since
by
(5),
is
i\
and
^lUJ-^/f^- )'
459
efficients
tj being known, the cocan easily be found from (1) by the method of undetermined coefficients. It is convenient in the computation to let
The form
of the series
r-it-w,
after
(6)
which
(1)
becomes
The
= a T>+a T + Z
i 2
i
+au T +
in
By
direct substitution
|_>
>
and comparison
" 7
of coefficients,
found that
J'
(9)
t S
= ct Ct
1 I
+
I
clt u.i
a t H u '
I
"3 63
'
a t
1894
4851
a r H
3293
j^i-r
^7007
a r
-(1
/x)
a = arbitrary constant.
for use
from
(7).
can be developed by eliminating the After the transformation (6) the integral (2) becomes
=.8(l-)(!-i),.
On
using this equation to eliminate
2
from
(7),
the result
is
found to be
Now
it
and
(9)
that
=cr[l +
ar
+ 2 a^
= j 2
2y
]'
2
g=2cr[l +
2aT 2
+20'+lKT '],
2
^=
2
2/
2J
+a f* (2; -j + 6)aw _
2
j;
+S{So-*+i)'(2y-2i!Hr i)<ilA,.ItV],
00
2;
+*2 (i+^vZ+Sflo'-Hi)^-^]J=3
;=i
*
*=2
(gy =2cV [l +
+4a
4ar 2 +4aV
+ 2 jg (i+l)a
j=4
L
A
2;
r^
2K-
2J
2{S *+i)c/-*+i)<v*W=2
J
4G0
PERIODIC ORBITS.
series
v Hence, on equating to zero the coefficient of t after these substituted in (10), it is found that
/-*
have been
i(2i+3)o,,=
-a(2f+j+2)a
_ iJ i
(11)
very simply for all values oij greater than unity. The result of applying (11) and reducing the coefficients to the decimal form is
coefficients
].
terms alternate and a general property which will be needed in establishing the existence of the closed orbits of ejection in the problem of three bodies (230). It follows at once from (11) that the part of the coefficient Ojy which comes from the first term on the right is opposite in sign
this solution is written the signs of the
atJ has
a factor.
This
is
to the coefficient of the preceding term. Since the sum of the subscripts of the product terms in the right member of (11) is 2j, their product has the same sign as the coefficient of a,,_ 2 Therefore, a,, and Ojy_, are opposite in
.
j.
a,,
The
unite.
expression ~ for
'
If
appears in aw _ 2 has a branch-point at t = t,, where three branches v =r w+in* the three distinct branches are = pe 1]
)
*-* V=, *i
If
W^ ^,
1
.
&^*
v=t l ttH
*\
(12)
and negative, <p = ir and the second of these expressions alone is real. Iittj is real and positive, <p = and the first of them alone is real. Consequently the analytic continuation of the real branch of the expression for through t = t} leads to a complex value, and this value remains complex Therefore the motion is not strictly as t, remaining real, increases to +
ttj
is
real
periodic.
not one of ejection the branch-points of the functions which express the coordinates in terms of the time are not on the real axis. If the major axis is kept fixed, and if the eccentricity is varied from zero to unity, an examination of the equations from which the character of the functions can be determined shows that the singularities start from both
is
positive
lines
and approach
approaches unity. At these singular points two branches of the function permute except when, for e = l, two of them have united on the real axis, and then three branches permute. If e increases beyond unity, each of the branch-points divides into two which move equally
the real axis as a limit as
in opposite directions along the real axis,
and
for e
= oo
461
The Integral. Equation (2) holds for all values of t, and therefore the series (9) is substituted in it the result is an identity in r. The conditions that the coefficients of the various powers of r shall be identical
wHn
in the left
tests of the
putation of (9). The integral also gives the relation between the arbitrary a and the greatest distance By direct substitution, it is found that
The
and
interval
from ejection to
collision is
(5), (9),
(13) to be
3 '' 2
" P=
KJ-Jq)
w)
orbit,
(14)
V2(1-m)
\/2(l~-V)
is infinite.
225. Orbits of Ejection in Rotating Axes. In the problem of three bodies the motion of the infinitesimal body will be referred to rotating axes.
In the demonstration of the existence of closed orbits of ejection in the probit will be necessary to use some of the properties of the orbits of ejection in that of two bodies. For this reason the orbits now under consideration will be referred to axes rotating uniformly with the period 2r. Suppose the ejection takes place at t = ^ and along the x-axis, where the
by x and
y.
Then x and y
(t t,).
are given
x=
+cos(
Q,
y=
sin
(15)
Those
orbits
which re-enter
are of greatest interest in the present connection. The condition that they shall have this property is that their period in t from ejection to collision shall be a multiple of 2w. This condition becomes, by virtue of (14),
(y
where j
is
3/2
= 2iV2TT=^),
(16)
a positive integer.
Figs. 15, 16, and 17 show the curves for j equal to 1, 2, and 3, at least as to general form, in full lines for ejection along the x-axis in the positive direcThese curves tion, and in dotted lines for ejection in the negative direction.
for the three values of j are not
drawn
same scale, for it follows from (16) 2/ One of the important proportional to j
to the
*.
4f>2
PERIODIC ORBITS.
The differential 226. Ejectional Orbits in the Problem of Three Bodies. equations of motion for the infinitesimal body when the finite masses describe
circular orbits are, in canonical units,
<Fx
2 dy
dt
]
= dU
dx'
(t
d*y
dt
2
df
= dU ^ ndy dt
,
dy'
(17)
2 2
V=\{x>+y>)+
+r
-
n=
(x+fx) +tf,
= (x-l+M) r|
+2/
'dxV.(dl\>
u +<D--*
_,
(18)
463
x+(j.+x^iy = (p+V^Tq)
r
2
,
uV^iv
dt=(p +q )da = p
2
2
(20)
2
da.
dp da
//
dH
du
dq' =
dH
dv
2
2
da
2 2
du = da
p)
dH
dp
dv
dH
dq
2
da
6
=
i|(
W +2p
2
5)
+(?;-2p
r2
(21)
V = I -p +9=- M '2
It follows
=l-2{p*-q>) + (p*+q%
p = q = \ix+n = y = 0, and that u and v are finite for r, = 0. Therefore the form of H shows that the differential equations (21) are regular in the vicinity of p = q = 0. Consequently, p and q are developable
from
(20) that
Fig. 16.
Fig. 17.
It follows from (20) that x, y, and vanishing with a. tti, considered as functions of a, have the form (19); and therefore x and y expressed in terms of r, defined in (6), have the form
as
power
series in a,
x+n = c[r*+ay+
solutions in the vicinity of
(t
t 1 )'
/
].
y = c[b2 T*+b 3
T+
The
(22)
character of the
'= r
obtained without difficulty from (17). variable and expanding the expression for
T
these equations
2
become
drx _ dx Z
dr*
dr
UT
[(x+nY+yT
5
+9 M T [2(x+M)+30r+ M
T
2 )
-fi/
]'
|
(23)
Z
dr*
dr
+bT dr
TV
[(x+tf+yT*
W T VL
+ 6 W+U +
Mil
PERIODIC ORBIT&
It will
now be supposed
initial
is
Therefore the
x(0)+/x
conditions are
=>
g:t
"
F L
CT
const.
(24)
With the
erty of symmetry.
conditions (24) the solutions have an important propLet them be written in the form
x+ m = t7(t),"
Now make the x M umX
?/
= t<<7(t),
^=Mt),
jjjU(r)'.
(25)
transformation of variables
v=- v
t--t
<k__dxi
dy_
dy.
Equations (23) are not changed in form by this substitution. Therefore the solution of the transformed equations with the initial conditions
*,(0)
- yM
=^P=o,
p^f|
=a
are
Xi-tJ/W,
where
/,
<?,
fc-tfcfr,),
4>
^-^W
T<p(T)=
fr^W,
by
the
1
^,
and
same symbols
T*f(T)
A
in (25).
l
Therefore
-T
<p(t
= +T<p(-t),\
(26)
tV(t)
= +t
3
1
+(t
)= -tV(-t).J
x and dy/dr are even functions of t and that dx/dr and j/ are odd functions of t. The first equation of (22) contains only even powers of r, and the second contains only odd powers, starting with a term of the fifth
It follows that
degree as the lowest. With the initial conditions (24), the solution of (23)
is
found to be
(x
M)=c[r
+ar
s
-faV+[-i+|a + | M ]r
3 7 3
+---}'
j-
y=
c=
/*>" +
u)l l/3
'
= arbitrary
constant,
in the left
members
according as the initial projection is in the positive or negative direction. The constant of the integral (18) is given by the equation
[()+ $)>(*+*>
(28)
+
l(-t-
+ M) +
2
l/i
Z/ ]
[l-2(x + M) + (x + M)
+ rJ
465
from
member of
this equation
can be developed
c= cT->+c +c T*+cy+
2
Since
2
this
i
equation must be
l
C_ = C = C = C =
6
=0.
a check on the computation of the coefficients of (27) of (27) in (28), it is found that
By direct
substitution
C= C
The
force function
is
=-fc a+ M (2+M).
in the
(29)
sometimes used
symmetrical form
F=(l-M)(r1+^)+M(r1+^)=t/+M(2-M),
instead of in the form given in (17).
Then
2
the constant
C becomes
(30)
C=-?c a+3 M
228. Recursion Formulas for Solutions. The second terms in the right members of (23) give rise to a large part of the labor of constructing the solutions. They can be eliminated by use of the integral, and relatively
simple recursion formulas can be developed for the construction of the solution after the terms of lowest order have been found. The integral (18) becomes in the notation of (23)
(31)
+ (z+m)
On
}-9rC.
multiplying the first equation of (23) by x+n and the second by y and adding the results; and then multiplying the first by y and the second (x+m) and adding the results, it is found that by
T( , +M)
^ +T,^_ 2(x+M) g_
=
9r-[(,+M)
2
3
+^]-
[(
^ffr
2
+ 9 M t J2(x+m)
5
-2/
+3(x+m) -!(z+m)2/
"}'[
9M t5
2/
J+3(x-+m)2/+6(x+/x)
466
PERIODIC ORBITS.
On
eliminating the second term of the first of these equations the integral (31), the simplified equations become
by means
of
fr'[(
I + (I )'+
,)
+4(x+M)"
(33)
=
The
solution (27)
27M t5
{(x+m)2/+2(z+
m ) 2/+
may
oo
z+M = cr [l+
s
2a
2 ,r
']>
y=
cA-l +
^]'
(34)
a,
bt = a = arbitrary constant.
The next
step is to form general expressions for the terms involved in (33). Nearly all of the terms written are of the second degree in x+M and y; those which are of degree higher than the second are all multiplied by the factor
H and
be in general of little importance. They will not be included formula and must be added to it when it is used. Since these terms contain t 11 as a factor, they will not contribute much to the solution unless it is carried very far. The general expressions for the terms of the second degree in x+ix and y are found from (34) to be
will
in the general
2J
^(k+l^k+Da^^A,
b 2i - 2t - 6
rt/^ 2cv{lOr
^(2f-3j+U)b _y +
1
2J
J ^(k+2)(2k+5)bu
(&+l)a 2t a 2,_ 2t r4>
A'
2J
2/
+ JJ JJ
2;
_ c r^+
%^{2k+b)b
2t
hv -n-A<
6r 3 (x+ M
)g
= -6cs
r'[-5T
-2[5a ,2
-(2y-l)62J _ 6]r^
GO
J-4
-.
>-5
t-1
467
Gr'yf
-
2 [0'-2K_ -&
6
2 2 ,_ 6
J=i
]r
'+
2 2 (k+l)au
J=i
b21 -*-*A>
*
t=l
f (1+2m)t
2 (x+m) = -
y f C (l+2 M )r {r +22^-eT +
2 3 6
J IX^-e^'}'
f (l-jtffV-
fc
(1-m)t {t
I2
-22Vi ^+ S S^V-a^}'
2
(f-Y -2cv{l+2
"
J 0+l)a ^+ 2 2 (*+DO'-*+i)^-T*'i
2
i=l
j=2
Jt=l
(g) =|c
r'{25r
-102(2i-DV
+
=
22
2 ^+ 5 ) (2i-2A-
i)&A--eT
2'
[;
y$*
2cv{-l-
2 [0 + l)(2i+l)a
w -62;
]r"
!t
+ 2S(Hl)(2Hl)a
r(x+ M
BT
u }
)
=1
>
j-i
( fc
..
22
;=2 t=i
t-1
+ 2 ) (2A;+5)a
2;
_ M 6 2 ,r^
-
/-I
2 ^+1)VH
J
2t
6 w _ 2t _ 6 r4.
2(x+ M
)^=2cV|-5-
2 [5a ,-(2i+5)6 Jr H 2 2(
2 2 2 2;
2 *+ 5
K-
6 T
*}'
-6r (x+ M )
^ = - 12cv{l+ 2 0'+2)
6
2J
2 2 (k+\)a
J-I
ik
av _ H
A,
2[^-*-^-sV+ L J i=4
22^-- t4'
6
-l
468
PERIODIC ORBITS.
substituting these expressions in (33) ' and tw+ respectively, it is found that
)-\
On
t v+s
coefficients of
ls
-^[(2k+5)(2j+2k+3)b +12(k+l)an]bv _
tt
ili
1-i
1-4
+62(2fc+5)&2ta2v _, t _ 6
1=4
+f (l+2M)2 a**,--.
*-l
-\-(ln)b
*=i
ik
bt j_ u _ n
+ quantities
coming from
terms
_ ik
/-i
lt 6 kt
+ 122(*+Dua w -u+22J(*+l)(2*+5)a w _ M
,_4
12(i+2)&2,_ 6 +
62J(2*+5)6A--.
;-4
and
These formulas are to be used when j is 3 or greater, and care must be taken in adding terms coming from the higher powers of (x+m) and y in the
right
are
,
x+ii = rcosT*+x
i fx'
^63
4851-
a*T*+
'7007
aV
.
,
_ 2,418,092
3,972,769
55,964,945 6_12_l ~
T 67,540,473
_ 38,481,084,886
32,937,237,333
"
*"
'
"|
J'
(36)
^nT
_l_
/4392
380
27
Vv
"1
469
The Conditions
for Existence of
|o
,
Stries (34)
is
and
p.
The
coefficients of the
various powers of t are polynomials in a and p, so far as p occurs explicitly, and they also involve p implicitly through c. These coefficients are expansible
as
in a
o
|
series in a
/3
and
if
|/3|
^p,
z+p |p| ^p
^.R.
They may be
dx
written
x = Pi(P,h',t),
^=p,G8,m;t),
y=*p t (P,v;T),
dv *g -p(0,j!
r),
(37)
p t are power series in /3 and p. Now a will be determined so that when p From (5) and ejection to collision shall be 2jir.
where p lt
.
is
is
2/s
=-^(2j)-
(38)
and that for t = T the coordinates of the x' infinitesimal body are x y0) y' where the accents denote derivatives with respect to t. Suppose now < p < p and let the values of the coordinates atr = Tbe
Suppose
T<R
and p =
,
x=x
+p
l ,
x'
= x' +p
y=y
+p
y'
= yl+P>.
(39)
The
conditions that these values of the coordinates shall belong to an orbi t of ejection are
+&=pM p;
t
T),
T).
arJ+A = P.08,M;
Since the right
T),
y'o+P< = p
(40)
tf, n;
j
members
and
p, it
power
series in
/3
of these equations are expansible as converging follows from the definitions of x x' ya and y' that
,
, ,
0i
= g,(/3,/i),
. .
.
A = 9i08,m),
ft
= g,(i8,
m),
A-g
G8,M),
(41)
where
g4 are power series in /3 and p, vanishing with /3 and p. If the infinitesimal body crosses the x-axis perpendicularly at any time
?,
,
the orbit
is
definitions of x
symmetrical with respect to the x-axis. x'^ y and y' that, when p = 0,
,
,
It follows
from the
x'
(P/2)
= 0,
2/
(P/2)
= 0,
(42)
It will be shown that the period in t from ejection to collision. from when is distinct be satisfied conditions can p zero, and thereanalogous fore that closed orbits of ejection exist in the restricted problem of three
where
is
bodies.
470
PERIODIC ORBITS.
Suppose n
initial
is
distinct
,
In order to leave the period Zo+ft 2/ +ft 2/+ft introduced by the transformation arbitrary an undetermined parameter b is
conditions x +/3,
t
=t
(1
+ 8),
(43)
where
t, is
the
new independent
.
Now if the orbit for n = does the solutions can be expanded as power n
variable.
the moduli of
ft
ft
5,
and if Q is arbitrarily chosen in advance, and m can be taken so small that the solutions
is
T^t,(1 + 5)^Q.
The quantity Q will be taken equal to the period from ejection to collision when
M-6.
In order to complete the discussion in regard to the convergence it is necessary to show that none of the orbits in question for /x = passes through
the position of /x. Suppose r = t is the time at which the infinitesimal body crosses the positive x-axis on which, for it = 0, the body it lies at the distance
unity from the origin. It is necessary to show that for none of the values of defined in (16) is equation (4) satisfied by = 1 and t t^mr, where ^ is
the time of ejection. It follows from (4) that the larger is the shorter is the time required for the infinitesimal body to pass from ejection to the distance unity. Therefore, if for the smallest value of belonging to the problem, viz.,
|
|
reaches the distance unity in less than t, then it will always be at a distance greater than unity at t ^ = t and all multiples of w until it reaches the greatest distance Consequently, it can not pass through the
$
it
.
= 2,
point occupied by n while receding from 1 n; and since the path referred to rotating axes is symmetrical with respect to the z-axis, it can not pass through the position of m on its return to 1 m- In making the computation
it is
convenient to
let
p.
tt
equa-
tion (4)
becomes
<-<
1
=^[T-4V 7=7+2sint
i-
(-l+2p)](^ )
x
found from this equation that if = 2 the value of t t for = 1 is 0.187T, which is less than t. Therefore none of the orbits in question for /x = passes through the position of mIf the infinitesimal body is moving in an orbit of ejection and crosses the a;-axis perpendicularly at any time, then it follows from the symmetry of its motion that its orbit is also an orbit of collision. Therefore sufficient
It
is
dx
P
,
,
If
5
the initial conditions are x +ft ^o+ft 2/ +ft 2/o+ft and the parameter has been introduced by (43), these equations become
,
471
|^=P 03
1
,...,/34 ,5,m;/Y2)
= O,
y-P (ft,...,A,,>;P/2)-0i
i
(44) It
where
and P2 are converging power series in follows from (42) that Pj and P2 vanish for ft =
ft
ft
5,
and
/x.
=ft= = ^ = 0.
5
is
If ft,
ft are
an orbit of
ejection.
Therefore, upon substituting the series for these constants in (44), sufficient conditions for the existence of closed orbits of ejection become
^ =Q
where
(/3,
5,
P/2)
= 0,
5,
y=
(ft 8, m;
P/2)
= 0,
5
(45)
Q and Q
t
are
power
series in ft
and
m,
= n = 0.
The
coordinates can, therefore, be developed as power series in ft 5, and y. and the moduli of these parameters can be taken so small that the series
converge for
\t\
<P, where
The proof
of the existence of closed orbits of ejection resolves itself into the demonstration that equations (45) have solutions when /* is distinct from zero.
and 5 are both also zero, two bodies in which the because, period in r from ejection to greatest distance depends upon ft and in which the distance depends upon 8. Therefore equations (45) have one or more solutions for /3 and 8 as power series in n, vanishing with n, according as the functional determinant is distinct from zero or is zero for /3 = 6 = /x = 0.
These equations are not
satisfied
by
/x
unless
to that of
to
Since the functional determinant involves derivatives only with respect Then the /3 and 8, the n may be put equal to zero before forming it.
in question is
determinant
dx
A=
da'
dx d8 dy d8
0_Oo = 8 = 5=0
.
dy
da'
Before forming the elements of the first column 5 may be put equal to zero, and before forming the elements of the second column 0=a ao may be put equal to zero.
is
where x
It follows
from
(15) that
be.
when
=/x=
and
is
t1
= 2jw
the value of y
is
zero whatever a
may
Therefore dy/da
zero
becomes simply
A= (dx/da)(dy/d8).
t tx
for
formed
is
t-t
=T =T
3
3
l
(l
+ 8) =- P + ^(l 8)
3
=j*(l + 8y.
a
(46)
The second
it is
be computed first. Upon putting a found from the second equation of (15) that
factor of
will
3 3
= 8 = 0,
472
Before computing the
zero.
first
PERIODIC ORBITS.
the parameter 5 follows, from (46), (15), and (9), that
factor of
may
be put equal to
Hence
it
aT
da
It
da
in
^2i
aT
4851
OT
J?r P/2
Of
223 that the signs in this series alternate and that a is negative for those orbits which lie entirely in the finite part of the plane. Therefore dx/da is distinct from zero for all values of a under consideration. It follows from this discussion that A is distinct from zero for
was proved
/3=o-a=5=0,
t=jtt,
existence of closed orbits of ejection can be uniquely satisfied for |/x| suffiThere is a closed orbit of the type in question for ejection ciently small. in both the positive and the negative direction for all integral values of j
or a
of (16) depends.
In the special case in which the finite masses are equal, a closed orbit of ejection for j = 2, with ejection in the positive direction,* was discovered
from numerical experiments by Burrau in two interesting memoirs. f Since in his problem y. had the large value 0.5, it is not to be expected that the results of this analysis would agree very closely with the results of his computations. Hence the comparison will be made only for the constant of the Jacobian integral. Upon taking into account the difference in his units and those employed here, it is found that his Jacobian constant CB
,
equation
of (29)
by
the equation
it is
Burrau's computation gave-2CB =2.2528; and for /z = 0.5, t?i?=2V2(l-n) found that C = 2.38, and the agreement is fully as close as would be
It follows from these numbers that a larger value of the conexpected. stant a, corresponding to a smaller value of belongs to the undisturbed In the undisorbit having the period 2ir than to that computed by Burrau.
,
turbed orbit the greatest distance to which the infinitesimal body recedes is, t = t, and it is then on the 2; it has this value at t by (16), negative The greatest distance found by Burrau in his computation half of the z-axis. was 1.9972, or a little less than that in the undisturbed motion.
x
small value of
is ejected toward or from the body n with a be disturbed so that on its return it will revolve |, in the This can be seen when the motion around 1 /* positive direction. is considered in fixed axes, for under the conditions postulated the disturbance is positive all the time that the infinitesimal body is going out and If it is ejected farther, it will be accelerated by n in the negative returning.
If
will
Burrau's orbit of ejection was from the body called m here, but permuting 1 n and p and changing the positive directions of the axes, the statements are correct. fRecherches numcriques concernant des solutions periodiques d'un cas special du probleme des trois corps, Astronomische Nachrichlen, vol. 135 (1894), No. 3230; and ibid, vol. 136 (1894), No. 3251.
473
While
in general the
body
will
on
its
return,
it
may
George Darwin has discovered one such orbit by numerical experiment* having the period ir. The ejection was from /x in the direction of 1 n, and the body collided with /* going in the same direction. f This orbit is one of a
Sir
pair which together are the limit of certain periodic orbits, though they are not periodic themselves, either physically or mathematically. The constant
belonging to this orbit in the units employed here is 20/11 = 1.818. The values of the masses used by Darwin were 1 ju= 10/11, /x = 1/11. It follows
from
for this period, and from (30) that 1.716. In this case the belonging to the undisturbed orbit is larger than that belonging to Darwin's orbit. The value of is 2 1/3 = 1 .26 the greatest distance in Darwin's
(16) that
l/3
;
=2
C=
diagram,
is 1.3.
231. Conditions at an Arbitrary Point for an Orbit of Ejection. Since the motion of the infinitesimal body is regular for all finite values of t and all finite values of the coordinates except those for which it collides with
one of the
masses, { it becomes a matter of interest to determine in case whether the trajectory is one of ejection or collision for a any special finite value of t. It is sufficient, as Painleve conjectured and as Levi-Civita
finite
proved,! that the coordinates and velocities shall satisfy one analytic condition in order that the orbit shall pass through one of the finite masses for a finite value of t. This conclusion will be established here in a different way.
Suppose n is zero and consider the problem of defining the initial conditions for an orbit of ejection so that it shall pass through the point in question, and so that the components of velocity at the point shall satisfy as
conditions as possible. The velocity in rotating axes at any distance from the finite mass 1 n is the resultant of the velocity with respect to The velocity with fixed axes and that due to the rotation of the axes.
many
respect to fixed axes at any finite distance can be made any finite quantity by a suitable determination of the constant a, or the equivalent constant
.
Consequently, an arbitrary speed, or one of the components of velocity, with respect to rotating axes at any distance can be secured. Suppose the speed at a given distance has been assigned; then it is possible to determine the initial direction of ejection so that the orbit of
through any point having the given distance, for it is possible to do it in fixed axes and the rotation simply changes the direction of ejection by an angle which is proportional to the time required for the body to reach the distance in question. It is clear from this that when = the conditions of ejection can be so determined that the infinitesimal jt
the
body
will pass
*On certain families of periodic orbits, Monthly Notices of the Royal Astronomical Society, (1909), p. 134. fSee further remarks on this orbit at the close of 234. tPainlev6, Lecons sur la Theorie Analytique des Equations Difffrentielles, p. 583.
Acta Malhematica, vol. 30 (1906), pp. 306-327.
vol.
70
474
PERIODIC ORBITS.
Of shall pass through any assigned point with any assigned speed. the define an two coordinates and to four the orbit, viz., quantities required can be taken arbitrarily and the fourth three of direction and motion, speed
body
determined by the condition that the orbit shall be one of ejection. The determination of the fourth quantity is double because the body has the same speed twice, once when it is receding from 1 n, and once when it is returning toward 1fi. Suppose that, for m = 0, an orbit of ejection passes through the point I/3 = T. If r represents the speed x T y T with the speed v T = Vx't+v't at (t with respect to fixed axes, then, since the component of velocity due to the
is
'
rotation of the axes equals numerically the distance of the point from the origin, the relation between v T and T is
'
4-1+&
,
(47)
the greatest distance to which the body recedes, Equation (2) determines and (13) gives the constant a,,. Equation (4) gives the value of T, and the direction of ejection is T degrees in the negative direction from the line n and the point (x T y T ). Let the angle of ejection be O joining 1 Now suppose that n is distinct from zero, but small. Let the initial Let a new independent variable t values of a and 6 be a +/3 and o +7and a parameter 8 be introduced by (43). Then the solution can be written in the form
.
z=Pi(/S, y,
s,
r,),
y=p,(P,
y,
s,
m; n),
^=
where
p,
,
. .
.
Pi(0>7,
4
5,
m; rj,
series in
^=p
/3,
4 (/3,
t,
5,
m;
t,),
p are power
y,
8,
and
m-
The moduli
t
,
of these
parameters can be taken so small that the series converge for <! r ^ T. The conditions that the body shall pass through the point (x T y T ) with
the velocity v T at
Tj
= T are = 0,
T)-y r = 0, V^+f-v T = Q. (48) Since these equations are satisfied by = y = 8 = n=O, they can be written as power series in y, 8, and vanishing with = y = 8 = n = 0, of the form
Pl (l3,y,8,n;T)-x r
p 3 (/3,
y,
5,
/3,
fi,
j3
P (A%l,ff
t
T)
= 0,
P (0,y,8,;T) = O,
2
3 (/3,
7 ,5,m; T) = 0.
(49)
Equations (49) are not satisfied by /x = unless also 13 = y = 8 = 0. Therefore they have solutions for j8, y, and 8 in terms of n which vanish for n = 0. If the determinant of the linear terms in 0, y, and 8 is distinct from zero the solution is unique. In treating the problem it is convenient to use derived from Let equations (49) rather than these equations themselves. <p represent the angle between the positive end of the x-axis and the line from the origin to the point (x T ,y T ). Then let Q l; Qiy and Q, be defined by
Q^P.cos^+Pjsin^,
Q, =
(50)
475
lies
yT )
(0,y,S,
x;T)=O,
= 0,
Q,(0, y,
5,
T)
(51)
are identical with those of (49), for the two sets of functions are linearly related with non-vanishing determinant.
The determinant
Q which
t
b is
3Q.
A=
476
PERIODIC OIIBITS.
T and
3
Q, becomes
0; 7").
Q,
(o,+0,
5;
n-(a,+0,
(55)
Therefore dQ/d/3 = d/d/3, dQ,/d5 = d/dS, the first of which is positive by which were derived in 223. The second one of these the properties of = (i = 8 = 0, the partial derivatives is positive or negative according as, for n
infinitesimal
If <,= T derivative of
3 t
.
body
receding from, or approaching toward, the origin at 3 then dQJdb, which is the has its greatest value for l
is
tt =T
with respect to
t, is
zero.
It follows
from
(47), (48),
/
and
i
y = m = 0,
Q, =
+0,0; T)Y
in (52) are
(56)
which appear
dt'(a
+0,8;T)
as
(57)
dQ " =
3
dp
dQ _
3
dp
Vl+te'K+O,
(9) it is
0; T)Y'
dS
Vl + te'K+0/0;
T)f
From equation
S'
found that
1
= 2c(l + 5)V
[(l+2a(l + 5)M-fa
(l
+ 5)V;+
].
coefficients alternate.
||
Since a
= +4rJf
is
rj+
'
ff -4^+100.1?-^0^+
(58)
negative, the first of these expressions is positive; since the velocity decreases or increases with increasing time (according as the infinitesimal body is receding from or approaching toward the origin) the second
negative if the body is receding from, and positive 3 If t is zero for t is an approaching toward, the origin. even function of 5 because the motion is symmetrical with respect to T3 as Therefore if the In this case the second of (58) is zero. initial time.
of these expressions
if it is
is
'
=T
infinitesimal
body for n = P = 5 =
is
not at
its
greatest distance at
t^T
3
,
the form of
is
A=
+ +
(59)
where the upper sign is to be used if it is receding from, and the lower if it Since dQt/dy is distinct from zero, A is approaching toward, the origin. is distinct from zero. Therefore in this case equations (48) are uniquely This means solvable for p, y, and 5 as power series in n, vanishing with n.
477
that if an arbitrary point in the xy-plane, and a velocity greater than that of this point with respect to fixed axes, be selected, then there exist two orbits of ejection passing through this point such that, for (x sufficiently
small, the infinitesimal body will pass the point with the given velocity. The direction with which the body passes the point depends upon the initial This is Painleve's conditions, of which it is a regular analytic function.
of three bodies. the velocity chosen equals that of the arbitrary point with respect to fixed axes, so that, for n = 0, the point in question is at the greatest distance to which the infinitesimal body recedes, then the determinant A is zero and the solution is multiple. The reason for it is that the two soluIf
tions
in
The question
t= ^(l-r-5),
of their existence for large values of li will the solutions of (23) may be written
now be
considered.
x=f
and the conditions
(a
+P,
5,
M5
r,),
5,
rj,
in
r,
(60)
for
are
^
It
-fi(%+fi, *,l;*)-0,
*()*/(+&*,
*)-0.
(61)
for
can be solved |pj sufficiently small, equations (61) in series with as and that when /S ll, vanishing li, converging power these results are substituted in (60) the latter become power series in n
and
which converge for 0^t^7t provided \p\ is sufficiently small. Suppose the series for the closed orbit converge for li = p, and that the The solutions of (23) values of ao +0 and 5 for this value of m are c, and 5 are expansible as power series in t for values of a, 8, and n in the vicinity If |o o <r 1; \8 8i\<r3 |/x /*,|<r, the series converge if of a,, 5,, and ftT t <T, where T is any arbitrary quantity less than the period from ejection The result will have the form of (27) where r is replaced by to collision. Each term of (27) can be expanded as a power series in a a lt T t (l + 5). < 1 p, . The solution 8 m /*, Mi which will converge provided \p Si may be written in the form
,
x-
^(a-a,, 5-6j,
x
/x-MiJ n)*
2/
= Ps(a-a
5X
/x~Mi
r,),
(62)
where y and p2 are power series in a ,, 5 ^, and li m^. Suppose the _ values of x, y, and their derivatives at t =T for a c^ = 5 = 5 = /x Mi = Let their values for an arbitrary set of values of and y'r are x T y T x' T a Mi satisfying the inequalities which insure convergency aj 8 5i and m
x
1
.
>
478
be x T + 0,
as
,
PERIODIC ORBITS.
y T + 0,
x' r
+
,
and
5,
,
2/^
4-
/3 4
Then
0,
are expansible
power
series in a
a,
m
r
Mi of
the form
(<-l,
4),
A-(-a
where the
g,
*- 4iJ'-J%;^
;
(63)
vanish f or a
a!
=5
5,
=m
Mi
= 0.
consider a solution with the initial conditions x T 0, yr t = = = = for O the infinitesimal 0, 2 3 4 body does z'r+A, y'r+ft- Suppose not pass through the position of n for I5t1 ^t. Therefore the solutions
,
,
Now
+P
4
.
which
,
will
converge for
At
t,
= 7t
are sufficiently small. the expressions for the coordinates become, making use of (63),
0,
, .
.
Z = -Pi(0i, y = F2 (0i,
x'
,04
,0
4
r)
,04
(64)
-)
5-5,, /x-Mi).
Conditions that the orbit shall be closed with the period 27r/(l
+ 5)
are
(65)
Q
It has
(a-a,, 5-5,
/x-/i,)
= 0.
and
vanishing with n
true unless the solution becomes multiple. there is one real solution for both positive
is
of n
p,,
There are three solutions altogether for \n\ sufficiently small because , defined in (16), has three values for which the conditions of a closed orbit of ejection can be satisfied, but only one of them is real. Consequently the real solution can not disappear by uniting with one of the others unless they first unite and become real. Then, if two of the real solutions should unite and become complex, there would be one real one left. That is, there to 1, is one real closed orbit of ejection from 1 m for all values of m from The has been made for the the value unity. argument period excluding 2ir, but it is entirely similar for any multiple of 2ir. It has been tacitly assumed in the argument that none of the orbits of ejection under consideration passes through the position of n for any value of n; for it was only under this condition that the convergence of It has been proved that the closed orbits of ejection the series was assured. do not pass through m for n sufficiently small. Since the coordinates in these orbits are regular analytic functions of n, it follows that if any one of
position of m for
any value
The motion it will pass near n. the vicinity of a finite body when referred to rotating axes
always in the
479
retrograde direction, and the orbits in question are always symmetrical with respect to the x-axis.
Consider the motion with respect to it in a closed orbit of ejection from 1 Whether the ejection is toward or from it the motion with respect to it fx. in those parts of the orbit which are near to it is direct instead of retrograde. Therefore, the orbit can not pass near it without first developing folds so
that a line from
is
it
it
three times.
It
extremely improbable, though not absolutely certain, that this sort of development could take place. It is probable that the real orbits of ejection which exist for it sufficiently small continue in the analytic sense for all to unity, and that they do not pass near it. values of it from
the orbits in which the ejection is toward it are symmetrically opposite to those in which the ejection is away from it, and these conditions
For
it
l-JJ'C.
v-l-l
S-l
/
Fig. 18. Fig. 19.
when
it
is
small.
When
it
which surround 1 it diminish in size and preserve their approximate forms, while those which surround it approach circles whose radii are f'3 where For it 1 e, where e is very 2jir is the period, and whose centers are at it. There are, of small, the orbits have the form shown in Figs. 18 and 19. closed orbits of from both of the finite bodies. course, ejection
,
233. Periodic Orbits Related to Closed Orbits of Ejection. There are periodic orbits passing near one of the finite bodies of which the closed
orbits of ejection are the limits. Suppose it = with infinitesimal the body respect to the finite
of
body
it.
Let the
infini-
cross the x-axis perpendicularly at < = near the body 1 it, and let the initial velocity be determined so that the period is 2w, or a mulThen the motion with respect to the rotating axes is periodic, tiple of 2t.
tesimal
body
ISO
PERIODIC OKBITS.
is
symmetrical with respect to the rc-axis, and crosses it perpenThe limits of these orbits, as the nearest at the half period. dicularly approach to 1 m becomes zero, are the closed orbits of ejection.
the orbit
motion near the finite body in both the positive and the retrograde directions, but in both cases the motion in the remote parts of the orbits when referred to rotating axes is in the retrograde direction. Therefore, those in which the motion in the of finite is direct the have loops, while the others do not. The body vicinity of classes of character the two orbits for periods 2x and 4?r are shown in Figs. 20 and 21. There are, of course, orbits of a similar character which are symmetrically opposite with respect to the y-axis. Suppose the initial conditions for one of the periodic orbits in question when n = 0, are x(0) = a, z'(0)=0, y(0) = Q y'(0) = b, and represent the coordinates for this solution by x x' y and y' The distance a is small
!
, , , .
The
and the
by
ju-
Fig. 20.
Fio. 21.
x(0) = a+a,
from zero and that the initial conditions are = 6+|8. Let the variable t be introt/'(0) duced by t = r(l + 8), Where 8 is an undetermined parameter. Then the solutions of the differential equations of motion, which are regular functions of the coordinates and 8 in the vicinity of x = x x' = x y = y y' = y[, 5 = for ^ t ^ jit (j an integer), can be written in the form
suppose
fx
Now
is
distinct
= 0, x'(0)
= 0, 2/(0)
x=x
5,
y=y
(a, ^, 8,
(a,
(3,
fi;
t),)
8,n; t),\
where p p 4 are power series in a, P, and 8, vanishing with a, /3, and 8. the moduli of o, 0, and 5 can be taken so small that the series Moreover, converge for ^ r ^jV, where is any integer. Sufficient conditions that the orbit for /x distinct from zero shall be
t
,
. .
,;'
2jir
;
are
ps (o,
0, 8,
M jV)
= 0,
p,(a,
p,8,;
jir)
= 0.
(67)
481
The problem
is
to
show that
a,
/3, and 8 can be determined so that these an arbitrary /x sufficiently small. In the
can be taken arbitrarily when the other is determined in terms of j except for sign one sign belongs to the direct and the other to the retrograde orbit. It will be supposed that a is the arbitrary. Therefore it may be supposed that it absorbs the undetermined a and leaves only two parameters in (67) besides the arbitrary /x. Equations (67) can be solved uniquely for /3 and 8 as power series in n, vanishing with /x, provided
;
dp?
djpi
A=
3/T
djh
~d8
(68)
djh
d8
dp'
is
/3-5
is
=M=
formed
ju
distinct
from
zero.
to zero, and therefore A depends only upon the two-body problem. Before the second column is formed /8 can be put equal to zero. When n = 3 = the period in t is 2jv; hence at the half period the infinitesimal body is
of x
is
an even function
of
it
tjw.
Now
,
the
is
parameter serves only to vary the period in t (keeping therefore equivalent to varying t from the half period.
8
fixed in t)
and
When t is near jw, can be determined so that tjir=JT(l+8)jir consistently with the definition of r. Therefore p 2 (0, 0, 8, /*; jir) is an even function of 5, and consequently dp 2/ 68 = Ofor/3 = 5 = /x = 0. Hence the determinant 5 becomes
A = ^^i. The second
factor,
is
it is
(69)
factor
is
distinct
the derivative of
zero.
first
for a constant
this derivative
distinct
from
If
In considering the
to zero.
r\
expression for x
and becomes
factor of (69) the parameter 8 can be put equal represent the coordinates referred to fixed axes, the
sinf.
=jir
is
The problem
is
r^Mjl +
;
dl
a/3
d/3^d/3
(70 ) {<U)
is
assumed conditions.
482
PERIODIC ORBITS.
The
expressions for
',
and
rj,
as given
are
= a[cos#-e],
wliere
= ^l-ei smE,
E is
w is the mean angular motion in the for t =jw and /3 = 0, it follows that
orbit.
the major semi-axis of the orbit, and Since sin' = and cosE= 1
483
234. Periodic Orbits having Many Near Approaches. -The orbits considered in the preceding article are characterized by the fact that, at least
for small values of
1
fj.
ll,
body
continually recedes until the mid-period, which is a multiple of ir, and then returns symmetrically. Orbits will now be considered in which
it
the infinitesimal body recedes from and returns toward before they re-enter.
ix
many
times
body is started near 1 ll m and ll; and suppose the on, and perpendicularly to, the line joining 1 initial conditions are such that its period is commensurable with 2ir without
Suppose
fx
is
zero
infinitesimal
P = 2tt P
where p and q are relatively prime
to the rotating axes
is
(72)
integers.
respect
T = Pq = 2irp.
(73)
In this period the infinitesimal body runs through q of its periods with respect to fixed axes and the movable axes make p rotations.
Now
suppose that
ll
is
distinct
initial
conditions
are given slight variations, but of such a character that the infinitesimal body is started at right angles to the line joining the finite bodies. A new independent variable t and a parameter 5 are
introduced by the relation t = r(l+5). The solutions can be developed as power series in and the increments to the initial condiix, 8, tions, and the moduli of these quantities can be taken so small that these series converge for t<T/2. Then the conditions that the solution shall be periodic are that the orbit shall cross the
x-axis perpendicularly at r = T/2. These conditions have the form (67) and all of the properties of (67) which were used in proving the existence
and character
of their solution.
Therefore, the
Jig. 22.
periodic orbits which are in question exist. Now suppose that the initial distance from
1
fx,
which was arbitrary, is made to approach During approach to zero the distances to the other near apses vary, but there is no apparent reason why all these apsidal distances should vanish at the same time. In fact, from the lack of symmetry it is doubtful whether any two of them arc simultaneously zero.
zero as a limit.
this
IS
PERIODIC ORBITS.
The simplest orbits of the type under consideration are those for which p=l, q = 2. Their general form for retrograde motion in the vicinity of the finite body 1 // is shown in Fig. 22. If the distance from 1 n to a becomes zero, the orbit of ejection discovered by Darwin is obtained. The
m to b becomes zero is one that is question whether the distance from 1 hard to answer. Certainly it can not be answered affirmatively with complete rigor by numerical experiments, though the existence of certain classes of periodic orbits can be proved in this way. If, for perpendicular projection from a given point on the z-axis with a certain speed, the next crossing of the x-axis is at an angle which is greater than ir/2; and if, for a perpendicular projection from the same point with a different speed, the next crossing is at an angle less than ir/2, then, from the analytic continuity, it can be inferred that there is an intermediate speed at which the crossing will be
perpendicular. But in the present case these conditions are not present, and all that can be said is that when the distance from 1 n to a vanishes, the
distance from
fi
to b
is
is
Chapter XVI.
IN
THE RESTRICTED
In the problem of two bodies there are whose dimensions range from infinitely great to infinitely small. They form a continuous series geometrically and their coordinates are continuous functions of the various parameters by which they may be There are orbits in which the direction of motion is forward, and defined. others in which it is retrograde. The two series are identical only when the orbits are infinitely great and when they are infinitely small. In the restricted problem of three bodies* the orbits which are analogous to the circular orbits in the problem of two bodies are those which revolve around one or both of the finite bodies and which re-enter, when referred to rotating axes, after one synodical revolution. Those inclosing but a single finite body were treated in Chapter XII, and it was shown there that the deviations from uniform circular motion are due to the attraction of the second finite mass. Those orbits which revolve around both finite masses, and which are analogous to circular orbits, were treated in Chapter XIII, and it was shown there that the deviations from uniform circular motion are due to the fact that the finite masses are separated by a finite distance.
a confrom those inclosing both finite masses and having infinitely great dimensions to those revolving around the two finite masses separately in orbits of infinitesimal dimensions. There are no difficulties for very great or for very small orbits, but since in some way the infinitely great are eventually divided into two series which become infinitely small, it is clear that there is a region in which the resemblance to the two-body
of the present chapter is to trace, so far as possible,
The problem
problem
is
very remote.
of
The
The method
treatment
is
The differtions with respect to the parameters upon which they depend. ential equations which define the motion are, when referred to rotating axes
and
in canonical units,
d2x
dt
2
9 dy_
dt
(I
-aQQe+m)
ns
V
(x-l+it)
r2
3
'
dt
2+Z dt
u
is
V
r2
2
V
=(x-l+ M
2
(1)
r? =
*
{x+n)-+y\
three bodies
+2/
that in which there are two finite bodies and one infinitesimal,
485
circles.
486
PERIODIC ORBITS.
(2)
The
general solutions of (1) can be written in the form a 4 m; t), y=f2 (oi, *-/i(i,
.
. . , ,
a 4 m; 0,
,
(3)
a 4 are the initial values of x, y and their first derivatives. where ai, The conditions that the solutions (3) shall be periodic eliminate three of the
.
.
four
a,,
y = F2
(a,
nt),
(4)
(for example, the Jacobian a,, constant C), and n is a constant depending on a and n and so associated with The problem under consideration is to follow i that the period is 2ir/n.
where a
is
one of the
or a function of
them
the solutions as
It will a, n, or n varies through its possible range of values. in the to found convenient discussion use sometimes one and sometimes be another of these parameters as independent. The correspondence between them is not one-to-one, so that a series of orbits may branch at a certain point with respect to one of them and not with respect to another. The functions in question are highly transcendental and it has not been found possible to follow them with complete rigor through branch-points and infinities by direct processes. But the fact that the orbits may have a branching with respect to one parameter and not with respect to another makes it possible sometimes to establish the existence of critical forms indirectly. There are orbits whose coordinates are complex and whose periods are real. With varying values of the parameter a they may become real. This does not arise in the problem of two bodies. It makes it necessary to include in the discussion certain orbits which are complex for the values of the parameters in terms of which they are most conveniently expressed. Sir George Darwin discovered several classes of periodic orbits by numerical experiments.* Burrau found the limiting form of the orbits of one
family of oscillating satellites by the same process. f Many other families have been discovered by computations carried out in connection with this
work.
All these examples illustrate the theories to be set forth here and place them on a solid basis at points where they fall short of complete rigor because of the difficulty of following orbits by analytical processes
through critical forms. Before taking up the direct synthesis, some of the properties of the periodic orbits which have been previously given will be enumerated and some additional ones will be derived.
236. Periodic Satellite and Planetary Orbits. It was shown in Chapters XII and XIII that there are periodic orbits encircling each of the finite bodies separately, and others encircling both of them together, which are closed
*
t
Acta Mathematica, vol. 21 (1897). Astronomische Nachrichtcn, vol. 186 (1894), No. 3230; and
ML,
vol.
487
The
grade. parameter are developed, corresponding to very small and very large orbits respectively, there are three classes in which the motion is direct and three in which it
is
of the
retrograde.
Only one
and one
real for small values of the parameters. Darwin's computations that at least in some cases the complex orbits may become real.
is
show
that
One
they
all
of the
is
period. these properties persist, for the solutions are expansible in integral or fractional powers of the parameters, and the property in question is a consequence
cross the line joining the finite bodies perpendicularly at every half If the parameters upon which the periodic orbits depend are varied,
Therefore, the whole series of orbits dimensions will possess this property, unless indeed they branch into two series which are symmetrical with respect to the line joining the finite bodies. If the coordinates of a real periodic orbit are analytic in a parameter, then the orbit can not disappear without becoming identical with another periodic orbit;* and a complex orbit can not become real without becoming
of the character of their coefficients.
from
infinite to infinitesimal
complex
orbit, the
two becoming
real as
they become
Real orbits disappear and appear in pairs with the variation of the parameters in terms of which their coordinates are analytic functions.
237.
The Non-Existence
Appeal
will
be
made
necessary to prove that the orbits which they have shown to exist are not isolated examples which exist only for special values of the
masses and the other parameter on which they depend. Suppose that for ju = Mo equations (1) admit the periodic solution x = Fi
(no; not),
y = F 2 ( lx
initial
riot),
(5)
is
2w/no.
The
,
conditions are
2/(0) =2/0,
x(0)=x
x'(0)=x'
y'(0)=y'
(6)
The
initial
integral
ft
W+yV)+a*+f*
^ly^ +lft
"l '2
7)
The
it
orbit in question will not pass through one of the finite bodies, for then would not be strictly periodic, as was explained in 222. It will not have
infinite branches, for
*
any
then
it
Let
M&hodcs
83.
INN
PERIODIC ORBITS.
There are two problems to be considered: (A) To determine whether or not periodic orbits exist for /x = mo+X, C = C which reduce to (5) for X = 0; and (B) to determine whether or not periodic orbits exist for /u = Mo, C = C +y, which reduce to (5) for 7 = 0. If periodic orbits exist when the parameters
,
are varied separately, they also exist when they are varied simultaneously. = (A) Let n no+\ and take as initial conditions
x(0)=x
+ a,
x'(0)=x'
+ a',
y(0)=y +P,
y'(0)
=y'o+P'.
(8)
Since the periodic orbit (5) does not pass through a singular point of the differential equations, the solutions can be developed as series of the form
=Fi (moJ WoO + a +pi(a, = *YGuo; lht) + a'+p (a, y = F ( Mo n t)+p +p 3 (a,
x'
2 2
;
a', P, a', p,
a',
j8,
P'
X; 0, X;
t),
j8',
(9)
P', X;
t),
^ = F '( Mo
2
Wo0+/3' +P4(a,
a', 0, P',
X; 0,
p 4 are power series in a, a', p, P' and X which vanish identi= 0. Moreover, if any finite T cally with these parameters and are zero for t is taken in advance the moduli of these parameters can be taken so small
where p t
. . .
that p\,
0<*^7\
.
. .
The
integral (2)
.
can be written
,
F(p u
where
Pi
a,
a',P, P',
.
.
\)-F(0,
.
0, a, *', 0,
#, X)-0,
Equation
(10) (10)
is
F is a power series in p u
by p =
x
. .
pA
t
a, a', P, P',
and
X.
identically satisfied
= p = 0.
Sufficient conditions that the solution (9) shall be periodic with the period P = 2r/n are
p,(a,
X; X;
p 2 (a,
It follows
18',
P) =0, P) =0,
p 3 (a,
7^(a,
/S,
P', X;
P) =0,' P) =0.J
(ID
can
is
a', p, p',
'
and
'
X,
vanishing
= p = p = 0. If x were zero and y were not zero the identically for pi If x and y were both zero solution would be made for p instead of for p
'
'
2.
the origin of tune would be shifted so that at least one of them would be disThis is always possible unless they are identically zero. tinct from zero.
But they
when the infinitesimal body is at an has been shown in Chapters V and IX that
Consequently,
it
equithese
is
may
= <p(pu p
if
2,
p.J a, a', p,
,
j8',
X),
*>(0, 0,
0; a, a',
j8,
P',
X)0.
(12)
Therefore,
tion of (11)
p lf p 2 and p 3
this relation
is
p 4 is redundant and
are periodic with the period P, then by virtue of also periodic with the period P. Hence the fourth equa-
may
be suppressed.
489
tions (11) are not satisfied by X = with a, a', orbits would be periodic with the same period
consider the solution of the first three equations of (11) for a, a', terms of X. The parameter 8' is superfluous and may be taken equal to zero. This amounts to a definite determination of the initial time. Since a is real (5) periodic solution the coefficients of pi, p 2 and p 3 are real. Equa-
Now
in
and 8
and 8
when n =
by X =
Similarly, they
unless
all
They
by
three of the parameters a, a', and 8 are a = a' = 8 = and X arbitrary, for then fixed
conditions would give a periodic orbit for all distribution of mass between the finite bodies. This is certainly not true for fi near zero. Therefore, the three equations can be solved for a, a', and 8 as power series in
If the powers are integral the solution integral or fractional powers of X. will be unique and real for both positive and negative values of X. If the
will
and negative values of X. If the powers are even fractional, there will be two real solutions for X either positive or negative, depending on the signs of the coefficients of certain terms, and all the solutions will be complex for X negative or positive, depending upon the signs of the same coefficients. Hence in all cases there are real periodic solutions for small values of X which reduce to (5) for X = 0. (B) Let C = C +y and take the initial conditions (8). Also let
-(1+8)t, an arbitrary parameter and t a new independent variable. solutions in this case have the form
where
8 is
(13)
The
X
y
2/'
=Fi
(moJ
;
r)
+a
+pi(o,
a', 8, 8',
y, 8; r),
t),
x' = Fi'(fi
noT)+a'+p 2 (a,
rhT)+8 +p 3 (a, n r)+/3'+p 4 (a,
=F
=F
(14)
t),
(n
2 '0uo;
y, 8; r),
where p u p4 are power series in a, a', 8, 8', y, and 8. The parameter y Thereis determined in terms of a, a', 8, 8', and 8 by equation (2) at t = 0. if is not fore it may be omitted from p u A Or, zero, equation (2) ,p y determines 8' as a power series in a, a', 8, y, and 5, vanishing with these It follows from the It will be supposed that 8' is eliminated. quantities. integral that the last equation corresponding to (11) is redundant, and it will be suppressed. The constant a' is superfluous and may be taken equal to Then the conditions that the solution (14) shall be periodic with the zero.
. .
.
'
period
in r are
Pl (a, 8, y, 8;
P)=0,
p 2 (a,
8, y, 8;
P)=0,
p 3 (a,
8, y, 8;
P)
= 0.
(15)
Consider the solution of equations (15) for a, 8, and 8 in terms of y. They are not satisfied unless all four of these parameters are zero, and consequently solutions for a, 8, and 5 as power series in integral or fractional powers of y exist, and the circumstances under which they are real are strictly
490
238.
PERIODIC ORBITS.
Double Orbits with Changing Mass-Ratio of the Finite Bodies. In Chapter XI, p. 359, it was stated that Darwin's computations show that the two orbits which are complex for small values of ///, or for large values of the Jacobian constant C, unite and become real for a certain value of C. In this computation the ratio of the masses of the finite bodies was 10 to 1. The question naturally arises whether a corresponding
The Persistence
of
double periodic orbit exists for other ratios of the The conditions for a double periodic orbit
finite
masses.
first
will
be developed.
(16)
Suppose
x-fito,
is
x'=h'{t),
y=f
(t),
y'=V(t),
a periodic solution of equations (1) for m Mo, having the period P. All orbits except those around the equilateral triangular points, which will be given special consideration in 242, are symmetrical with respect to the In them the origin of time can be so chosen that the initial conx-axis.
ditions are
x(0)=x
and from
(2),
x'(0)=0,
.
2/(0)
=0,
y\0)
=/,
(17)
C= C
Now
x(0)=x
it will
+ a,
x'(0)=0,
y(0)=0,
y'(0)=|/'o+/9,
C = C +y.
(18)
The constant
can be expressed in terms of a and y by means of (2), and be supposed that |8 is eliminated by this relation. Let a new independent variable t be defined by
t
= (l+8)r,
Then the
(19)
*-Ji (, y = p 3 (a,
where p u
. .
V, I; t),
y, 8; r),
x'
J/'
=p =p
2
4
(a, y, 8] t),
(a, y, 8;
r),
,_
p4 are power
series in a, y,
and
8.
The
in t are
p 2 (a,
It will
y, 5;
P/2)=0,
with
p 3 (a,
y, 8;
P/2)
= 0.
(21)
as the independent variable and equations. Suppose equations (1) initial conditions (17) are integrated as power series in 8. The terms indeof 5 will be with the P. pendent periodic period Non-periodic terms will
enter only when the terms in the right members at some stage of the inteThen t times periodic terms will gration contain terms with the period P.
appear in the solution, and terms multiplied by t 2 and higher powers of I will not enter until later stages of the integration. Terms of this character
491
for all 5
;
would be periodic
the coordinates would be constants with respect to I. The expression for y = p 3 is an odd function of I with the initial conditions (18) Therefore the term in t is multiplied by an even function, that is, a cosine term which does not vanish at t = P/2. Therefore p 3 carries 5 to the first degree
that
is,
.
and
this
a power series in a and y vanishing for Suppose 7 is taken as the independent parameter and that (22) is solved for a in terms of y. A necessary and sufficient condition that (16)
where
P is
a = y = 0:
C is
that
(23)
AP a
for a
= P (a,y)=0
l
= 7 = 0.
Now
Suppose this condition is satisfied. suppose n = fi -\-\ and consider the question of the existence of a
double periodic solution for this value of fi. The double periodic solution will exist provided the equation corresponding to (23) is satisfied. Suppose the initial value of x is
x(0)=x
where
<x
+a + a,
(24)
a parameter which remains to be determined. The steps corto those leading up to (23) can be taken in this case, and the responding equation corresponding to (23) becomes
is
P
for a
2 (<r,
X)=0
(25)
= 7 = 0, where P 2
series
The
orbit
a power series in a and X vanishing for a = X = 0. contains terms in <r alone; otherwise, not only would every
is
whose
initial
conditions were
x(0)=x +a,
where x(0)
orbit.
is
*'(0)-0,
2/(0)
=0,
it
y'(0)=y'
(26)
fractional
of X, vanishing with X, the multiplicity of the solution being equal to the degree of the term of lowest degree in <r alone. Hence, in the = (that is, for analytic sense, if there is a double periodic orbit for X
power
m = mo), then there is a double periodic orbit for every value of n = n +\. If the solution of (25) has the form
<r
= Xp(X),
,
(27)
a power series in X, then there is a single double solution of the But if the solution is in X } then there series for every |X| sufficiently small. are two real double solutions when X has one sign and none when it has the That is, for X = two real double periodic orbits unite and disappear other.
where p(X)
is
492
PERIODIC ORBITS.
And in general, double periodic solutions appear or identical for certain values of n, a result analobecome which disappear in pairs, gous to Poincare^s theorem respecting the appearance and disappearance of
by becoming complex.
real periodic orbits.
consider the real periodic satellite orbits which are re-entrant after a single synodical revolution. Darwin's computation shows that for the mass ratio of 10 to 1 there is one, and but one, real double periodic orbit in which the motion is direct. Hence, there is no other double periodic orbit
Now
with which
it could unite to disappear for any value of n. The same result also is a consequence of the analysis of Chapter XI, where it was shown that for n = 0, and therefore for all n, there are but three
From the fact that there orbits of the type in question. are only three periodic orbits of the type in question it follows that there can not be more than one direct double periodic orbit; and from its existence
real
and complex
for
10 M = /ii) it follows that there is one direct double periodic The orbits of inferior planets orbit for all values of n from zero to unity. differ from those of satellites only in the ratio of the masses. Therefore, there are also double periodic orbits of inferior planets in which the motion
n = l /u,
forward direction. one of the masses which revolve in circles becomes zero, those orbits around the other which are complex for small periods are complex for In this case there are no double orbits all periods from zero to infinity.
is
in the
When
except those of infinite and infinitesimal dimensions. The question arises as to the character of the double orbits for very small values of the second finite mass. Consider the totality of real circular orbits around one of the
finite
bodies
when
is
zero.
becomes
finite, the periodic orbits are continuous deformations of the circular orbits with the exception of that circular orbit whose period is 2t.
through the point where the second mass becomes finite, and the This means that the orbit It is conjectured that the complex orbits have itself has a discontinuity. corresponding discontinuities; that when the second finite mass is very small there are three real orbits about the larger mass in which the motion is in the forward direction and which have a mean distance near unity and a period near 2tt; that there are three corresponding real orbits of small dimensions about the smaller mass; and, finally, that there are three similar orbits of the nature of superior planets. For increasing values of the Jacobian constant two of the three in each case unite and form the double orbits. Consequently, if this conjecture is correct, and if the three types of double orbits are followed as one of the finite bodies approaches zero as a limit, the one around the larger finite mass and the one around both finite masses approach the unit circle, and the one around the smaller finite mass approaches
It passes
zero dimensions.
493
The question
test for
/x
=
(7
In
ing with
= 3.58 and
53 orbits were computed for various values of C, startending with (7 = 3.086. For (7 = 3.58 there were two
which were geometrically and which intersected the x-axis near alike, For smaller values of (7 the correspond.785. ing periodic orbits were more nearly identical. For (7 = 3.086 they were sensibly identical. The
direct periodic orbits
much
computation for
this value of
wn.
23.
=y l
C = 3.086,
Double periodic
orbit.
494
M = Ji
PERIODIC ORBITS.
495
The
(Fig: 26,
x A,
orbit
defined
by m=2> C = 2.50, x
t
page 496)
C = 2.95,
l^
496
PERIODIC ORBITS.
The
power
series
mt-ti
with the
initial
conditions
= =
?j
l
'
= ?/ =
i-4?(t-t y+%(t-t
a Bo,
,
,
y+
24
\-^Bq-\-AoB 1 +BqB 2 24
(30)
{t-uy+
direction cosines of the normal to the curve of zero relative velocity at the point (x u yi)
A
is
and
of zero relative velocity at the point (x u yi). take a new set of axes (u, v) with origin
Now
with u having the direction of the tanFig. 26. ends If the it. to v and positive perpendicular gent of the new axes are chosen so that the cosine and sine of the angle from the end of positive end of the -axis counted counter-clockwise to the positive of t;-axis end the if the B and the w-axis are proportional to A and positive is 90 forward counter-clockwise from the positive end of the w-axis, then the
at (x u
j/i),
,
(t
j 4+
(31)
= Bo
l)
2]
^t
^+
positive for both positive and negative values of (t-ti), because the coefficient of {t-Uy can vanish only at an equilibrium Therefore the positive end of the w-axis extends from the point point. The value of v is positive for (xi, yi) into the region of real velocities. small negative values of t t u and negative for small
The value
of
is
Therefore, if motion along the positive values of curves of zero relative velocity is taken as positive when the
region of real velocity is on the
orbits,
left,
ttu
the
motion in
is
the
body crosses the tangent In Fig. 27, C is a to the cusp in the positive direction. is the orbit near curve of zero velocity, P is a cusp, Fig. 27. the cusp, and T is the tangent to the orbit at the cusp. Now suppose the orbit having the cusp is a periodic orbit. If it has no other double points than at the cusps, and if it is inside of the curves of zero relative velocity, then it revolves around one of the finite bodies in the
direction; that
the infinitesimal
positive direction.
If it is outside of the curves of zero relative velocity, it Inrevolves with respect to the rotating axes in the retrograde direction.
497
periodic orbits of superior planets revolve in the retrograde direction with respect to the rotating axes. But the orbits with cusps, if they have no other double points, revolve in the forward direction with respect to fixed
axes, because at the cusps they rotating axes.
to Cusps.
z(0)=x
is
C = Co,
(32)
Therefore
x'(P/2)=0,
l/(P/2)=0.
(33)
Suppose
it
has a cusp at
x{t l )
u or
=x
l ,
x'(t 1 )
y{k)
= yi,
way
*'(ft)0.
(34)
If
character in the vicinity of the cusp will be changed. The nature of these changes will now be considered. Suppose the initial conperiodic, its
ditions are
x(0)=x
and that
+ a,
x'(0)=0,
2/(0)
=0,
y'(0)=y '+p,
C = C +y;
(35)
t=(l+6)r,
where r
is
(36)
5 is
new independent
variable
and
The
solutions can be
expanded
5)=0,
as
x'[(l+8)P/2]
x
= Pi (a,
0,
y[(l+8)P/2]
=p
(a, 0,
5)=0,
(37)
where p and p 2 are power series in a, 0, and 5, vanishing identically with a, Now consider the solution of (37) for and 8 in terms of a, vanish0, and 5.
ing with a. The solution is always possible either in integral or fractional = 5 = 0. But this powers unless the equations are identically satisfied by means that all orbits in which the infinitesimal body crosses the x-axis near
period.
'
all
the orbits
respect to a, in the analytic sense, until the place of crossing, x , is small. But then the methods of Chapter apply and it is known that the
XIV
Therefore equations (37) are not identically satisperiod depends upon x = 5 = 0, and they can be solved for and 8 in terms of a. The solufied by tions will have the form
.
= a ,/ "P,(a
"'),
= a v "P
(a
l/
"),
(38)
8 is dis-
of pi
If
positive
In general
it
will
be unity.
'.is
PERIODIC ORBITS.
In the computations of Darwin 7 was taken as the parameter which The change can be made here because (2) is uniquely defines the orbits. solvable for a as a power series in /3, 7, and 5 unless
.
(1-ju)Qcq+m)
/i(xo-l+M)
_n
2/o
= 0.
But these
The
equalities are satisfied only at one of the collinear solution points. orbits in the vicinity of these points will be omitted from this discus-
sion because they belong to quite another category. If a is eliminated by means of (2), and /J and 5 by means of (38), the solutions will be expressed as power series in y Up and the values of the coor,
dinates at r =
ti
are
x(r) T . h
= *, + 7 1/p0i,
3,
x'{r) r
^ - +7
l/
"6 2
y{rU= yi +y^e
where
du
.
.
,\
d4 are
power
series in
y".
t
expressions for the coordinates in the vicinity of the relations (39), can be expanded as power series int ti.
The
u satisfying
solution
is
The
(40)
;}
where
1/p = T 1/p &i = 7 04, a = (l+5)&i+Kl+5) 4 & 2 =-(l+5)a +Kl+5) fio, a 3 = f(l+5)6 +|(l+5) [4ia - f(l +5)a +i(l +5) B &s l+5 = l+7 1/piV /p ),
ai
2,
+4 6 a +B
2
1
(41)
2 b,],
1 ],
where
A Au A B B B
,
and
P
2
is
a power series in y Vv
2
The transformation
(31) gives
6) 2
It
u = 7 1/p (Mo +e*B (r - J,) + (a A +b B ) (t - h) +(a 3A +b iB )(T-t y + 1/p v 7 (Mo - B B Q ) (t-U) + (b A - a B Q ) (r - *0 + (b zA -a B )(T-t y+ .... follows from (41) that for 7 = W = iGV+J5o )(T-* ) + V= _I(^ o2+5o )(t _ <i)
1
(42)
(43)
determines the points at which the periodic orbit from the second of (43) that (t ti) = is a but not a triple quadruple solution for 7 = 0. Therefore there are three
r
The equation
It follows
solutions of v =
for
1/p
,
vanishing with
7.
or fractional powers of
(44)
499
The
others depend
second equation of (42). Unless d4A remaining solutions have the form
of the
member
the two
h=
If
1/2P
tx
J/2P
vaj+bs
'
(45)
which will be exceptionally if at all, the corresponding solutions exist but may be in integral powers of y 1/p It has been remarked that p will in general be unity. When it is odd
.
K = 0,
the periodic orbit with the cusp at (x u yi) is a multiple orbit. If p is even, two orbits which are real when 7 has one sign unite for 7 = and disappear by becoming imaginary when 7 has the other sign. When p is odd there is a
single real orbit for
7 both positive and negative. It is clear that only If it were exceptionally, if at all, will a double periodic orbit have a cusp. so in any particular case the value of y. could be changed, when it would no longer be true. Therefore it will be supposed that p is unity. It follows from (45) that the second and third intersections of the curve
is positive with the w-axis are real for 7 positive or negative according as or negative, and that they are not real when 7 has the other sign. When the second and third intersections of the curve are real the curve consists of a small loop as is indicated in Fig. 28; and when they are complex the curve has a point near which the
curvature
Fig. 29.
is
When
sections of the curve with the w-axis, one occurs before t\ and one after t t
.
It follows
from
this
discussion
Fig. 28.
Fig. 29.
that if a periodic orbit for a certain value Co of the Jacobian constant has a has a point, near a curve of cusp, then for a slightly larger (or smaller) value it zero relative velocity, in the vicinity of which there is very sharp curvature; for
diminishing (increasing) values of C the point of sharp curvature approaches the cusp form on the corresponding curve of zero velocity, which it reaches for C = C ; and for still further diminishing (increasing) values of C it has a small In Darwin's computations examples of loop near a curve of zero velocity. It follows, of course, from the symperiodic orbits with cusps were found. that if there is a cusp metry of the periodic orbits with respect to the x-axis = Jfc. at x = x,, y = y u then there is also a cusp at x x u y=
241.
The Persistence
Suppose for
of
Finite Bodies.
Cusps with Changing Mass-Ratio of the m = Mo equations (1) have a periodic solution
?y(0)=0,
y'(0)=2/'o,
x(0)=x,
x'(0)=0,
(46)
500
*(*i)=z.,
If
x'{t
PERIODIC ORBITS.
l
)=0,
y{h)=Vx,
IfW-.
(47)
represents the period of the solution, the expressions for x' satisfy the equations
and y
x'(P/2)=0,
y(P/2)=0.
(48)
Now
ixiriodic
suppose m = Mo+X and consider the question of the existence of a Let orbit having a cusp for this value of n.
I-(144K
The
ditions that
it
x(0)=x
shall
+ a,
x'(0)=0,
y(0)=0,
y'(O)=y'o+0.
X.
(49)
solution can be
expanded as power series in a, )3, 5, and be periodic in t with the period P are
j8,
The con-
x'(t) t=i
/2
= Pi(o,
8,
X)
= 0,
in a, 0,
6,
2/(t) t=/V2
= p 2 (o,
j3,
5,
X)
= 0,
(50)
vanishing with a, 0, 5, and X. Unless the initial conditions (46) define a double periodic orbit these equations can be solved uniquely for /3 and 5 as power series in a and X, vanishing with a and X. The results will have the form
and
X,
/S
=g
(o,X)
*-(*, X),
(51)
where q and g 2 are power series in a and X, vanishing for a = X = 0. Suppose /3 and 5 are eliminated from the solutions by means of equaThe results will be expanded as power series in a and X. The tions (51). values of the coordinates at t = U will be
x
Xto^mxt+Pfa
</(r),=
X),
x'(r),=
= 0+2V(a,
J
= i/i+P
'
\),\
(a, X),
*'(T)-*-0+P/(a X)J
where
Ph
P/,
t
x
P P
2,
The values
series in t
are power series in a and X which vanish with a and X. = U can be expanded as power
The
results are
. .
x(t)=Xi+Pi(o, X)+ a(r-Uy+ x'(t)=0 +P (o,X)+2o(t-i) + l/(r)=y +P,(e, X)+ b(r-t y+ y'(r)=0 +P/(o, X)+26(t-i) +
/ 1 1
i
(53)
.
The
U_
arc
l
= P./(a,\)+2b(t 2 -t )+ .... (54) These equations are not satisfied by X = 0, because a and b contain terms which depend upon x and y alone. Neither are they satisfied by o = 0, = 0, unless orbits crossing the x-axis at x = x are periodic for all values ti of m and have a cusp for the same = But it is known that the points at
,
= P,'(a,X)+2a(< -< )+
x
t2
tx.
which the periodic orbits cross the x-axis depend upon the value of ju- Therefore equations (54) can be solved for t 2 t and a in integral or fractional
y
powers
of X.
will
be in integral powers of
X.
If
the
501
is in integral or odd fractional powers of X, it is real for both positive and negative values of X. If the solution is in even fractional powers of X, there are two real solutions when X has one sign and only complex solutions
when
it
a real cusp exists for any value of n unless two real cusps become identical and disappear by becoming complex. Since an orbit is uniquely defined by the conditions for a cusp, as well as by any other initial conditions, cusps
if
It follows
from
H, it will exist
by becoming complex only when two orbits become identical. Darwin's computations showed that in the case of one of the orbits which was complex for large values of the Jacobian constant ("satellites of Class C") there were periodic orbits without loops near the cusp form, and others for smaller values of the Jacobian constant having loops. It follows from the results of 240 that between the two orbits there exists one having two cusps which are symmetrically situated with respect to the x-axis; and it follows from the discussion of this article that the orbits with cusps exist for all values of n unless a cusp develops on another orbit which later becomes identical with this.
disappear
Properties of the Periodic Oscillating Satellites near the In Chapter IX, Dr. Buck has treated the Equilateral Triangular Points. periodic oscillating satellites which are near the equilateral triangular points,
242.
Some
using in a general way the methods of Chapter V. It will be necessary for the purposes of the latter part of this chapter to develop a few additional properties of these orbits; and the most important of them can not be established
Chapter V, but follow from the methods of Chapter VI. differential equations will be transformed by letting m = Mo+X. For motion in the vicinity of the equilateral triangular points they are
by the methods
of
The
W
where
2
dt
~4~ (1
2lM,)y
A
mY
'
(55)
d2 y ,o dx _ 3V3 n
_ na
r_
X and
Y are
of the second
in x, y,
and
X.
In this article the character of the small oscillations will In treating them and Y may be provisionally put equal to zero. The characteristic equation on which the nature of the solutions depends is
be discussed.
S2 -j,
4
M -2-^(l-2 4
o)
^-bS
+^Mo(l-Mo)=0.
(56)
2S-3^3(1-2mo),
The
S*~l
roots of this equation are all purely imaginary or complex in conjugate When or is not greater than zero. pairs according as 1-27/z (1-mo) is solutions is zero there are two pairs of equal purely imaginary 1
-27mo(1 -Mo)
502
of (56).
It will
PERIODIC ORBITS.
be supposed for the present that p has such a value that the are roots of (56) pure imaginaries and distinct, and that p has such a value that X = p-p is very small.
Let the roots of (56) be
a and p are real and
+crV^l,
a<p. There are two periodic solutions of the linear the period 2w/a and the other with the period 2ir/p. with one terms of (55), = If x = 0, y = ci>0 at t 0, the solution having the period 2ir/a is
x = ^~ sin
Oi
at,
2/
= nr- sm
1
vt+Ci cos
at,
(57)
01
where
h0l_
2(r
^p'
Oi
2
n ai
- ~ 3 V/3
4
(l-2/lo) -
PR
^qrs
i
infinitesimal
body
2
,
is
an
,
ellipse
whose equation
is
+i x
2
2d
Ci
C\
^xy +
2/
=1
C\
and
if
major axis
of the ellipse,
it
between the positive end of the x-axis and the is easily found to be defined, except as to quad-
tan 2e =
i-a/-bS
2
]
<0
(59)
The
x'(0)=^>0, Oi
There are equations for the period
2/'(0)=^<0. Oi
2ir/p
and
is
replaced
by
p,
differ
c is replaced
by
2.
The motion
in these orbits
direction.
The
linear
+U n = !*
+^(1
=
)]
2
- 2mo)**/ +
W-C
a
2
)
(60)
Let the value of C for the orbits having the period 2w/a and 2x/p be Cp They are found from equations (57) to have the values
.
C and
a=
c1
[:-^(l+a
2
| [9-(<T +
2 2
9
4 )
-^(l-2po)
2 2
]
(61)
^=
It follows
,
c/[from
(l+a 2
].
and
p are
((
.
l-Vl-27po(l-po)
_ l + Vl-27/io(l-/io)
Since p is small a"1 is small and the limit of <r as mo approaches zero is zero. On the other hand p2 is near unity, and its limit as p approaches zero is 1.
503
is
positive
and C,
is
is zero. Hence the value of the Jacobian constant is greater for the orbits whose period is 2ir/<r, and less for those whose period is 2x/p, than it is for the Lagrangian equilateral triangular point solution. Now consider the curves of zero relative velocity. They are known to be real only if the value of C is greater than that which belongs to the
of
Ca = C
Therefore they are real only for the equilateral triangular point solution. solution with the period 2-ir/a. Their equation is
a=fx
This
is
3V3\
>-(l-2
)xy+-ly\
is
(63)
the equation of an
ellipse,
given by
(64)
The
Since
a"-
is
small
when p
is
member of equation
is
(59)
is \/3small, the approximate value of the right V3(l 2p ) Therefore the orbit whose period is
,
2ir/a has its axes, for small /x nearly coincident with the axes of the zero of relative The z-axis is in the line velocity. corrresponding curves
joining the finite body m with the equilateral triangular point, is of course at right angles to it.
Let the coordinates in the orbit whose period axes be and y its equation is then
;
is
Ai =
1
2 [(ai
cos
sin 0)
2 2 +&! cos
0],
(65)
Bi = \[(ai
sin
0+cos ey+bS
sin
0].
A?+Brf = l,
/io)
sin
<p
cos
<p
+ l sinV],
(66)
B
It follows
i-[fsin
3V3
(1
2/i
)
sin
(p
from
(58)
that
when
no is small the
approximate values
for
of a! and 6i are y/l/z and zero respectively. = is -30, it Is found from (65) that Mo
A# = Ofi Mo
1
lim
Hm
2 2
_Q
(67)
limit of the ellipse for mo = is a straight line through the origin and the position of n, and for small values of mo the eccentricity is near unity.
The
504
PERIODIC ORBITS.
of
<p
is
also
30
when
mo is small.
Hence
x
it
is
x
.
same value as A /B and the corresponding velocity have not only the same orientation, but they
A/B
from
(61), (65),
and
A_ ~ A r
l
CX
= 4.
Ct
whose period is 2w/a to zero relative curve of the that of velocity corresponding to the same value This is actually the of C is equal to the square root of this number, or 2. the to the curves of zero of orbits linear dimensions ratio of the of the limit
The
relative velocity as mo approaches zero. The discussion so far has pertained to the linear terms alone of the
differential equations.
The
terms of the
by the methods of Chapter VI. Consequently, for small values of the parameter X they give close approximations to the periodic orbits and the corresponding curves of zero The period 2-k/o is very long for small values of ju relative velocity.
periodic solutions which can be shown to
.
Now
mate
consider the periodic orbits whose period is 2ir/p. The approxivalue of p for small values of juo is unity, and from the equations correit is
sponding to (58)
orbits also
found that
b2
= ^,
Q 2=
3-^f.
tan 20 =
-V3
when
no
It is
(65) that
Mo
Aj _ i "*'
2
Therefore in these orbits the length of one axis is twice that of the other. The limit, for mo = 0, of the eccentricity of the orbits whose period is 2v/a is unity and the limit of the periods is infinity; the corresponding limits for
the orbits whose period
243.
is
2ir/p are
V3/2 and
2t.
Analytic Continuity of the Orbits about the Equilateral Triangular Points. The periodic solutions as developed by the methods of The coefficients Chapter VI are power series in X*, and they involve ju
.
The
of the
power
series are
continuous functions of
mo-
The
J
when
and complex when it has the other. As X passes through zero from one sign to the other, two real solutions for X unite and disappear by do not to the becoming complex. They belong physical problem except when \ = n po, but since the orbits exist for every value of fx distinct from zero it is easy to get an understanding of the situation from the behavior of
X has one sign
505
the more general solutions when X does not equal m~ Mo- Or, since the coefficients are continuous functions of fi this parameter can be considered as with X that their so sum is That is, the solutions and the period p. varying
,
can be expressed in terms of n and X by replacing ju by p X. The two real orbits which unite and disappear for X = are not geoThis appears to be an exception to the theorem that metrically distinct. real orbits appear or disappear only in pairs. It arises because in the analysis adopted the conditions that the orbit shall be periodic give a double
determination of the same orbit.
the five equilibrium points. the solutions are developed
the
= 0.
arise only at
Moreover, the matter is quite different when by the method of Chapter V. When the parameter t passes through zero the orbits do not disappear, but the same series is obtained for both positive and negative values of e, the origin of time belonging to a different place on the orbit. In the symmetrical orbits around the equilibrium points which are on the z-axis the origin of time is displaced by half a period. In the non-symmetrical orbits about the equilateral triangular points the origin is shifted from one point where the arbitrary
initial condition, e. g.,
x'(0)=0,
is satisfied
where
it is
also satisfied.
integral exists when the right members of the differential equations are not limited to their linear terms. Hence, in place of (60), the right member is an infinite series in x and y. When the expressions for
The Jacobian
x and y as
in X*, the
C becomes
a power series
term of the lowest degree in X being of the first degree. Conse5 quently the series can be solved for X as a power series in ^C*, and the result substituted for the solution in powers of X 5 will give x and y expressed
converge for \C\ sufficiently small. As C goes whose period is 2ir/a change from real to complex, through zero the orbits and those whose period is 2ir/p change from complex to real. There is a branch on each of the series at C = 0, but the two series are distinct.
as
power
series in
C which
5
,
When
ju
satisfies
the equation
1 -27*,(1 -mo)
-0
The Existence
of Periodic Orbits
gular Points for Large Values of /x. been for values of /x such that 1 -27it(l -/*) is positive. If it is zero, there is a double solution of zero dimensions. Suppose now that n has such a value that this function of n is very little less than zero, and take a*o so that
l-27/x (l-Mo) is a little greater than zero. Then there are real periodic When X orbits with the periods 2t/<t and 2t/p for X sufficiently small.
506
is
PERIODIC ORBITS.
do not belong to the physical problem. The analytic continuation of the solutions with respect to the parameter X can be made until X = /x Mo unless they have some singularity for a real positive value of X. It is very improbable that there is an infinity in the solutions for a real positive value of X because an infinity implies either an infinite branch of the orbit or one passing through one of the finite bodies. The orbit could not acquire an infinite branch without winding infinitely many Even if it should pass times, in the rotating plane, about the finite bodies. would be bodies its of finite the maintained, as was continuity through one
less
than
ju
Mo these orbits
seen in the case of other orbits of ejection in Chapter XV. would imply the existence of other real orbits which could
branch-point
identical
become
with the ones under consideration. It also seems very improbable that Therefore it will be there is such a branch-point for small variations in X. assumed, as being probable, that the periodic solutions can be continued to the ones belonging to the physical problem for X = n mo- This applies both
whose periods are 2t/<t and to those whose periods arc 2w/p. The possibility of their having acquired loops about one or both of the finite bodies by having passed through ejectional forms must, however, be
to those
This circumstance makes numerical verification difficult. as a limit, the value Now suppose the value of /z approaches 0.0385 of /x which satisfies 1 Mo) =0, and that the analytic continuation 27juo(l The expressions for the coordinates can be made with respect to X for all /x
admitted.
.
in the two classes of orbits are the same except that a and p are interchanged. a and p approach equality, and the correspondAs no approaches 0.0385 = A difficulty in attempting coming orbits approach identity for X /i fact that a certain determinant which is distinct arises from the plete rigor from zero in the proof of the existence of the solutions approaches zero as ju approaches 0.0385 .... But if it is admitted that the analytic continua.
. .
/*<>
made starting with any p it follows that even 0.0385 there is a double periodic orbit, and it than h larger surrounds a small real curve of zero relative velocity in the vicinity of one of the equilateral triangular solution points. As it is decreased toward the limit 0.0385 the dimensions of this double periodic orbit diminish toward zero as a limit. There is in this analysis a double determination of a double periodic orbit, just as of a single periodic orbit, and the two determinations coincide when it has zero dimensions. Consequently it can disappear at zero dimensions without uniting with another double periodic orbit. If h increases the double periodic orbit persists, according to the principles of 238, unless it becomes identical with another double periodic orbit. If there were another double periodic orbit with which it could unite it would envelop neither of the finite masses and would have two distinct branches which are symmetrical with respect to the z-axis. It is improbable in the extreme that there is another such double periodic orbit, which would mean
tion with respect to X can be
if
,
is
little
507
the existence of four single periodic orbits of the type under consideration. There are only two periodic orbits which shrink on the equilateral triangular
points, and others had loops about a
of the type could arise only
finite
The
existence or
two single periodic orbits which branch from it for values of the parameters which define the orbit, for example the Jacobian constant C, its
It should
,
be added, of course, that the two series of orbits may branch at the double orbit when considered with respect to one parameter, and form a continuous series when considered with respect to another.
245. Numerical Periodic Orbits about Equilateral Triangular Points. In accordance with the principles of 244, two periodic orbits about the
to \, equilateral triangular points should exist for all values of n from and for all values of C near that belonging to the equilateral triangular
equilibrium points. The only way they could cease to exist for at least some value of C would be for all of them to pass through an ejectional form for
every C. These orbits have an axis of symmetry only when p*. It is very difficult to establish by numerical processes the existence of a periodic orbit when it has no axis of symmetry because, for a given initial point, there are two arbitrary components of velocity, and interpolations must be made from a two-parameter family. Therefore the computations were
restricted to the case m
= 2-
It follows
from the
.
in this case the orbits in question have the fine x Since n = \ is far from the values (0 <; n ^ 0.0385 of the orbits in question
=
.
.)
was established by direct processes, those found by computation can not be expected to have much geometrical resemblance to those found by analysis. Since the surfaces of zero relative velocity expand with increasing C and unite on the rr-axis, it follows that either the periodic orbits about the equilateral triangular points unite in pairs and disappear with increasing
values of C, or they pass through the collinear equilibrium points with Therefore it seemed best to start computations for values infinite periods.
of
not much greater than that belonging to the equilibrium point, viz, 3. In attempting to discover periodic orbits about the equilateral triangular In 17 of these C was taken equal to 3.03; points 40 orbits were computed. in 16 it was taken equal to 3.20; in the remaining 7 it was taken equal to
3.284.
The
initial
were x = 0, y
values of the coordinates and components of velocity ' = 0, x ' determined so as to give the arbitrarily chosen, y
adopted value of C. The computation was continued until x became again equal to zero, and the approach to periodicity was determined by the approximation of yd to zero.
508
C = 3.03,
PERIODIC ORBITS.
Period
=2X4.388 =8.776
(Fig. 30).
509
there are two periodic orbits about the equilateral triangular points differing considerably in dimensions and periods; for C = 3.20 there are also two periodic orbits which differ less
for
C = 3.03
510
in dimensions
PERIODIC OKBITS.
and
periods;
and
for
C = 3.3284
there
is
The computations to a periodic orbit, though one was not actually found. indicate that the two series of periodic orbits unite and disappear for some value of C slightly smaller than 3.3284. But there is an orbit so nearly
periodic for
C = 3.3284
that
it is
511
finite
body
512
PERIODIC ORBITS.
which intersects the x-axis perpendicularly is symmetrical with respect to the x-axis. Hence, it follows that if one of these orbits of
Any
orbit
ejection intersects the x-axis perpendicularly, then it is a closed orbit of ejection of the type treated in Chapter XV. Computations were first made for m=z to discover orbits of the type
characterized
by j= 1 with
ejection
proved in Chapter XV that such an Such an orbit was found for m = i, but its its period is approximately 2r. period was about 8. Another orbit, also of a similar type, was discovered whose period was about 14. One of these orbits is undoubtedly the limit, for decreasing values of C, of the oscillating satellite about the collinear equilibrium point, as Burrau's calculations have indicated. The value of C corresponding to the equilibrium point for n = \ is about 3.46, and the
values of
in Fig. 15. It was orbit exists for small values of m and that
2.84.
The
the origin of the other orbit of this type. It is of a periodic orbit about 1 n consisting of a double loop and having a double point on the x-axis. Such orbits were treated by Poincare in Les Methodes Nouvelles de la Mecanique Celeste, Chapter XXXI. The ana-
former beyond the ejectional form for decreasing values of C is also a periodic orbit with two loops. For greater or smaller values of C the latter will have also the character of an oscillating satellite, but it can not reduce to the equilibrium point because there is only one orbit The results for n = \ are given in the tables of page 511. of this type. For /i = | similar results were found. Since orbits of this type have not been computed heretofore, the results for the four orbits will be given
lytic continuation of the
for
enough values
of
The corresponding
n=
l,
are
513
:.n
If
PERIODIC ORBITS.
n were zero and the infinitesimal body were ejected from 1 n either toward or from m in such a way that its period would be ir, the orbits described in rotating axes would consist of two parts symmetrical with respect to the These curves are the limits sepaz-axis, as shown in Figs. 40, a, and 41, a. m = 0) in the rotating plane, as is shown in Figs. 40, b, and 41, c. As n increases a dissymmetry develops with respect to the line through 1 fi perpendicular to the z-axis. Suppose the orbits are followed as n increases in such a way that they shall remain Then orbits of the type Fig. orbits of ejection in one way or the other. will some of the characteristics of both types a 40, a, go into types having
rating
two types
c of Fig. 41. That they partake of the characteristics of type c instead of those of type b was proved by computations for both n = \ and n = 4 The two following tables give the results for ejection from 1 m toward
and
n,
with n having the values \ and |. The first orbit lacks a little of being closed, but an exactly closed orbit exists between this one and the one which
for
was computed
C = 3.478.
515
516
certain cases
for
PERIODIC OKBITS.
by making use
of properties of
symmetry.
is
>
n = \,
to the
intersects the y-axis perpendicularly Hence it follows that if, for /* = t/-axis.
Any
then 1 fx intersects the y-axis perpendicularly, lision with the second finite mass.
it
After 14 computations had been made, an orbit of ejection from 1 n and collision with m (m = 5) was discovered in which the ejection was toward
n and in which the collision took place without the infinitesimal having it or /x. The following table gives encircled, in the rotating plane, either 1 the results at a considerable number of intervals from the time of ejection of
the infinitesimal
body from
ju
until
it
517
This orbit has a loop about the equilateral triangular point. It follows that there are two families of periodic orbits of the types shown in Fig. 46. len orbits were computed in an attempt to find one of them. The difficulties of making the calculations when the infinitesimal body was near one of the finite bodies were so great that wide departures from the orbits of Indications of such periodic orbits were ejection had to be attempted.
obtained, but none was actually found.
Axis
X-Axis
Fig. 45.
Fig. 46.
Ejection and of
in
Chapter
XV
Number of Closed Orbits of Orbits of Ejection and Collision when n = %. It was proved that when n is sufficiently small there are infinitely many
closed orbits of ejection, and reasons were given for believing that these The question of the existence of orbits of orbits persist for all values of nejection
and
collision
Fig. 47.
Fig. 48.
that there are infinitely many closed orbits of ejec= i- The differential equations of tion, and of ejection and collision, for M motion in fixed rectangular axes with the origin at the center of gravity of
It will
now be shown
d?x_
dl
2
i(x-xi )
jy
finite
i(x-Xj)
ry
d?y_
dt
2 t
h(y-Vi)
ry
h(yy*)
ry
(68)
where,
if
the
= 0,
*>
Xi=-|cos,
z 2 =+icosJ,
Vi=
h s in
Vi=
+h
sin
I.
(69)
518
PERIODIC ORBITS.
Now
let
x = r cos
0,
and y = r
sin
0,
after
which equations
(68)
become
(70)
where
r*
J
:r
+2/
2
,
n^^+i+r
it is
cos (6-t),
t
r 22
= r +\-r
2
cos (0-t).
(71)
Suppose
known
that, at
= T,
|'>0.
(72)
r*2,
Since
^+^}<^
it
and
i[^-^]<^r?
first
and
r(gf
IS
always positive,
follows
from the
d P>
of (70) that
(r
_i).v
V 6\
The
(ty>^+^iF +K=F(r)
Suppose
(74>
-57
>0
at
<
= T.
t.
Then
K in
Computations were made in which the infinitesimal body was ejected from 1 n both toward and from n with initial conditions corresponding to
K>0, and
?>2
(It
with
-r-
positive.
Hence
in
it
infinity.
Moreover,
follows from the second equation of (70) that, when referred to rotating axes, the infinitesimal body revolves infinitely many times about the finite
bodies,
H=
distance from the origin continually increases. Now consider, for example, a closed orbit of ejection from 1 n, for \, in which the infinitesimal body makes at least one circuit about the
its
and
finite
body
m-
once, but it does not cross the ?/-axis perpendicularly. Moreover, from the symmetry of the orbit with respect to the z-axis, that if
follows
crosses
the
half
an angle
a.
tt/2
+ a,
is less
Suppose it is the latter. Now suppose the initial conditions of ejection are changed so as to increase K. The orbit will cease to be a closed orbit of ejection and will tend toward one which winds out to infinity with continually increasing r. The angles at which the orbit crosses the
than
ir/2.
axes are continuous functions of the parameter defining the initial conditions, as for example. Hence the intersection with the y-axis which was at the
519
less than t/2 for the closed orbit of ejection will be exactly for a certain value of K. The orbit will be therefore an orbit perpendicular of ejection from 1 /x and collision with n.
and
consider an orbit of ejection from 1 n and collision with /x, crossing the x-axis at least once. From the symmetry of these orbits with respect to the y-axis it follows that they cross the x-axis an even number of times
if such an orbit crosses the x-axis once at an angle ir/2 +/3, where a positive quantity, then it also crosses it at an angle, of 7r/2 /3. If the initial conditions are so changed as to increase the constant K, the orbit ceases to be an orbit of ejection and collision, the angle corresponding to /3 eventually becomes 7r/2 greater than ar/2, and therefore, since it is a continuous function of K, there is at least one value of for which it is Such an orbit is a closed orbit of ejection. exactly w/2.
Now
and that
is
any value of K, there is a closed orbit of ejection, then for some larger value of K, corresponding to a smaller value of the Jacobian constant C, there is an orbit of ejection and collision; and that if, for any value of K, there is an orbit of ejection and collision, then for some larger value of K, corresponding to some smaller value of the
It follows
from
if,
for
is
Hence, for
fj.
= \,
closed orbits of ejection and collision. They are And since it has been all distinct because they have distinct values of K. for an orbit of ejection, r increases continuously to shown that, when
many
K>0
infinity, it follows
K corresponding to these
may be characterized by the number of times they cross the y-axis. For ejections in both the positive and the negative direction there are closed orbits of ejection from each of the finite bodies, and also orbits of ejection from one and collision with the = other, which cross the y-axis 2(2/+l) times, j 0, 1, 2,
classes of orbits are
bounded.
The
orbits
249.
The
(c)
the Evolution of Periodic Orbits about Equilibrium Points. evolution of the periodic orbits about the equilibrium points (a) and
On
1 n and m was traced, for of Burrau's values C, by computations from small ovals to the decreasing = 1 in = For % they are shown in Fig. 35, and for m ejectional form.
/j.
Beyond these forms they have loops about 1 fx. The periodic orbits about the equilibrium point (&) between 1 /x and n undergo corresponding evolutions. In the case m = 2 as the orbit, for decreasing values of C, becomes an orbit of ejection from one body
Fig. 37.
it
becomes an orbit
in Fig. 45. this
of
collision
is
shown
has loops about the finite bodies, and the motion in With decreasing values of C these loops is in the retrograde direction. these loops probably enlarge, the loop about each body eventually becoming an orbit of collision with the other body. In this case the orbit of
Beyond
form
520
PERIODIC ORBITS.
ejection and collision intersects the y-axis six times, tions being perpendicular.
two
of the intersec-
As C decreases, the ejectional and collisional form passes into a loop about the second body, which in turn expands and becomes an ejectional and collisional form with respect to the first body. In this manner the loops of the periodic orbit, with decreasing values of C, pass through ejectional and collisional forms, first with one finite body and then with the other, in a never-ending series, the ejectional and collisional forms being those shown to exist, for n = \, in 248, in which the ejections from each body are in the direction away from the other. They are characterized by the fact that they
2(2j+l) times, j = 0, 1, 2, If the finite masses are unequal the evolution of the periodic orbits is in a general way similar, except that the ejectional forms for the two masses do
cross the t/-axis
. .
not occur for the same values of C. Periodic orbits about the equilateral triangle equilibrium points have been shown in Figs. 30 and 34. With decreasing values of C they probably
and pass through ejectional forms, but ejectional forms in which the direction of ejection is not along the x-axis. Consequently they can not be discovered by numerical processes. If this conjecture is correct, for still smaller values of C they possess loops about the finite bodies, and for still smaller values of C the loop about each of the finite bodies may pass through an ejectional form with the other. There is, however, no evidence to guide conjectures.
increase in size
250.
On
There are
three direct periodic satellite orbits, two of which are complex for large values of the Jacobian constant, but all of which are real for smaller values
Suppose the orbits about the finite body n are under consideration. With decreasing value of C they can pass through ejectional forms. In fact, Darwin's computations showed that two of them were approaching such forms, one by approaching ju from the positive direction and the other by approaching it from the negative direction. The motion of the infinitesimal body when it is near collision is nearly the same as it would be if the mass of the second body were zero. Consequently its properties can be inferred from a consideration of the motion in the neighborhood of an ejection in the problem of two bodies referred to
of C.
rotating axes. It is clear that if the ejection is in the positive direction, the curve near the point of ejection lies on the negative side of the rr-axis, while if the ejection is in the negative direction the curve near the point of ejection
the ejection is in the positive direction, the two families of periodic orbits which are near the ejectional orbit both intersect the z-axis in the negative direction from the point of ejection, and the small complete loop about the point of ejection is then
lies
of the z-axis.
When
is
described in the
521
positive direction; while if the ejection is in the negative direction, the two families of periodic orbits which are near the ejectional orbit both intersect the x-axis in the positive direction from the point of ejection, but in this case
the small loops are both described in the same directions as in the other case. The x-axis consists of three parts, viz, that extending from oo to the
n to n, and that extending from y position of 1 ju, that extending from 1 to +00 If a periodic orbit intersects the x-axis perpendicularly in any one of these three parts before it goes through an ejectional form, then it will
.
same part
after it passes
through the ejectional form. Moreover, the branches of a closed orbit of ejection extend from the finite body with which there is collision in the direction opposite to that in which the neighboring periodic orbit intersects
the x-axis perpendicularly. In the case of the direct periodic orbit about 1 m which enlarges in the n from the negative direction, the positive direction and approaches 1 ejection is in the positive direction, the collision is in the negative direction,
and the orbit has the form shown in Fig. 42. The computation shows that it has two loops and, therefore, that it had two cusps symmetrical with
After this respect to the x-axis before it arrived at the ejectional form. orbit passes beyond the ejectional form, with decreasing values of C, it acquires a loop about 1 m> which intersects the x-axis perpendicularly in
the negative direction from n and which has a double point on the x-axis between 1 n and p.. Consider the further evolution of the periodic orbit. If the small loop
about 1 m should again pass to the ejectional form, the ejectional orbit would be exactly of the type of that from which the loop developed. It is improbable that such an additional ejectional orbit exists for another value of C. Now consider the possibility of that part of the orbit which crosses the x-axis perpendicularly in the positive direction between 1 n and n passing through an ejectional form. It can not pass to an ejectional form with n because, in accordance with the general conclusions respecting the motion near a point of ejection, the branches of the curve near n would lie in the positive direction from it, and the partial loop about n just before the ejectional form was reached would be described in the retrograde direction. But this branch of the curve could evolve to an ejectional form with 1 /x, when the orbit would have the form shown in Fig. 47. With decreasing values of C this orbit acquires an additional loop about 1 n, which is described in the retrograde direction and which intersects the x-axis perpendicularly in the negative direction between 1 y. and y.. This loop can expand and take an ejectional form with n, then acquire a loop about n, which can become an ejectional form with 1 n, and so on, being an ejectional form first with one of the finite masses and then with the other in a neverending sequence. The ejections from 1 n are all in the negative direction,
522
PERIODIC ORBITS.
It is probable, though not the positive direction. qualitatively the course of evolution of the direct satellite
all in
orbit
start
was made.
Now consider the direct satellite orbit about 1 /x which enlarges in the negative direction and which approaches the ejectional form from the
shown
The ejectional form was found by computation and is With decreasing C this orbit acquires a loop about 1 p, which may pass to the ejectional form with n, as shown in Fig. 48. The other
positive direction. in Fig. 44.
branch which crosses the z-axis perpendicularly may pass to the ejectional form with 1 it. With decreasing values of C this orbit acquires a small loop about 1 m which never again passes through the ejectional form. But the loop about n enlarges and becomes an ejectional form with 1 n with the ejection in the negative direction. Then follows a loop which becomes an ejectional form with n, followed by a loop about n which becomes an ejectional form with 1 n, and so on, first with one finite body and then with the other in a never-ending sequence. The ejections from 1 ft are in the negative direction, and from ju they are in the positive direction. There is a third direct satellite orbit whose evolution has not been traced. Only a conjecture can be made in regard to it, and that conjecture is that it acquires cusps and then loops, probably about the region of the
equilateral triangular points.
251.
On
sider the retrograde periodic satellites about 1 n. orbits, only one of which is real for large values of C.
the Evolution of Retrograde Periodic Satellite Orbits. ConThere are three such
238, indicate that only one of
the retrograde periodic orbit about 1 n is small and nearly circular in form. As C diminishes the orbit increases in size and departs widely from a circle. Consider the question of its passing
of
through an ejectional form with 1 fi. the ejectional form by shrinking upon
If
1
it
the periodic orbit should approach from the positive or the negative
direction, just before arriving at the ejectional form it would make a partial loop about 1 m in the retrograde direction, and just after passage through
it would make a complete loop about 1 ju in the positive was seen in 250, in connection with a consideration of an ejectional orbit in the problem of two bodies referred to rotating axes, that this is impossible. Hence the retrograde satellite orbit about 1 a* can not become an ejectional orbit with 1 n, at least until after it has passed through an ejectional form with m-
But
it
Now
1
/x
consider the possibility of the retrograde satellite orbit about Since it intersects the x-axis passing through a collisional form with n.
1
it
between
and
it
negative direction, in
such an orbit must be one in which the collision is in the which the ejection is in the positive direction, in which
523
the partial loop just before collision is described in the positive direction, and in which the complete loop just after collision is described in the retrograde
direction.
precisely the way in which such a limiting form can be orbit passes through this form. An orbit of this ll and m was type, with the roles of 1 interchanged, computed and is shown in Fig. 39.
is
This
After the retrograde periodic orbit about 1 ll passes through an ejecform with ll, it acquires a retrograde loop about ll which crosses the x-axis in the positive direction between 1 ll and ll. This loop enlarges and
tional
m, after which it acquires a retrograde loop about 1 ll, which, in turn, enlarges and passes through an ejectional form with fi. This process continues, the form becoming ejectional first with one finite mass and then with the other in a never-ending sequence. In all of these orbits the parts near the ejection points are on the negative
1
side of
1 ll or the positive side of /x, and never between 1 orbits of these series which are closed orbits of ejection with
ll
and
it
ll.
The
are a part
which were shown to exist in Chapter XV for sufficiently small values of ll; and those which are closed orbits of ejection with ll are the
of those
corresponding orbits for the other finite mass. Consider first the orbits of the type under consideration which are orbits All these orbits are orbits of ejection ll. of collision and ejection with 1
negative direction; they have double points on the x-axis in the positive direction from ll, and intersect the x-axis perpendicularly Only in the
in the
are therefore only those orbits of 226 by ejection in the negative direction and by even values of j; those characterized by odd values of j have a different origin. The orbits of the type under consideration which are orbits of ejection
1
ll.
They
lx
negative side of
also intersect the x-axis perpendicularly only on the ll. On interchanging the roles of 1 it and ll in 226,
ll
intersect the x-axis perpendicularly are of the type of the part of those
of collision
Those which are characterized by odd values of j on the negative side of 1 ll. They under consideration which are orbits
satellite orbits about 1 ll, with decreasing values of C, go through an infinite series of ejectional forms with 1 ll, the ejections all being in the negative direction, and these orbits ll in the are those of the orbits treated in 226 which are ejected from 1 of The are characterized even values and which j. by negative direction retrograde periodic satellite orbits also go through an infinite series of ejec-
being in the positive direction, and these orbits are those which can be shown to exist by the methods of 226, and which are characterized by ejection in the positive direction from ll and by
ll,
the ejections
all
odd values
of
j.
satellite orbits
about
524
m.
PERIODIC ORBITS.
and they go through. a similar series of critical ejectional forms with both 1 n and m- The ejections from 1 n and n are also respectively in the negative and positive directions, but those which are ejectional forms with 1 n which are of those while are characterized by odd values ejectional forms
,;',
with
ju
are characterized
by even values of
two
finite
226, in ejectional forms from both finite bodies, of the type treated in in from one is the direction from which the ejection the other. body opposite
the Evolution of Periodic Orbits of Superior Planets. It was shown in Chapter XII that for large values of C there ai^ two periodic orbits in which the infinitesimal body makes simple circuits alrout both of the
252.
finite
On
bodies in the retrograde direction. When the system is referred to There are also fixed axes, one of the orbits is direct and one is retrograde.
four orbits in which the coordinates are complex, two of when referred to fixed axes and two being retrograde.
whether or not either pair of the complex orbits becomes real with decreasing values of C. Since none of these orbits was computed, very little is positively known about their geometrical characteristics or about their evolution. It was shown in 248 that there are two infinite sets of orbits which are orbits of ejection from one finite body and of collision with the other. One set is characterized by the fact that the ejection from each finite body is in the direction away from the other. Reasons were given in 249 for believing that they are limiting forms of the analytic continuations of the oscillating satellites about the equilibrium point b. The other set is characterized by the fact that the ejection from each finite body is toward the other finite body. These orbits are probably limiting forms of the analytic continuations of retrograde periodic planetary orbits. The probable series of changes to the first limiting form is shown qualitatively in Figs. 49 and 50. Beyond the limiting form the orbits acquire loops about each of the finite bodies.
There are two retrograde periodic orbits of the types of superior planets. Probably they both undergo evolutions to limiting forms of the types deThis conjecture is supported by the fact that two closed orbits of scribed. found b)*- computation, 247, for a related type of orbits. were ejection
...
i".
Fig.
r>0.
IN
=SOOK POCKET
I-!
&
Sci.
Applied