Professional Documents
Culture Documents
Treatise On Cal Cul 00 Geor
Treatise On Cal Cul 00 Geor
Treatise On Cal Cul 00 Geor
NO..
LIBRARY
OF
Arthur
D. Butterfield
tiLCi*
{rAl^C*^
(Kambrfoge:
PRINTED BY
C.
J.
CLAY,
M.A.
A TREATISE
ON THE
BY
GEORGE BOOLE,
D.C.L.
;
HONORARY MEMBER OF THE CAMBRIDGE PHILOSOPHICAL SOCIETY PROFESSOR OF MATHEMATICS IN THE QUEEN'S UNIVERSITY, IRELAND.
&ambrt&<je:
MACMILLAN AND
AND
23,
CO.
PREFACE.
In the following exposition of the Calculus of Finite Differences, particular attention has "been paid to the connexion
of
its
Differential
Calculus
is ^in
my
Mr
And
it
has been
me much
sheets.
In offering him
my
am
free
work
will be found to be
from important
errors.
GEORGE BOOLE.
8,
i860.
Gordon
Bell
http://archive.org/details/treatiseoncalculOOgeor
CONTENTS.
PAGE
CHAPTER
CHAPTER
Differences of Elementary Functions, 6.
I.
.
factorials,
n.
16.
Generating Functions,
Differences of o, 20.
Exercises, 25.
14.
Laws and
A and -j-
Secondary
Form
CHAPTER
Nature of the Problem,
distant,
33.
III.
.
OF INTERPOLATION
28.
.28
equi-
29.
Not
Areas of
Curves, 36.
CHAPTER
Meaning of Integration, Summation of 48.
45.
IV.
.
FINITE INTEGRATION
Periodical Constants, 47.
.45
Condi-
Integrable Forms,
59.
Series, 56.
Connexion of Methods,
Exercises, 63.
CHAPTER
Definitions, 65.
69.
V.
.
.65
66.
Supplemental
Exercises, 79.
CHAPTER
Development of 2wx, 80.
Bernoulli's
YI.
.
.
80
Applications, 84.
Limits of the Series for 2w*, 91. Otherforms of 2w,, 94. Exercises, 97.
CONTENTS.
PAGE
CHAPTER
Genesis, 99.
VII.
. .
EQUATIONS OF DIFFERENCES
Linear Equations of the
constant Coefficients,
first
.99
orders, 10 1.
Linear Equa-
tions with
106.
Analogy with
Differential
Equations,
118.
Fundamental
Exercises, 123.
CHAPTER
VIII.
EQUATIONS OF DIFFERENCES OF THE FIRST ORDER BUT NOT OF THE FIRST DEGREE .125
. .
Law
of Reciprocity, 132.
Exercises, 150.
CHAPTER
IX.
.
.151
158.
Finite Solu-
Exercises, 178.
CHAPTER
X.
.
OF EQUATIONS OF PARTIAL AND OF MIXED DIFFERENCES, AND OF SIMULTANEOUS EQUATIONS OF DIFFERENCES 179
Equations of Partial Differences, 182.
191.
Method
of Generating Functions,
Simultaneous Equations,
205.
CHAPTER
Direct Problems, 209.
218.
XI.
. .
208
Functional Equations,
Exercises, 229.
CHAPTER
Exercises, 241.
XII.
.
GEOMETRICAL APPLICATIONS
Miscellaneous Examples, 242.
.232
Answers
to the Exercises
245
FINITE DIFFERENCES.
CHAPTER
I.
The Calculus of Finite Differences may be strictlj 1. defined as the science which is occupied about the ratios of the simultaneous increments of quantities mutually dependent. The Differential Calculus is occupied about the limits to which such ratios approach as the increments are indefinitely diminished.
In the latter branch of analysis if we represent the independent variable by x, any dependent variable considered as a function of x is represented primarily indeed by < (x), but, when the rules of differentiation founded on its functional character are established, by a single letter, as u. In the notation of the Calculus of Finite Differences these modes of expression seem to be in some measure blended. The dependent function of x is represented by ux the suffix taking the place of the symbol which in the former mode of notation is enclosed in brackets. Thus, if u x (/> (x) then
,
ux+h =
w inx
S
(j>(x
+ h),
^>(sin^),
and so on. But this mode of expression rests only on a convention, and as it was adopted for convenience, so when convenience demands
it is
laid aside.
The
and
step of transition from a function of x to its increment, still further to the ratio which that increment bears to
the increment of x, may be contemplated apart from its subject, and it is often important that it should be so contemplated, as an operation governed by laws. Let then A prefixed to the expression of any function of x, denote the operation of taking the increment of that function correspondB. F. D.
if
[CH.
I.
ing to a given constant increment Ax of the variable x. Then, representing as above the proposed function of x by u x
we have
and
Aux
-r~^= A#
u x+ a x -ux
~n
A#
A#
is
the fundamental
But
ought
there
to
is
be noted.
du
-7- is
not a
true fraction, nor have du and dx any distinct meaning as symbols of quantity. The fractional form is adopted to express the limit to which a true fraction approaches. Hence
-7,
and not
d, there represents
a real operation.
is
But
in the
Aw Ax
a true fraction.
Its
nu-
merator Au x stands for an actual magnitude. Hence A might itself be taken as the fundamental operation of this Calculus, always supposing the actual value of Ax to be given and the Calculus of Finite Differences might, in its symbolical character, be defined either as the science of the laws of the operation A, the value of Ax being supposed given, or as the science of
;
mental difference above noted between the Differential Calculus and the Calculus of Finite Differences, the term Finite ceases to be necessary as a mark of distinction. The former
is
Though Ax admits
it is
of
usually given to
unity.
any constant value, the value There are two reasons for this.
First, the Calculus of Finite Differences has for its chief subject of application the terms of series. Now the law of a
ART.
2.]
OF FINITE DIFFERENCES.
however expressed, has for its ultimate object the determination of the values of the successive terms as dependent upon their numerical order and position. Explicitly or implicitly, each term is a function of the integer which exseries,
presses
its
position
in
the
series.
And
thus, to revert to
language familiar in the Differential Calculus, the independent variable admits only of integral values whose common difference is unity. In the series of terms
2
I
92
2
.
q2
A2
It is an explicit function of x. the general or xth term is a? but the values of x are the series of natural numbers, and
A#=l.
Secondly. When the general term of a series is a function of an independent variable t whose successive differences are constant but not equal to unity, it is always possible to replace that independent variable by another, x, whose common difference shall be unity. Let <j> (t) be the general term of the series, and let At = h; then assuming t = hx we have
At
= hAx,
it
whence
Ax =
1.
Thus
variable
suffices to establish
finite
unity.
At
the rules of the calculus on the of the independent the same time it will be noted that this
difference
and A.
shall therefore in the following chapters develope the theory of the operation denoted by A and defined by the equation
We
general operation
Aux _ ux+h - ux Ax h
where
Ax = lu
12
CHAPTER
II.
1.
The
operation denoted
by
is
capable of repetition.
For the
same kind.
indices, the
2 to
AAu =AV.
In like manner
AAV
and generally
AA-X=AX
the last
th
(1),
ux
If
member being termed the n difference of the function we suppose ux = xs the successive values of ux with
,
their successive differences of the first, second, will be represented in the following scheme
Values of x
1 1
4 64
61
30...
6...
216...
ux
8
19
27
125
91
Aux
37
A\x
12
6
18
6
24
6...
AV
It may be observed that each sum of differences may either be formed from the preceding sum by successive subtractions in accordance with the definition of the symbol A, or calcu2 lated from the general expressions for Aw, A u, &c. by assign-
ART.
2.]
5
5
,
&c.
Since u x
x we
have
A ux = 6.
3
It
stant.
may also be noted that the third differences are here conAnd generally ifux be a rational and integral function
the
of x of
nth
degree,
its
nth
differences
will be constant.
For
let
ux
then
-1
- axn - bxn~
l
b l9 b2
&c, being
constant
coefficients.
Hence Au x
is
rational
and integral function of a? of the degree w1. peating the process, we have
Ke-
n 2 and
;
so on.
Finally
we
shall
have
... 1,
A\ = aw(n-1) (w-2)
a constant quantity.
Hence
also
we have
A=1.2...w
2.
(2).
by A,
operation or series of operations denoted when the subject function u x is given, there are certain elementary cases in which the forms of the results are deserving of particular attention, and these we shall next consider.
While the
2
,
...
An
[CH.
II.
Let ux
= x (x 1)
(x
2)
...
(x
m+ 1).
Then by
definition,
When
by a
sin
sin (x
+ h)
sin (x
+ 2h)
...
sin [x
+ (m 1) h],
the factors are usually called factorials, and the term in which they are involved is called a factorial term. For the particular kind of factorials illustrated in the above example it is common to employ the notation
x{x-l)
doing which,
...
(x-m+l)=xW
(i),
we have
Ax^^mx^
Hence, x
[m~ x)
(2).
A xW = m(m-l)a;(
2
ro - 2)
and generally
An x^ = m(m-1)... (m-n+l)^'^
2ndly. J
(3).
Let ux
x(x +
-r l) ...
-,
(x
+ m-1)
=-r
Then by
definition,
1 1
...
A Wx =
=
{x+1)
(x
+ 2)
(x
+ m)
I
x(x+l)
...
(x
+ m-1)
(A
\x + m
I) xj (x
+ 1)
(x
+ 2)... (x+m-1
(4),
x(x+l)
Z^
...
(x
+ m)
'
ART.
2.]
x(x+l)
we have
...
(x
+ ml)
1
)
#(-)
A("
M
)
= -^'m'
(5).
Hence by
AV
and
this
ffl
>
= -m(-m-l)
=
(-l)
n
...(-fw-n
...
+ lJaT"^
'n
)
m(m + l)
(m
+ 72-l)^-m
(6),
may be
regarded as an extension of
(3).
3rdly.
factorials,
we
find
. . .
i\U xUx_ x
[U x+1
Ux-m+t) * U xUx-t
Ux-m+2
A
and
WxMz+
Ux+m-i
u*
~ "**"
U xUx+\
Wx+2?w
(8),
in particular if
ux
ax + h,
= am
x
Luxux_
. . .
ux_m+1
u x_ x
. .
wx_ m+2
(9),
=
U x Ux+i
Ux+m-i
U x Ux+i'" Ux+m
(10).
In like manner
we have
u
T
.
we may
Alog. =
So
also
log(l+^)
= log a*"x-m+i
a
(U).
A log (,_,
4thly.
. . .
ux. mH )
(12)
To
[CH.
II.
We
have
= (a-l)ax
Hence
(13).
AV=(a-l)V,
and generally,
Hence
also, since
(14).
(15).
To deduce
+ b)
+ b)
(ax
+ b + a) sin (ax + b)
(
ax
+b+
,
i)
a
7r +
sin
lax + 6
By inspection
we
see that
A
And
A*
+ a + it)
(16).
generally,
sin (ax
^^L
^
2
h),
An cos(a^ + 6) = ('2sin|Ycosjax + J +
sines
^l...(18).
These results might also be deduced by substituting for the and cosines their exponential values and applying (15).
3. The above are the most important forms. ing are added merely for the sake of exercise.
The
follow-
ART.
4.]
To
-1
ux
ux+1
sm ux
cos
cos ux+1
sin (ux+1
- ux
Au
a)-
Next,
A tan
-1
ur
easily
shewn
that
A tana# =
cos
a#
1
cos
a(x
=
+ l)
=-=
2
-t
(oj,
+ a?x + a x*
(4). v
'
When
the increment of
is
tion denoted
by
-r
merges, on supposing
Ax
to
become
remain unchanged,
following are illus-
by
-=-
The
mode in which some of the general theorems of the Calculus of Finite Differences thus merge into theorems of the Differential Calculus.
10
[CH.
II.
Ex.
We have
A sin x _ sin (x + Ax) sin x
Ax
Ax
2
* smj Aa?
1
/
(
sin
03+
Ax+tt\ -
Ax
And, repeating the operation n
An w A sm x _ n ~ {Ax)
.
times,
sin ^ (2 v 2 Ax) y
sin
V
oj
+n
/
'
(Ax)
'"
is
da?
a
^=
Aa*
S in(-
+ f)
(2),
known theorem
Again,
we have
x+Ax cr*
- x
a;
Ax
Ax
.Aar
-os-vAnd
hence, generally,
(A)
AV =(^)"become
infinitesimal, this gives
by
the
(4).
Supposing Aa;
to
2r = (loga)"a'
But it is not from examples like these to be inferred that the Differential Calculus is merely a particular case of the Calculus of Finite Differences. The true nature of their connexion will be developed in a future chapter (Chap. viii.).
ART.
5.]
11
Attention has been directed to the formal analogy between the differences of factorials and the differential coefficients of powers. This analogy is further developed in the following proposition.
5.
To
of
develope
of the
th
1st.
Assume
<l>(x)=a
+ bx + cx + dx
{2)
{3)
...+hx {m)
(1).
legitimacy of this form is evident, for it represents a rational and integral function of x of the ntb degree, containing a number of arbitrary coefficients equal to the number of coefficients supposed given in <f> {x) And the actual values of the former might be determined by expressing both members of the equation in ascending powers of x, equating coefficients and solving the linear equations which result. Instead of doing this, let us take the successive differences of (1). find by (2), Art. 2,
.
The
We
(2),
2)
...(3),
Aw <(x) = m(m-1)
And now making x =
...1A
(4).
and representing by A$(0), A 2 <(0), &c. what &c. become when x = 0, we have
</>
A <(#),
2
(0)
= a,
A0 (0) = b, A
2
</>
(0)
2c,
A m (/>(0) =
Whence
determining
a, b, c,
1.2...mA.
...
h,
we have
.
(5 ).
12
[CH.
II.
we
by proceeding
employ-
Ax = h,
4>{x)\
x(x h)
2
+
where the brackets
after reduction,
a? is
(A
<
(a))
^-A)(*-2A)
~iT2.3
JT^n
{ }
-+** w
to
Taylor's theorem is the limiting form to which the above theorem approaches when the increment Ax is indefinitely
diminished.
and
of functions.
+~ Ux= dx~ + 2
_dux
d*u x
Y ~dx
x + 2l~d^ + &C
d su
'
we
of
a function with its successive differential coefficients. these fundamental relations spring many general theorems expressing derived relations between the differences of the higher orders, the successive values, and the differential
From
coefficients of functions.
As concerns the history of such theorems it may be observed that they appear to have been first suggested by particular instances, and then established, either by that kind of proof which consists in shewing that if a theorem is true for any particular integer value of an index n, it is true for the next greater value, and therefore for all succeeding values or else by a peculiar method, hereafter to be explained, called the method of Generating Functions. But having
ART.
7.]
13
been once established, the very forms of the theorems led to a deeper conception of their real nature, and it came to be understood that they were consequences of the formal laws of combination of those operations by which from a given function its succeeding values, its differences, and its differential coefficients are derived.
Ex.
Kequired
to express
its
suc-
cessive differences.
We have
ux+1 =ux +Aux
.*.
;
u x+2 = ux +
Aux + A
we
(ux
+ Aux
.
= ux +2Aux +A*ux
Hence proceeding
w*+3 =
as before
find
ux + 3Aux + 3A 2 u x +
V
A*..
by the agreement of their with those of the successive powers of a binomial, the general theorem
These
special results suggest,
coefficients
= ux + nAux + fa"
71
1.2
..
n +-
(n
1) (n
la g
~3
2)
., p -A\+&c
(1).
Suppose then this theorem true for a particular value of n, then for the next greater value we have
u x+n+i
= ux + nAux +
n in K
1)
l2
n
A2 A ux
2
+
+ Aux + nA%+
n
(n
1)
(n
2)
tt^ 1.2.3
&c>
p A ux + &c.
., 5
&^ AX+
14
[CH.
II.
= ux +
*\ a 1) Am,
(n+l)n
n (n AX + (n + j^g
A
l)
1)
*
A v AV.+ &c.
the form of which shews that the theorem remains true for the next greater value of n, therefore for the value of n still succeeding, and so on ad infinitum. But it is true for n = 1, and therefore for all positive integer values of n whatever.
8.
We
by
the
method of generating
Definition.
series of
If
powers of
.
represented by function of ux
is capable of being developed in a the general term of the expansion being ux f, then </> (t) is said to be the generating And this relation is expressed in the form
(t)
t,
</>
(t)
= Gux
1
Thus we have
since
A. %
is
the coefficient of
in the development of e\
it/
In like manner
6
-r=G
t
1.2... (#
+ !)'
f
in the
since
J
.
.
.
a)
is
the coefficient of
development
Ji
[SC "J~
of the
first
member.
And
generally, if
Gux
<$>
(t),
then
<*w-*^
Hence
therefore
ch^-ip.
2 ).
Gux+ -GuM\-l)<f>{t)
t
But the
first
member
is
obviously equal to
GAux
therefore
<?A,=
(i-l)^(0
(3).
ART.
8.]
15
And
flA>.-(j-l)"*W
tion
Wconsidera-
To apply these theorems to the problem under we have, supposing still Gux $ (t),
= Gux + n GLux +
u x + n&ux +
(w
l)
GA\i x + &c.
n (n
2
^ AX+ &c|
W
(W
Hence
.
1)
(1).
made
Although on account of the extensive use which has "been of the method of generating functions, especially by the older analysts, we have thought it right to illustrate its general principles, it is proper to notice that there exists an objection in point of scientific order to the employment of the method for the demonstration of the direct theorems of
the Calculus of Finite Difference; viz. that is, from its very nature, a symbol of inversion (Biff. Equations, p. 375). In applying it, we do not perform a direct and definite operation, but seek the answer to a question, viz. What is that function which, on performing the direct operation of development, produces terms possessing coefficients of a certain form? and this is a question which admits of an infinite variety of answers according to the extent of the development and the kind of indices supposed admissible. Hence the distributive property of the symbol G, as virtually employed
16
in the
[CH.
II.
above example, supposes implied in the mere definition of supposed to have reference to the the one member as in the other ; tions being supplied, it becomes a indirect character still remains.
the symbol.
same system of indices in and though, such convenstrict method of proof, its
9. We proceed to the last of the methods referred to in Art 6, viz. that which is founded upon the study of the ultimate laws of the operations involved. In addition to the to denote the opesymbol A, we shall introduce a symbol ration of giving to a? in a proposed subject function the increment unity ; its definition being
Dux =ux+1
Laws and
1st.
(1).
D,
and
-=
The symbol
A
+
is
Thus
(2).
A
For
(ux
(ux
+ vx + &c.) = ux+1 + vx+l ... - (ux + vx ...) = UX+1 -Ux + Vx+1 -Vx ... Aux + Avx ...
(3).
2ndly. The symbol is commutative with respect to any constant coefficients in the terms of the subject to which it is applied. Thus a being constant,
= aAux
And
we
have, a,
(4).
from this law in combination with the preceding one, b,... being constants,
(5).
ART.
9.]
17
A MAX=AM+
\
by
(6),
m and
n being positive
rn,
indices.
For,
of the index
A mA\= (AA...W
(A
M+
...(m
+ n)
times} u x
= A X.
These are the primary laws of combination of the symbol A. It will be seen from these that A combines with A and with constant quantities, as symbols of quantity combine with each other. Thus, (A + a) u denoting Au + au, we should have, in virtue of the first two of the above laws,
(A
u
(7),
the developed result of the combination (A + a) (A + were a symbol of quantity. in form the same as if
b)
being
The index law (6) is virtually an expression of the formal consequences of the truth that A denotes an operation which, performed upon any function of x, converts it into another
function of
Perhaps
tition
;
the same operation may be repeated. might with propriety be termed the law of repeas such it is common to all symbols of operation,
x upon which
it
except such,
subject
to
if
repetition.
which they are applied, as to be incapable of It was however necessary that it should be disbecause
it
tinctly noticed,
The laws which have been established for the symbol A are even more obviously true for the symbol D. The two symbols are connected by the equation
Z>
= l + A,
ux
(8),
since
Dux = ux
B. F. D.
+A^= (1 + A)
18
[CH.
II.
and they
by
the relation
D=e*
(9),
For
~
1)Ux X
1 d'
d u
f^dld + =
1
+ dx+2dx^
dz
27zdx*
p + &C ')
U*
e*u x
D, A, and
<*
dx
are connected
by the
o equations
Z)
= l+A = 6^
(10),
fact that
and
by
means of
and
-j-
-=- it
combine,
other, as if they
(Differential Equations,
In the following section these principles will be 10. applied to the demonstration of what may be termed the direct general theorems of the Calculus of Differences. The conditions of their inversion, i.e. of their extension to cases in which symbols of operation occur under negative indices, will be considered, so far as may be necessary, in subsequent chapters.
Ex.
its
To develope ux+n in a series consisting of u x and 1. successive differences (Ex. of Art. 7, resumed).
definition
By
Dux:
ux+2 = D*ux
&c.
ART.
10.]
19
Therefore
= i>X=(l+A)%
0).
|'1
+ wA+ !^li)A' +
w(w -
(w
- 2)
A'...
= u. + n
Ex.
values.
2.
^+
^
n-1
2 \8
AX + W(m
~1
(re
- 2)
2 -3
AX+&c.(2).
its
To
express
A nu x
in terms of
wx and
successive
Since
Au x = ux+1 ux = Du x u x we
,
have
Aux = {D-l)ux
and as, the operations being performed, each side remains a function of x,
Anux =(D-l) n ux
= inn - nD +
n
1)
^g
i>
n "2
+ &c.| wx
y2
w x+n _ 2 ...+ (-
1)X
(3).
Of
the
We have
A nxm =(x+nyn -n{x+n-l) m +
Now
first
n( "~ lS)
j.
(x+ n -
w
2)
-&c...(4).
A
n
rt
be adopted
to express
what the
;
member
n
when x =
then
= n -n(n-iy
1.2
n{n-l)(n-2) m
n (n-1) (n-2)
1.2.3
(w
3)
&c
^
V
;
are of frequent The systems of numbers expressed by occurrence in the theory of series. Professor De Morgan has
An
22
20
II.
given the following table of their values up to A from the above theorem. (Dlff. Calculus, p. 253) A
o
o^
10
calculated
A2
o
2
A3
A4
A5
A**
A7
A
c
A^
A 10
M
<l
I
;i
6
14
o4
36
24
240
o5
o
l!
'
30
62
150
120
'
540
1806
1560
1800 16800
720
o7
126
8400
40824
15120
T91520
1905120
5040
141120
o8
o o
1
!
254
579 6
126000
40320
1451520
'
834120
2328480
362880
'
3628800
From
(2) Art.
we have
Aw 0*=1.2...w,
and, equating this with the corresponding value given
by
(5),
we have
1 .2
. . .
n = nn - n
(n
- l) n + "fo" 1
)
(
_ 2 ) + & c
(6)
Ex. 3. To obtain developed expressions for the n th difference of the product of two functions u x and v x
.
femce
&uxv x = u x+i
v x+1
- ux v x
,
D applies
to
to v x alone,
,
we have
A" MA
= (M-1)\
(7).
It now only remains to transform, if needful, and to develope the operative function in the second member according to the nature of the expansion required.
21
Thus if it be required to express A uxv x in ascending differences of vx we must change D' into A'+ 1, regarding A' as operating only on v x . then have
,
We
A\A =(D(1+A')-1)\^
- (A + DA!) nuxvx
" A n + nA^DA' + "fa-V A W D A' + &c.i
2
ux v x
Remembering then that A and operate only on ux and A' only on v x and that the accent on the latter symbol may be dropped when that symbol only precedes v x we have
,
uxv x
=A
u x .vx
+ wA*-X+1
At?.
+
the expansion required.
^A"X,Ax + &c
= ax
.
(8),
As
Then
since
A
we have
n 'r
ux ^ r
=A
=a
n
n-r
x+r
=a A
r
n-r x
x+r
{a-l)
n-r
,
by
(14), Art. 2,
A naxvx = ax {(a-l
vx
+ n(a- l^aAv,
(9).
+ ^(-irVA\-+&c.)
Again,
if
is to be ordered according to sucnecessary to expand the untransformed operative function in the second member of (7) in ascending powers of and develope the result. find
the expansion
,
cessive values of vx
it is
We
^A = (~
we have
l)
K* - nux+1 vx+1 +
71 [71
v
~ 1
is to
D into + A,
+ A',
(11),
now
to
be
22
[CH.
II.
Ex.
4.
To
ux
.
express
A nux
efficients of
1st.
By
(10), Art. 9,
A = edx - 1.
Hence
(12).
we have
(^i r= ^ +
=
on expansion,
n
t
_ + __ +
n+1
1
&c
.J
(13),
+A
+A
n+2
2
t
+&c,
coefficients.
Av A
Hence
and therefore
^=(i)"^ + ^(ir^ + ^(i)^ +&c -- (u) The coefficients A A ...&c. may be determined in 11.
-
various ways, the simplest in principle being perhaps to develope the right-hand member of (13) by the polynomial theorem, and then seek the aggregate coefficients of the successive powers of t. But the expansion may also be effected with complete determination of the constants by a remarkable secondary form of Maclaurin's theorem, which we shall proceed to demonstrate.
(t)
in positive
and
integral
is possible,
may
be expressed
in the form
</>(
becomes when
0.
23
First, we shall shew that if <f>(t) and ^fr(t) are any two functions admitting of development in the form a + bt + cf +
&c,
then
*(2 + M- + ()*M
0, after
<
15) >
For, developing all the functions, each member of the above equation is resolved into a series of terms of the form m A fd\j tn while in corresponding terms of the two members
i
m and
n will be reversed.
if
Now
and
n
(
-y
t
J
is
equal to
is
greater than n,
to
1.2...ra if
is
at the
same time
if
t
factor.
Hence
= 0,
dt)
\dt)
'
true, or,
and therefore under the same condition the equation adopting the notation above explained,
(15) is
**(o) = f*(o)
Now by
*(*)
(16).
Maclaurin's theorem in
its
known form
= <Mo)+^(o).*+JU(o).
^+&c
(17).
= ^.(0)+^g)o.* + ^g)o
2
.
^+
&c....(18),
theorem
(17),
(18)
may
conjugate.
24
II.
As
(e
l) n n being
,
= 0,
e
"=( iv *U
we have
(*
-1
r
'
)
An
l) n
= An
+ An
if
1.
2
.
f- + A"0 -^ + &c.
3
.
But
if ra is
An
is less
than
n,
and
to l.2...n
Hence
.^ +1 +
e ',i) ( ;
r+f 1.2...W +
An
n+1
3-^ 1.2...W
An
M+2
+2
+2
The
coefficients of this
expansion are
now
given by the
that table
is
Hence
An
therefore, since
A nu =
dn+1 u
dx
(e
l)
rt
u,
we have
d n+*u
,, A
.
d nu
An
n+1
Aw
n+2
As
a particular example
l)...(m
we have
A nxm = m (m -
- n + 1) xm ~ n
1
...
+ AM
...(21).
v.
The reasoning employed in the above investigation proceeds upon the assumption that n is a positive integer. The very important case in which n = 1 will be considered in another chapter of this work.
Ex.
of
u.
5.
To express
d nu
-j-^ in terms
25
Since
= 1 + A, we
;|=log(l
therefore
+ A),
+ A)!*
now be developed
(22),
= f-^V dx)
K
{log
(1
in as-
Aw _ = AM __ __ +&c +lf
du
2
= 1, we have Aw Am
3 4
(23 ).
As a particular application, suppose u=x(xl)... (x w+1), then adopting the notation of Art. 2,
Aw = nx \ A2u = n(n-l)x
{n
~l
[n ~ 2
,
&c.
Therefore
^
dx
" = na!*"* - n n
(
1)
a*-*
^("-iH-^Mfc
o
(24))
coeffi-
It would be easy, but it is needless, to multiply these general theorems, some of those above given being valuable rather as an illustration of principles than for their intrinsic importance. Other examples will occur in the course of
this
work.
EXEKCISES.
1.
sin a
sinaicsina (x+1)
'
26
EXERCISES.
[CH.
II.
A 2* sin -, = 2*+2 sm
tan
^
-,
^sm
^J
Atan-,=
cos-,
A cot (2'a)
2.
sm (2
a)
Shew
that
A may
wa v 'x xv x+\
3.
If
Ax=a
/t
A
'
and
Aux = ux+a ux
prove that
Uj)
^ = .
4.
Am, + +a~" x
n IV
\IV (n
.; X - a* 2.
a)
tl/l
n (na) (n A ^+" ^ i AX 2. 3.
.
IV VI
\iv
v
2a) iUI L
3 3
AX + &c
A
viz.
cfydx*
=a
[log {a
+ a A) \ y
What complementary
member ?
5.
function
must be added
to the second
Shew
that Sir
m
is
=/(l) +/(1
+ A)
t+f{\
+ A) 0*.
A+
8
&c,
a consequence of the
6.
Theorem
of Art. 11.
If
x=
prove that
(d\ n
A(T
A'
w
2
d2
d3
7.
An
n+1
= ^( n+1 A n
)
n
.
EX.
8.]
EXERCISES.
27
Au
term will be
An(f+q AP P x A
9.
d**quXAI
?
2
dxp df
viz.
3
ax
a.x*
2
2
2.3
a
2.3.4
3
= ex (a + xAa + | A
the symbol
^A
+ &c.
Aan = a n+l an
.
What class of series would the above theorem enable 10. us to convert from a slow to a rapid convergence?
11.
Shew by
in ascending powers of
may
be ex-
(e 0) ,
(e 02)
^_ +(6i03)f:|_
four terms of the expansion.
And
first
28
CHAPTER
III.
OF INTERPOLATION.
The word interpolate has been adopted in analysis to 1. denote primarily the interposing of missing terms in a series of quantities supposed subject to a determinate law of magnitude, but secondarily and more generally to denote the calculating, under some hypothesis of law or continuity, of any term of a series from the values of any other terms supposed
given.
As no series of 'particular values can determine a law, the problem of interpolation is an indeterminate one. To find an analytical expression of a function from a limited number
of its numerical values corresponding to given values of its independent variable x is, in Analysis, what in Geometry it would be to draw a continuous curve through a number of given points. And as in the latter case the number of possible curves, so in the former the number of analytical expressions satisfying the given conditions, is infinite. Thus the form of the function the species of the curve must be assumed a priori. It may be that the evident character of succession in the values observed indicates what kind of assumption is best. If for instance these values are of a periodical character, circular functions ought to be employed. But where no such indications exist it is customary to assume for the general expression of the values under consideration a rational and integral function of x, and to determine the coefficients by the given conditions.
This assumption rests upon the supposition (a supposition however actually verified in the case of all tabulated functions) that the successive orders of differences rapidly dimi-
In the case of a rational and integral function of the nth degree it has been seen that differences of the n
nish.
of
+ 1 th
ART.
1.]
OF INTERPOLATION.
29
all succeeding orders vanish. Hence if in any other function such differences become very small, that function may, quite irrespectively of its form, be approximately represented by a function which is rational and integral. Of course it is supposed that the value of x for which that of the function is required is not very remote from those, or from some of those, values for which the values of the function are given. The same assumption as to the form of the unknown function and the same condition of limitation as to the use of that form flow in an equally obvious manner from the expansion in Taylor's theorem.
and of
2. The problem of interpolation assumes different forms, according as the values given are equidistant, i. e. correspondent to equidifferent values of the independent variable, or not. But the solution of all its cases rests upon the same principle. The most obvious mode in which that principle can be applied is the following. If for n values a, b,... of an independent variable x the corresponding values ua ub ,... of an unknown function of x represented by u x are given, then, assuming as the approximate general expression of ux
,
,
ux =A+Bx+Cx\..+Ex
n~1
(1),
a form which is rational and integral and involves coefficients, the data in succession give
n arbitrary
u= A + Ba + u> = A + Bb +
Ca\
+ Ean~\ n Cb\.. + Eb ~\
.
a system of n linear equations which determine A, B...E. To avoid the solving of these equations other but equivalent modes of procedure are employed, all such being in effect reducible to the two following, viz. either to an application of that property of the rational and integral function in the second member of (1) which is expressed by the equation A nux =0, or to the substitution of a different but equivalent form for the rational and integral function. These methods will be respectively illustrated in Prop. 1 and its deductions, and in Prop. 2, of the following sections.
Prop. 1. Given n consecutive equidistant values uQ u^.. u n _ of a function u x to find its approximate general expres,
sion.
30
OF INTERPOLATION.
[CH.
III.
By
Chap.
II.
Art. 10,
.
m(m
and x
1
j
l)
A9
Hence, substituting
for x,
for
m,
we have
ad
inf.
ux =u
+ xAu
-{
nr [rp
A m + &c.
2
x.
Z
is
Anwa=0,
ux = u Q + xAu
^~ A\
2)
. .
x(x-l)...(x-n +
1.2...(n-l)
(2j>
It will be observed that the second the expression required. member is really a rational and integral function of x of the degree n 1, while the coefficients are made determinate by the data.
In applying this theorem the value of x may be conceived to express the distance of the term sought from the first term in the series, the common distance of the terms
given being taken as unity.
Ex.
3*16
= -4996871,
= '4969296, = '5010593
we have
the following
numbers and
Uo
4969296
differences
%
4996871
13722
%
5010593
4983106
13765
A A A
up
to
13810
2
3
-45
2
first
-43
The
its
differences
A\.
Now
the
common
ART.
3.]
OF INTERPOLATION.
31
being taken as unity, the value of x which corresponds to Hence we have 3*14159 will be '159.
u = 4969296
+ '159
x 13810
;i
- 45
(159)
('159-1) (-159-2)
1.2.3
we
find
^='4971495, which
is
the first difference only been employed, which is equivalent to the ordinary rule of proportional parts, there would have been an error of 4 in the last decimal, and second differences would have reduced
true to the last place of decimals.
this error to 1.
Had
the values given and that sought constitute a whatever may be the position of the value sought in that series, it is better to proceed as
3.
When
follows.
Let u ,u x ,u2 ... u n be the series. w the principle of the method, A w = Art. 10,
fi (ft
0,
^""Vi-r
\ 2
un_2
...
+ (-l) n uQ =
(3),
ux
...
un
may be found
we have
ua
-2u + u, = 0;
l
.:
u,=^^
(4).
is
To supply
we have
uQ
...
kw
6w 2
i
4^ 3
+ u^ =
*)
*( u
+ u*)-( uo+ u
5 ).
3-
OF INTERPOLATION.
r oo
[CH.
III.
Ex.
Eepresenting as
is
usual
e~
n ~l
d6 by T
(n),
it
is
by
finding approxi-
mately log
(-
logT(w),
*
\ogT(n).
*
i 12
4
l
74556
>
18432>
'55938,
'13165,
-42796,
-08828,
l
-i
-32788,
JO
l
-05261.
-i
Wj
Let the w2 ug
. . .
series
,
of values of log T(w) be represented by Then prothe value sought being that of w5
.
ceeding as before,
we
find
8.7
or,
8.7.6
ui
+w -8
9
K+m
8)
28 (3 + w7)
- 56
(w 4
+ w + 70w =
fl
whence
5
= 56
(u
+ *Q - 28
K+m
7)
+8
K + tQ -
(m,
^9 )
70
'
uaf &c,
their values
we
find
log
rg) = -24853,
To shew the gradual closing of the approximation number of the values given is increased, the following
are
as the
results
added
ART.
4.]
OF INTERPOLATION.
Data.
Calculated value of w .
a
33
w4
uG
6 ,w 7
-25610,
'24820,
%,%
w w 8 w4
\, u v usi
-24865,
-24853.
4. By an extension of the same method, we may treat any case in which the terms given and sought are terms, but not Thus, if u t uA u5 were given consecutive terms, of a series. = 0, 3w 2 = would give and u3 sought, the equations
,
,
A\
8
u4 3w 3 u5
- 3w
+ 3w ux = 0, + 3m3 - % = 0,
,
we have
(7),
3^-8^ + 6^-^ =
and hence u3 can be found. But it is better to apply the general method of the following Proposition.
at
once
Prop. 2. Given n values of a function which are not consecutive and equidistant, to find any other value whose place is given.
u n be the given values, corresponding to respectively as values of x, and let it be required to determine an approximate general expression for u x .
, , . . .
Let u a u b u e
a, b, c ...n
We
Art.
1.
shall
assume
this
expression rational
and
integral,
there being n conditions to be satisfied, viz. that for ... x = n, it shall assume the respective values u a u b ... u n the expression must contain n constants, whose values those conditions determine.
Now
x == a, x = b
,
'x
(8),
C by
a,
. . .
34
OF INTERPOLATION.
[CH.
III.
The substitution of another but equivalent form for (8) enables us to dispense with the solution of the linear system.
Let
ux
(9)
to n terms, each of the n terms in the right-hand member wanting one of the factors x a, x b,...(x n), and each being affected with an arbitrary constant. The assumption is legitimate, for the expression thus formed is, like that in (8), rational and integral, and it contains n arbitrary
coefficients.
+ &c.
Making x a, we have
ua
therefore
=A
(a
b)
(a
c)
...
(a
n)
(a
b)
(a
c)
b,
...
(a
n)
'
we have
B=
w\ d so on.
^
(l>-a)(b-c)...(b-n)'
finally,
. . .
Hence
(x
{x b) = u\ a -b)
c)
(x
n)
(a-c)
...
(a-n)
+U
(x
a)(xc)...(x n)
...
(b-a) (b-c)
(b-n)'"
,
.
...
+u n
(x
-)
(n
(10),
Lagrange's theorem
for
As the problem of interpolation, under the assumption that the function to be determined is rational and integral, is a determinate one, the different methods of solution above All these methods exemplified lead to consistent results. are implicitly contained in that of Lagrange.
The following are particular applications of Lagrange's theorem.
ART.
5.]
OF INTERPOLATION.
35
Given any number of values of a magnitude as ob5. served at given times to determine approximately the values of the successive differential coefficients of that magnitude at another given time.
;
Let , b, ... n be the times of observation, uai ubi ... un the observed values, x the time for which the value is required, and ux that value. Then the value of u x is given by (10), and the differential coefficients can thence be deduced in the But it is most convenient to assume the time usual way. represented above by x as the epoch, and to regard a, b, ... n as measured from that epoch, being negative if measured
backward.
The
du
values of
cL
,
-y-|
member
&c. in the development of the second by 1, 1.2, 1.2.3, &c. successively. Their general expressions may thus at once be found. Thus
coefficients of x,
,
of (10) divided
in particular
we
shall
have
dux j* ax
,2
/i be ...n( T \b
...++n, c
(11),
du* -
be
...
~aW~
I +
n( r
^ hc
+U T~j + j+&c. cd
1.2(a-b)(a-c)...(a-n)
upon
is
founded
6. The values of a quantity, e. g. the altitude of a star at given times, are found by observation. Required at what intermediate time the quantity had another given value.
.
usual to consider the time as the independent problem it is most convenient to consider the observed magnitude as such, and the time as a function of that magnitude. Let then a,b,c,... be the values given by observation, u a uln wc) ... the corresponding times, x the value for which the time is sought, and ux that time. Then the value of u x is given at once bv Lagrange's theorem
it is
Though
(10).
32
36
OF INTERPOLATION.
[CH.
III.
solved by regarding the time Bepresenting then, as in the last example, the times given by a, b, n, the time sought by x, and the corresponding values of the observed magnitude by u a u b ... u n and ux we must by the solution of the same equation (10) determine x.
as the independent variable.
. . . , , , ,
The above forms of solution being derived from different hypotheses, will of course differ. say derived from different hypotheses, because whichsoever element is regarded as dependent is treated not simply as a function, but as a and thus rational and integral function of the other element the choice affects the nature of the connexion. Except for the avoidance of difficulties of solution, the hypothesis which assumes the time as the independent variable is to be pre-
We
ferred.
Ex.
mum
or
or
minimum being
taken, to find
its
minimum.
Let a, b, c, represent the times of observation, and ux the magnitude of the quantity at any time x. Then ua) u b and u c are given, and, by Lagrange's formula,
_M u* = -
(x
(a
b) (x c) U _ h) (a - c) +
(x
'
c)
(x
a)
(b
-c){b-a)
(x a)(x b) + K {c -a)(c-b)>
and this function of x is to be a maximum or minimum. its differential coefficient with respect Hence equating to
to x,
we
find
2{{b-c) u a + (c-
a) u,
{a
-b) u e ]
This formula enables us to approximate to the meridian sun or of a star when a true meridian observation cannot be taken.
altitude of the
Areas of Curves.
Formulae of interpolation may be applied to the ap7. proximate evaluation of integrals between given limits, and
therefore to the determination of the areas of curves, the con-
ART.
7.]
OF INTERPOLATION.
37
The application is convenient, as it does tents of solids, &c. not require the form of the function under the sign of integration to be known.
Prop. The area of a curve being divided into n portions bounded by n + 1 equidistant ordinates u u ,...un whose values, together with their common distance, are given, an
,
approximate expression
The
if
the
common
general expression for an ordinate being uxy we have, distance of the ordinates be assumed as the
I
uxdx.
Now, by
ux
(2),
A = u + xAu +
x(x
. 1) )
^
x
Hence
I
ux dx
=u
dx + Au
xdx +
(x
1) dx
-
+
and
n
Au
3
rT^j
n
2
n* a
I*?
A\
fn*
2\
A\
(?- +
8 .
l
-")t^
35
4
50
, .
A\
A
/'
,
15"
225ra*
274re
AV
6
(14).
+ &c
.
It will be observed that the data permit us to calculate the successive differences of uo up to A nu Hence, on the assumption that all succeeding differences may be neglected,
38
OF INTERPOLATION.
[CH.
III.
the above theorem gives an approximate value of the integral sought. The following are particular deductions.
1st.
involving;
all
-2
I
uxdx
,
= 2u + 2Au + i A\.
But Au
=u u
l
AV =
u2
2u + u
x
whence, substi-
\
J
M = S^Jh
by
If the
Ji,
J\*-*Sa
2ndly.
is h,
W).
common
distance
we
find in like
manner
*<* + .Jkj
3rdly.
like
*u
** + V*
(16).
we have
in
manner
jy^^K+u
4thly. portions
t)
+ u(.u1+ u + u^
e)
(17) _
supposition that the area is divided into six 7 equidistant ordinates leads to a remarkable result, first given by the late Mr Weddle (Math. Journal, Vol. ix. p. 79), and deserves to be considered in
The
bounded by
detail.
unity,
Supposing the common distance of the ordinates to be we find, on making n = 7 in (14) and calculating the
coefficients,
ART.
7.]
OF INTERPOLATION.
39
uxdx
:
= 6?/ + 18 Aw +
27
A\ + 24A\ + 123 A\
<
Now
differs
small fraction
tion we must suppose sixth differences small, since all succeeding differences are to be neglected, we shall commit but
a slight error
this,
if
we change
3 Aw
6
Doing
find,
and so
on,
we
on
reduction,
r
n
u x dx =
10
{u
+ u + w + w + 5 [u + u + 6m
.
5)
8 },
common
u x dx
/.
=
v
{m
tt
v4
+ v + 5 {u + uB + 6u
G x
)
3]
. . .
(19),
AV
when
140
fifth
it
errs in excess
by only
AV when
The
lowing
to find
the fol-
area being divided into six portions by seven equidistant ordinates, add into one sum the even ordinates 5 times the odd ordinates and the middle ordinate, and multiply the result
The proposed
by
of the
common
40
OF INTERPOLATION.
[CH.
III.
The two radii which form a diameter of a circle are and perpendicular ordinates are raised at the points Required the area of that portion of the circle of bisection. which is included between the two ordinates, the diameter, and the curve, the radius being supposed equal to unity.
Ex.
1.
bisected,
The
V3
~2~'
V8
3
'
V35 ~6~'
'
V35 V8 ~6~' 3
V3
'
2
is
'
The
area
which, on
-+
o
will be found to be
The
In
rule
for
equidistant ordinates
commonly employed
differences
Should the function to be integrated become infinite at or limits, an appropriate transformation will be
Ex.
2.
2
/
log
Odd.
infinite at the
lower limit,
= 6 log
sin 6
-f0 cot
6d0,
[ Jo
limits,
=-
f * 6 cot Odd.
Jo
The
now
calculated for
ART.
8.]
OF INTERPOLATION.
41
= 0,
we
= --, 6=
find
7T
27T
,
7T
0=17?
an(i
The
(^ 6 cot Odd
J
=- -69311.
is
known
to
be
-69316.
8.
employment of differences, and to calculate directly the If we coefficient of um in the general expression for judx.
diate
represent
,
the
,
equidistant
ordinates,
2n-\- 1
in
number, by
integrations
by
as-
= A un + A
Q
(u n+1
+ unJ + A
(-
(u n+2
+u _ )...+ A n (un +u
7l
),
where generally
A m 1.2.. .(n
'
ir
1.2.
+ r)
,.{n-r)
p
A
(o
-/)(i'-Y)...K-2/*)
s_ f
may
is
dv
()
even.
Application
to
Statistics.
9. When the results of statistical observations are presented in a tabular form it is sometimes required to narrow the intervals to which they correspond, or to fill up some particular hiatus by the interpolation of intermediate values. In applying to this purpose the methods of the foregoing sections, it is not to be forgotten that the assumptions which they involve render our conclusions the less trustworthy in proportion as the matter of inquiry is less under the dominion of any known laws, and that this is still more the case in proportion as the field of observation is too narrow to exhibit
42
fairly the
OF INTERPOLATION.
[CH.
III.
operation of the unknown laws which do exist. anomalies, for instance, which we meet with in the attempt to estimate the law of human mortality seem rather to be due to the imperfection of our data than to want of continuity in the law itself. The following is an example of the anomalies in question.
The
Ex. The expectation of life at a particular age being defined as the average duration of life after that age, it is required from the following data, derived from the Carlisle tables of mortality, to estimate the probable expectation of life at 50 years, and in particular to shew how that estimate is affected by the number of the data taken into account.
Age.
Expectation.
Age.
Expectation.
10 20
48-82
=w,
60
70 80 90
14-34
41-46
34*34
=u
=u
30 40
=u =w 5*51 = u
9-18
G
7
27'Ql=u4
.
3*28
=u
The expectation of life at 50 would, according to the above scheme, be represented by u 5 Now if we take as our only data the expectation of life at 40 and 60, we find by the
method of Art.
3,
5 =
If
find
"5
M4
4P = 20-97
()
we add
30 and 70,
we
(J).
If
we add
us =
we
find
..(c).
K + O - Jq K + O + ^ K+<) =20-75.
,
And
90,
if we add we have
and
W6=YqK+ M
M W 6)-Jq( 3+ t)
ART.
9.]
OF INTERPOLATION.
43
notice that the second of the above results is considerably lower than the first, but that the second, third, and fourth exhibit a gradual approximation toward some value not very remote from 20*8.
We
Nevertheless the actual expectation at 50 as given in the which is greater than even the first result or the average between the expectations at 40 and 60. may almost certainly conclude from this that the Carlisle table errs in excess for the age of 50.
Carlisle tables is 21*11,
We
is so.
a comparison with some recent tables shews that this the tables of the Registrar-General, Mr Neison* deduced the following results.
And
From
Ago.
Expectation.
Age.
Expectation.
10
47*7564
40-6910 34-0990
27-4760
20*8463
60 70
14-5854
9-2176
20 30
80
90
5-2160
2-8930
40
50
at 50, corre-
given in
20*8215,
(),
(5),
(c),
(d), will
be found
21-0307,
20*8464,
21*2073.
is less
We
the
than
those at 40 and 60. see also that the second result gives a close, and the third a very close, approximation to its value. The deviation in the fourth result, which takes account of the extreme ages of 10 and 90, seems due to the attempt to comprehend under the same law the mortality of childhood and of extreme old age.
mean between
We
When in an extended table of numerical results the differences tend first to diminish and afterwards to increase, and I think I have observed such a disposition in tables of mortality, it may be concluded that the extreme portions of the tables are subject to different laws. And even should those laws admit, as perhaps they always do, of comprehension
* Contributions to Vital Statistics, p. 8.
44
OF INTERPOLATION.
[CH.
III.
under some law higher and more general, it may be inferred that that law is incapable of approximate expression in the particular form (Art. 2) which our methods of interpolation
presuppose.
EXERCISES.
1.
following data
2. 3.
Find a
1, 2, 3,
x=
shall
and integral function of x which, when assume the respective values 4, 6, 10.
45
CHAPTER
IV.
FINITE INTEGRATION.
here used to denote the process function of x, we determine some other function of which the given function expresses the
1.
The
term integration
is
Thus
to integrate
ux
is to
&vx = ux
The operation of integration is therefore by definition the inverse of the operation denoted by the symbol A. As such it may with perfect propriety be denoted by the inverse form A" 1 It is usual however to employ for this purpose a distinct symbol, 2, the origin of which, as well as of the term inte.
gration by explain.
which
its
office
is
denoted,
it
will be proper to
One of the most important applications of the Calculus of Finite Differences is to the finite summation of series.
Now let u u v u2 &c. represent successive terms of a whose general term is u x and let
,
, ,
series
vx
(1).
initial term,
Vi =m+V+'"+ mm+\
Hence, subtracting
(1)
(2).
from
,
(2)
Avx =u x
.*.
^=A"V
A-1
,
It appears from the last equation that applied to ux , expresses the sum of that portion of a series whose general term is ux which begins with any fixed term u a and ends On this account -1 has been usually replaced by with ux_ x
,
.
46
FINITE INTEGRATION.
[CH. IV.
gration.
the symbol 2, considered as indicating a summation or inteAt the same time the properties of the symbol 2, and the mode of performing the operation which it denotes, or, to speak with greater strictness, of answering that question of which it is virtually an expression, are best deduced, and are usually deduced, from its definition as the inverse of the
as defined
by the equation
(3),
may
as defined
by
the equation
(4),
St^A-X
and as having for its object the discovery of some finite expression of which the right-hand member of (3) constitutes the development, it is interrogative rather than directive (Differential Equations, p. 377); it sets before us an object of inquiry, but does not prescribe any mode of attaining that object. And in fact our knowledge of the cases in .which %ux admits of finite expression rests ultimately upon an examination of the known results of the operation A, and a more or less direct reference of the form of the function ux to such
results.
one conclusion flows immediately from such reference A. It is that, whatever may be the form of ux the complete expression for %u x must contain an arbitrary constant, or, to speak more strictly, an arbitrary quantity which does not change when x changes to x+1, and which is therefore constant relatively to the kind of change denoted by the symbol A. For let vx be a particular form of %um and let vx + w x be any other form, this assumption being sufficiently general to include all possible forms if wx is regarded as an arbitrary function of x, then
to the effect of the operation
,
Now
Avx =ux
Hence
A(vx +w x )=ux
Aw =0
And
this
(5). is
wx
a function of
unity.
x which
when x
is
increased
by
ART.
1.]
FINITE INTEGRATION.
47
Since w x is thus constant relatively to A, we may with propriety represent it by c, and term it a periodical constant. With this convention we are permitted to say If v x be a particular value of Xux the complete value will be v x +c.
; ,
The necessity of a periodical constant c to complete the value of Xux may also be established, and its analytical expression determined, by transforming the problem of summation into that of the solution of a differential equation.
Let
%ux = y, then y
d
is
solely conditioned
for
by
the equation
ty = u x,
equation
or,
putting
lx
(*-!)* = .
(6)-
Now, by the theory of linear differential equations, the complete value of y will be obtained by adding to any particular value vx the complete value of what y would be, were ux equal to 0. Hence
Xux = v x + Cx em? + C
m*x
+ &c
(7),
2
,
Gv C
mv w
m
&c. the
dif-
-l
(8).
= 2i7r\J 1 these roots are included in the form i being When t we have 0, or a positive integer. and the corresponding term in (7) reduces to a constant. But when i is a positive integer, we have in the second member of (7) a pair of terms of the form
Now
all
which, on
making
C+G = A
f
i,
(C C)
\J
=B
it
is
reall
sm2i'7T.
Hence, giving to i
2,u x
vx +B
x
-\-
C-\-
cos 2ttx
2
+A
cos iirx
3
+A
cos ttx
+ &c.
(9)
sin 2irx
+B
sin 4,7rx
+B
sin Qttx -f
&c
portion of the right-hand member of this equation is the general analytical expression of a periodical constant as above defined, viz. as ever resuming the same value for values of x, whether integral or fractional, which
The
which follows vx
48
differ
FINITE INTEGRATION.
[CH. IV.
to
by
unity.
It
with only a particular set of values of x progressing by unity, and not with all possible sets, the periodical constant merges into an ordinary, i.e. into an absolute constant. Thus, if x be exclusively integral, (9)
do, as indeed usually happens,
becomes
= vx + c,
c
<j) (cos 2ttx, sin27rx). But this notanot only inaccurate, but very likely to mislead. It seems better either to employ G, leaving the interpretation to the general knowledge of the student, or to adopt the cor-
tion
is
rect
form
sin 2iirx)
(10)
We
shall usually
do the former.
Integrable Forms.
2. Of integrable forms reducible under any general law, the following are the most important.
1st
Form.
1).
x(xl)
...
(x
m-\-
II.
Art.
;
2,
we have
Ax
therefore
{mM)
={m + l)xm)
Xx {m) =
f
^
i
m+\ +
'-
G,
<*
or
,\ tx(x-l)...(x-m +
,
\ l)
Thus
also, if
ux = ax-hb, we have
( 2)
ART.
2.]
FINITE INTEGRATION.
49
For, by Chap. il. Art. 5, any such function is reducible to a series of factorials of the preceding form, each of which may be integrated separately.
Ex.
To
Chap.
determine
II.
$xm
By
Art 5
m xm = AO x
Therefore
~
2
>
x +
{2)
=- x ...+ x
(3)
{m)
.
-,
__
3
(3)
In the same
way we
find for
%ux
Xux =
C+u x + Aw
A\
j-y-jj
+ &c.
is
rational
and
integral.
There exists also another theorem which accomplishes the same end. We have
^_n =
(l
+ A)"X
n
= ux - ntMx + ^
Let n = x, then
A ^ - &c.
2
u
Therefore
^a;
= ^ - icA ux +
A ^ - &c.
2
Now this being true independently of the form of n x are permitted to change u x into Xux . If we do this, represent by the value of Xu we have
we
and
~ S. = (7+
m*
#(a?
-'
50
It is
FINITE INTEGRATION.
[CH. IV.
obvious that a rational and integral function of x by assuming for its integral a similar function of a degree higher by unity but with arbitrary coefficients whose values are to be determined by the condition that the difference of the assumed integral shall be equal to the function given.
may
also be integrated
3rd Form.
um^ xx+i
v
... '
'
u. x+m
where u x
is
of the form
II.
ax +
h.
Art.
2,
we have
corresponding to the
A
Hence
am
.
uxux+1
. . .
u x+m
am(u xuXl
r
. .
(4).
.uxvn ^)
It will be observed that there must be at least two factors No in the denominator of the expression to be integrated.
finite
for expression 1
ax
r exists. o
To
certain
are
re-
ducible.
Thus we can
integrate
any rational
form
^x+m
ux being of the form ax + b, and (x) a rational and integral function of x of a degree lower by at least two unities than
cf>
denominator.
For,
expressing
cf>
{x)
in
(x)
. . .
+Euxux+1
..u x+m _ r
cients, or
A,B... being constants to be determined by equating coeffiby an obvious extension of the theorem of Chap. II.
5,
Art.
we
find
ART.
2.]
FINITE INTEGRATION.
l -
51
?
SM=At
...
- + B2
+ Et 1111
(4).
(5),
Again, supposing the numerator of a rational fraction to be by at least two unities than the denominator, but intermediate factors alone to be wanting in the latter to give to it the factorial character above described, then, these factors being supplied to both numerator and denominator, the fraction may be integrated as in the last case.
Ex.
Thus ux
still
representing
ax + b, we should have
^
with the second
^
"x
XUX ^
w x+i -x+2 x+3
member
of which
we must
proceed as before.
As
is
all that is known of the integration of rational functions virtually continued in the two primary theorems of (2) and
it
(4),
is
Supposing then ux
= ax + b,
. . .
let
ux u x_ x
ux_ m+1
(ax
+ b) m\
(
={ax + by-m
then
3(+)-< + >;7 + o
whether m be positive or negative. with the theorem
6 ).
The analogy
of this result
f(ax +
is
bTdx= a{m l) +
{a *
^\ c
<j>(x) is
obvious.
4th Form.
rational
and
42
52
Since
FINITE INTEGRATION.
[CH. IV.
Aa
[a
1)
x
,
we have
a
1
(x) we may now employ either a method of <f> parts or a symbolical method founded upon the x relations between the exponential a and the symbol A.
To deduce Xa
by
integration
To
integrate
by
parts
we
have,
,
since
therefore
(7),
Applying
nx
(x)
this
we have
x+1
2*(*) a*= f
n -^-t^A^
{x)
= -
\(f>
(x)
- a^ax A(j>
(x
Thus the integration ofa cf>(x) is made to depend upon that x of a A<f) (x) ; this again will by the same method depend upon x 2 that of a A (j) (x), and so on. Hence <p(x) being by hypothesis rational and integral, the process may be continued until the This will happen after function under the sign 2 vanishes. n + l operations if <j> (x) be of the nth degree ; and the integral will be obtained in finite terms.
But the symbolical method above same result by a single operation.
referred to leads to the
By
known theorem
But ax etXo * a
into log a,
we have
ART.
2.]
FINITE INTEGRATION.
d_
53
Now %ax
cf>
(x)
(e**
1 )~V< (x)
therefore
by what precedes
Xa*<f> (x)
(x)
= az (ae^ -
l)-
1 <f>
{x)
-1
= ax {a (1 + A) ax
/
l}
(x), since e
Tx
=1+
A,
aA
V
1/
1
.
IV
This however
pleted get
by the
We
tax
<f>
(x)
+
The
of
<f>
(8).
series
(x),
within the brackets stops at the nth difference supposing <f> (x) of the ?iih degree.
5th Form.
+ b)
<j>
(x), sin
(ax
+ b)
<f>
(x),
where
<j>
(x) is rational
and
integral.
</>(x)
=l
and
rt
sin
a / - cos ax +
I
<2
7r\ +
-z
a A sin
(oas
7\ =2 sin + o)
o>
/
I
sin
aa?
+o
a+
-\
ir\
J
54
FINITE INTEGRATION.
[CH. IV.
we have
_
2,
cos [ax
+b
t
a
-\
2 sin
Hence changing
& into
Z>
/
(
we have
a+ ir
cos
aa?
+b
cos (*;
+ &) =
+<7
(9),
2smsm [ax +
tsin(ax
b
.
+ b) =
+ b)
(10).
2 sin-
The
values of
2 cos
may now
(ax
+ b)
<f>
(x),
sin (ax
<f>
(x),
expressing the trigonometrical functions in their exponential forms, applying the general theorem (8), and reducing the results to a rational form.
Thus
.
also
we might
mx
<j>(x),
integrate
(ax
+ b)
When a function proposed for Miscellaneous Forms. 6th. integration cannot be referred to any of the preceding forms, it will be proper to divine if possible the form of its integral from general knowledge of the effect of the operation A, and to determine the constants by comparing the difference of the conjectured integral with the function proposed.
Thus
since
ART.
2.]
FINITE INTEGRATION.
(x)
55
where
rational fraction
it is evident that if <j> (x) be a be such. Hence if we had to x integrate a function of the form a yjr(x), ty (x) being a rational fraction, it would be proper to try first the hypothesis that the x integral was of the form a <f> (x) <f> (x) being a rational fraction the constitution of which would be suggested by that of f(x).
yfr
<f>
= a<j> (x + 1)
yjr
(x)
Ex.
Thus
1)
it
may
.
x
X
2 (x
-.
+
1)
(x
, - '
if finite, will
as a factor.
2 And
proves x r
to
be
its
tan" cf> (x), &c, are of the </> (x) 1 respective forms sin" ty (x) tan" i/r (x) &c, ty(x) being an 1 algebraic function when cj)(x) is such, and, if tan" be employed, rational if <j>(x) be so, it is usually not difficult to
since
,
Thus
A sin"
1
conjecture
if finite
1
forms exist, of
2 sin"
yjr
1
-\/r
(x)
2 tan"
yfr(x),
&c.
(x)
being
still
it
supposed algebraic.
will be found that tan
-1
Ex.
Thus
p
tegrable in finite terms
(f
is satisfied.
+ qx
-t-
rx
is
in-
r = 4
2
(pr
1)
1
The
1
result is
1
Stan"
p + qx + rx
j=
2 G+ tan" r - 2rx
(11).
observations may be generalized. The operation does not change or annul the functional characteristics of the subject to which it is applied. It does not convert transcendental into algebraic functions, or one species of transcendental functions into another. And thus, in the inverse procedure of integration, the limits of conjecture are narrowed. In the above respect the operation A is unlike that of differentiation, which involves essentially a procedure to the limit, and in the limit new forms arise.
The above
denoted by
Ex.
Eequired 1
2
cos ax. cos a
[x+
tt*
1)
56
FINITE INTEGRATION.
[CH. IV.
The
integral
it
.
shews that
proper form x
P
must be trigonometrical, and its factorial form must involve cos ax as a denominator. The
~
V C.
is
tanaa?
sin
Summation of
Series.
3. The application of the Calculus of Finite Differences to the summation of series has been already referred to in Art. 1. From what is there said it appears that to determine the sum of any portion of a series the general expression of whose terms is known, we must integrate the term which follows that portion, and determine the arbitrary constant by observing with what term the portion begins.
Ex.
first
series 1
+2 +3
2
...
+ x we must
integrate [x
+
2
l)
2
.
Now 2 {x+l) = t{l+3x + x{x-l)} x(xSx (x-1) 1) A 1 +-*+-yTo determine G we observe that reduced to its first term 1. Hence
(7=0.
1
(x
1
- 2)
~
c.
is
when x =
the series
=1+
C,
and therefore
It is often more convenient to integrate the last term of that portion whose sum is required, thus obtaining the sum of the preceding terms, and then to add the last term to the result.
Thus
l
in the above
2
2 2
+ 2 +3 ...+^ = 2^ + ^
=^ir
on determining the constant.
x(xl) J
x(xl)(x 2)
r
is
+x
It must always be carefully noted what pendent variable. Thus if the series be
x a + xa+1 + x a+ \.. + xn
Accordingly, the true variable is n. with respect to n, we have
2W
denoting integration
xa
-}-
x u+1
...
+ xn = X nxn+1 = xt nxn
x-l +
0.
ART.
3.]
FINITE INTEGRATION.
57
the series re-
To
duces
determine
its first
G we
observe that
when n a
term.
Hence
x
I
a+l
.'.
C = xa
x-l
X-V
n+1
and we have
a
'
x-l
the
known
expression.
Ex.
To sum
the series
1*
-2 +3 -4
2 2
2
. . .
+ x
Here
Reducing by
(8)
Zux+1 = 2 (Art.
2,
l)*{x
l)
2
.
we have
finally
Stt* +1
=(-1)
Ex.
To sum
the series
1
. . .
'*
+ x(x + 2)'
Bw ^- S (+l)(a +8)
=s
.+ 2 (a?+l)(aj + 2)(a> + 3)
(x+l){x + 2)
(x
+ S)
1
(05-f 2)
(*
+ 3)
= CWhen
And
05
2(a5+l)
(a?
+ 2)
05+2
is
1 the tie
sum
of the series
therefore
C= -
on reducing
305
+505
4(05+1) (05+2)
58
4.
FINITE INTEGRATION.
[CH. IV.
When a series proceeds by powers of some quantity x, frequently happens that its finite value can be obtained for some particular value of x, but not for any other value. The following is an illustration.
it
Ex.
To sum, when
\\x
2.3
+
,
2\x*
3.4
3\x
4.5
p +&c.
to
n terms
(a),
The n ih
term, represented
by uni being
(n
1) {n
'
2)
we have
^ ZUn+1==
71
{n
l){n
+ 2)
Z + ^.
71
X
2)
"(w+l)(w +
.-.^
Now
assume
to w,
(n
"
we have
(n
+ l)(n + 2)
x\x
\
a (n+
1)
+b
a7i
n+
m
+b
(a
~ Xn a(x-\)7?+(2a+b)(x-\)n+(a+b)x-2b (w + l)(w+2)
'
a(x-l) =
Whence we
l,
(2a
+ b)(x-l) = 0,
1
7
+ b)x-2b = 0.
2
find
if
x = 4, and
(w+
Substituting in
there results
1) (w
+ 2)
(b),
ART.
5.]
FINITE INTEGRATION.
59
.
114 2T3
+ 374
2\_
"*
(n
"' C ''
Connexion of methods.
preceding article admit also of treatment developed in the treatise on Differential Equations (p. 435), and it is very interesting to compare the modes in which the same conditions of finite
5.
The
of the
method
distinct in form.
tration.
algebraic expression present themselves in solutions so totally The last example will serve as an illus-
If
we make x =
l\e e
2\e29
+
(ra
n\end
+ l)(ra + 2)
+l %D + ,)
J-^,
(n+l)0
(l+(i>
11
tf
-l
by means
of the theorem
we have
for the
sum
1 i +-J
+x fx ^.n+1 r*a?
xj
x1
M -x dx
,
A 4
r t
Now let x = 4. Then substituting 4 for x in the terms without the sign of integration and in the upper limit, we
have
n lf'x^-Xj l[ 4 -l dx--\ 45 +7 3 x1
4J
xn+*-x* dx x1 4J
2
60
4n
FINITE INTEGEATTON.
[CH. IV.
-l
3
+-
r'x^-x-x^+x*
!
4./
\
1
dx
,
-1
4U^r
iJ>
now be
1
1 /
(*fx
-x
xn+2 -xn+1 \
x-=i~)
dx
= 4- +
The
result
4"
integration can
effected,
4 V
+ 2/
+2
3'
as before.
But what
methods ?
is
It consists in the law of reciprocity established by the theorem of Chap. n. Art. 11, viz.
and yjr(t) being functions developable by Maclaurin's cf)(t) theorem, and t being made equal to after the implied operations are performed.
To
(a)
(a
+ 1) +
<f)
(a
+ 2)
. . .
<j>
(n) .
is
2<(rc
+ l)-2<Ma),
first
and
to
a in
is
,dtJ
\dtJ
e'-l
t
and therefore
= 0.
ART.
6.]
FINITE INTEGRATION.
61
Now
as
= 0,
Hi
d
=
since
(c*- 1)
_1
<
(*+n+ 1) - (--l)"
0(+a),
is connected with w in the one term and with a in the other by the sign of addition. And now, making t = 0, the expression reduces to
S^(n + 1)-S^(a),
which agrees with the previous expression.
to
inverse forms.
From
l)"
1
,
of
in the forms
d_
,
and more generally of 2 in the form (edx l)~ n flow certain theorems which may be regarded as extensions To comprehend the trueof some of the results of Chap. II.
dx (e
Any of the expression (e l)~ u x must be borne in mind. legitimate transformation of this expression by the development of the symbolical factor must be considered, in so far as it consists of direct forms, to be an answer to the question
which that expression proposes; in so far as it consists of inverse forms to be a replacing of that question by others. But the answers will not be of necessity sufficiently general, and the substituted questions if answered in a perfectly unrestricted manner may lead to results which are too general. In the one case we must introduce arbitrary constants, in the other case we must determine the connecting relations among arbitrary constants in both cases falling back upon our prior knowledge of what the character of the true solution must be. Two examples will suffice for illustration.
;
d dx
Ex.
1.
Let us endeavour
to
pression for
Xux
given in
(3),
Art.
62
FINITE INTEGRATION.
[CH. IV.
Now
to
Now this is a = co in
(3).
only a particular form of Xux corresponding To deduce the general form we must
if to
that constant
we
assign
we
%uxvx
in a series
We have
Xuxvx = {DD' l)"
1
u x vx
referring to
.
x only
as entering into ux
D'
to
x only
as
entering into vx
Now
Z)'
= A' + 1, =
(DA'
therefore
In applying this theorem, we are not permitted to introduce unconnected arbitrary constants into its successive terms. If we perform on both sides the operation A, we shall find that n the equation will be identically satisfied provided A% ux in any n~ term, and to in the preceding this term is equal imposes ^ ux 1 be retained withthe condition that the constants in S"" n out change in % u x And as, if this be done, the equation will be satisfied, it follows that however many those constants may Hence then we be, they will effectively be reduced to one. may infer that if we express the theorem in the form
l
+A ux_
2
% vx
(1),
ART.
6.]
FINITE INTEGRATION.
63
we
shall be permitted to neglect the constants of integration, n provided that we always deduce % vx by direct integration of n~l the value of % vx in the preceding term.
If ux be rational and integral, the series will be finite, and the constant C will be the one which is due to the last integration effected.
EXERCISES.
1.
Find by
integration the
2
sum
of the series
2
.
+2 +3
...+ 10
2.
Sum
5
the series
2.3.4.5
3.
+ ^
8
z
3'
2
3.4.5.6
~ + &c. + 4.5.6 .7
11
~.
t.
to
terms.
Sum
2
the series
2 .4
2
^3 + -3^ + 475 +
l
3 3
3
.4
.4
4 .4
5-.T
P + &C
t0
^ termS
4.
Prove that
2
.
5.
...
+ ax
~l
cos (x
1) 0.
successive orders of ngurate numbers are defined the xth term of any order is equal to the sum of the first x terms of the order next preceding, while the terms of the first order are each equal to unity. Hence, shew that the x th term of the nth order is
6.
by
x(x+l)
(a?
+ 2)
...
...
(x
+ n 2)
*
1.2
(w-1)
7. It is always possible to assign such values of 5, real or imaginary, (being the roots of an equation of the nth degree) that the function
{a
+ ffx + yx
...
+vxn )sx
64
shall
EXERCISES.
[CH. IV.
stants,
be integrable in finite terms, a, fi, ... v being any conand ux of the form ax -\- b. (Herschel's Examples of
Sum
1
the series
1
7tt\
sin 6
9.
+ sin
20
to
terms.
(different
from
%ux v x
viz.
%*ux vx
+
11.
rcjrc-l)
AX2"+2V2 - &c.,
a^ inf
Xux = (7+ ^v
"~
2
,
A ^ xi
+
12.
x{x-\)
.
(x
2)
2.3
^^
,
.
Expand
%<f)
(x)cos2mx in a
<j>(x).
65
CHAPTER
V.
SERIES.
series is said to be convergent or divergent accord1. ing as the sum of its first n terms approaches or does not approach to a finite limit when n is indefinitely increased.
This definition leads us to distinguish between the convergency of a series and the convergency of the terms of a The successive terms of the series series.
converge to the limit 0, but it will be shewn that the of n of those terms tends to become infinite with n.
sum
On
+ +I+ + 2 8
16
9 +&C
is
its
sum
terms.
Two cases present themselves. 1st. That in which 2. the terms of a series are all of the same or are ultimately all of the same sign. 2ndly. That in which they are, or ultimately become, alternately positive and negative. The latter jicase we propose, on account of the greater simplicity of its theory, to consider first.
Prop.
I.
and
is
are, or
series whose terms diminish in absolute value, end with becoming, alternately positive and negative,
convergent.
Let u x
B. F. D.
w +u u +
2 z
4
its
63
SERIES.
[CH. V.
terminal portion, the part which it follows being in the latter Then writing it in the successive forms case supposed finite.
^2 +
{u 2
3
*0
+ (5 --0 + &c
4
5
(*)>
-u )-(u -u )-&c.
1
(2),
u2
we
and
less
than u x
Ex.
Thus
the series
,
-2 + 3
+ - &c. ad
5
inf. J
is less
than
theory of the convergency and divergency of series whose terms are ultimately of one sign and at the same time converge to the limit 0, will occupy the remainder of this chapter and will be developed in the following order. 1st. fundamental proposition due to Cauchy which makes the test of convergency to consist in a process of integration, will be 2ndly. Certain direct consequences of that proestablished. position relating to particular classes of series, including the 3rdly. geometrical, will be deduced. Upon those consequences, and upon a certain extension of the algebraical theory of degree which has been developed in the writings of Professor De Morgan and of M. Bertrand, a system of criteria It may be added general in application will be founded. that the first and most important of the criteria in question, to which indeed the others are properly supplemental, bein^ founded upon the known properties of geometrical series, might be proved without the aid of Cauchy' s proposition but for the sake of unity it has been thought proper to exhibit the different parts of the system in their natural rela-
The
tion.
Fundamental Proposition.
Prop. If the function <j)(x) be positive in sign but 4. diminishing in value as x varies continuously from a to go
then the series
<f>
(a) -f
(/>
{a
1)
<p
(a
-f
2)
+ &c. ad inf.,
(3).
ART.
4.]
SERIES.
67
is finite
<f>(x)
dx
J a
or infinite.
For
since
<
(x)
diminislies from
}
again from
a+ltox = a + 2
ra+l
I
x = a to x = a + &c, we have
,
l,
and
<f>
(x)
dx <
<f>
(a)
J a
ra+2
I
(j>(x)
dx<(j>(a+l),
together,
J a+i
inf.
we
<j)(x)
dx <${a) +
<j)(a
+ 1)
...
J a
<fi
(x)
dx
>
<j)
(a
+1
J a
cf>
(x)
dx >
(j>
(a
+ 2)
J a+l
and so on.
[
+ l) +
(x)
<(a
+ 2) ...+&c
(5).
J a
Thus the
<\>
J a
series
${a)
(f)(a
2)
...
2) ...
differ by <f> (a) will differ from the former series by a quantity less than cf> (a) therefore by a finite quantity. Thus the series and the integral are finite or infinite together.
which
Cor. If in the inequation (5) we change a into a 1, and compare the result with (4), it will appear that the series
<(a)+<(a
+ l) +
<(a
+ 2)
...
ad
inf.
52
68
has for
SERIES.
[CH. V.
inferior
/OO
and
superior limits
f
00
<j>(x)dx,
and
$(x)dx
(6).
J a
J a
The application of the above proposition will be sufficiently explained in the two following examples relating to geometrical series, and to the other classes of series involved in the demonstration of the final system of criteria referred to in Art. 3.
Ex.
1.
The
geometrical series
l
+h+h
<
1,
+ti
+ &e.
if
ad
inf.
is
convergent
if
divergent
>
1.
The
0,
general term
is If,
the value of
is
x in
the
first
term being
I
simply whether
h dx
Jo
is infinite.
Now
Jfdx =
hx -l
log h
If h
>1
series is
expression becomes infinite with x and the diverging. If h < 1 the expression assumes the finite
this
j
.
value
-,
The
log h
If h =. 1 the expression becomes indeterminate, but, prox ceeding in the usual way, assumes the limiting form xh which becomes infinite with x. Here then the series is
diverging.
Ex.
2.
The
successive series
1
(a+l) m
(a
_J_
+
'
(a
1
+ 2)'
+ &C.
+ &C.
(7),
i7
a
1
(log a)"
+ l){log(a+l)}
1
+ &C-
ART.
4.]
SERIES.
69
m>\,
and divergent
m ~ 1.
The determining
integrals are
rdx *dx
)
a
f
>
r
a
dx
m x(\ogx) mi
>
r f*
]a J
xm
dx " x\ogx(log\ogx) m,
is
and
when m
1
equal to
1
1,
are
^-^
1
(loga^-flogg)
'
-"
m
than
1.
in
which x
1,
go
less
integrals
if m be m=l the
log
x log a, log log x log log a, log log log a? log log log a &c.
still
and
become
infinite
with x.
if
Thus the
vergent
if in
>1
and divergent
m ^ 1.
Perhaps there is no other mode so satisfactory for 5. establishing the convergency or divergency of a series as the direct application of Cauchy's proposition, when the inteBut, as this is not gration which it involves is possible.
always the case, the construction of a system of derived rules not involving a process of integration becomes important. To this object we now proceed.
First derived Criterion.
Prop.
The
series
is
supposed positive,
ratio ~^1 tends,
u + u x + u 2 ...ad inf., all whose terms are convergent or divergent according as the
is indefinitely increased,
to
u Ux
when x
a limiting
value
less
Let h be that limiting value and first let h be less than 1, and let k be some positive quantity so small that h + k shall
also
be
less
than
1.
Then
u
as
-^
it is
70
SERIES.
[CH. V.
possible to give to
that value
x some value n
and
lie within the limits h + Jc and h 7c. Hence if, beginning with the particular value of x in question, we construct the
three series
+ 7c)u n +(h + kyun + &c. u n + un+1 + un+2 +&c. un + (h-7c)un +{h- hfun + &c.
u n +{h
>
'
(8),
each term after the first in the second series will be intermediate in value between the corresponding terms in the first and third series, and therefore the second series will be intermediate in value between
l-(h + k)
which are the
finite
and
i-(h-k)'
first
values of the
and third
series.
And
to
the other hand, if h be greater than unity, then, giving k some small positive value such that h k shall also exceed unity, it will be possible to give to x some value n so
large, yet finite, that for that
On
and
all
shall lie
between h
+k
and h k.
ux
Here then
still
each term
after the first in the second series will be intermediate between the corresponding terms of the first and third series. But h + h and h k being both greater than unity, both the latter series are divergent (Ex. 1). Hence the second or given series is divergent also.
Ex.
3.
The
series
l
is
+ +
<
+
f
:
f
.
Q
all
+&c,
derived
t.
convergent for
values of
For
if
t '~1.2...x
Ux+1
'
ART.
5.]
SERIES.
71
then
m= x+1
t
and
this tends to
4.
as
a?
tends to infinity.
Ex.
The
series
a(a
+ l)
a(a
+ l)(g + 2)
t
is
is less
or greater
(jb
+1
) (?>
<+ 2 )
... (J
+ a-1)
'
Therefore
t^tf +
a?
and
this tends, a? being indefinitely increased, to the limit t. Accordingly therefore as t is less or greater than unity, the series is convergent or divergent.
If
the rule
fails.
Nor would
it
be easy to apply
number
of factors involved in the expression of the general proceed, therefore, to establish the term of the series. supplemental criteria referred to in Art. 3.
We
Supplemental Criteria.
6.
Let the
series
under consideration be
ua
+ u a+l +
u a+2
+ ua+3
...admf.
(10),
the general term ux being supposed positive and diminishing in value from x = a to x = infinity. The above form is adopted as before to represent the terminal, and by hypothesis positive, portion of series whose terms do not necessarily begin with being positive; since it is upon the character of the terminal portion that the convergency or divergency of the series depends.
It is evident that the series (10) will be convergent if its terms become ultimately less than the corresponding terms of a known convergent series, and that it will be divergent
72
SERIES.
[CH. V.
if its terms become ultimately greater than the corresponding terms of a known divergent series.
series
2,
"v
is
ux with
.
in (7),
Ex.
^ X
Then
a condition of convergency
is
1
Hence we
find
xm <
log
m<
and since
logic
'
log
a?
On
is
divergency
1
if
>
Xjn
x being
than
1.
indefinitely increased,
and
being equal to or
less
But
this gives
log
logo;
and therefore
log
logic
ART.
6.]
73
according as,
9 u x
-.
than unity.
test failing, let
(7).
But the
made with the second of the series in For convergency, we then have as the limiting equation,
the comparison be
uv <
(log
xy
find,
m being
Hence we
by proceeding
as before,
log
And
x
a
xu
?
-?
tends to
Should the limit be unity, we must have recourse to the third series of (7), the resulting test being that the proposed series is convergent or divergent according as, x being indefinitely
increased,
the
function
4-x
ad
The forms of the functions involved in the succeeding tests, Practically, we are directed to inf., are now obvious.
i
ux
Ix
, I
1
I
Ju LJu LL*Juiv x
'
xu x
'
xlxu x
lllx
q
'
llx
mix
(A),
and the
first
of these
which
tends, as
x is
indefinitely increased
74
SERIES.
[CH. V.
convergent or divergent.
The
criteria
may be
<
presented
in another form.
For
representing
in (A)
by u
x
(x) ,
and applying
CO
we have
1(f)
(x)
c/>'
(x)
t
1
'
_ x<p' (x)
c/>
Ix
<f)
(x)
(x)
'
X
llx
(<f)'
(x)
[c/>
(x)
x)
'
log x
= logcc \x
and
so on.
T
4>(x)
of functions (A)
is
replaced
by
the system
x <l>'( x )
J
X X
Ux
\}
It
^$)- )l
\ &Q
is
P)-
was
was
virtually under this form that the system of functions originally presented by Prof. De Morgan, [Differential
as follows.
Calculus, pp. 3257). The law of formation If n represent the nih function, then
pw =n(P.-i)
(ii).
There exists yet another and equivalent system of de7. termining functions which in particular cases possesses great advantages over the two above noted. It is obtained by substituting in Prof.
De Morgan's
forms
may
-
U 1
CD
for
->
The
Since
ux
</>
x)
we have
ART.
7.]
SERIES.
75
U *+ l
$ X
i
<j>(x+l)-<j>(x)
Now by
known theorem on
and
1.
Hence
ux
_ 1 = (j>'(x + 6)
_<j>'{x)
4>'(x
+ 8)
(12).
for,
Now
6 r
;,,.
has
since
uK
by hypothesis converges
(x)
as
is
indefinitely increased,
,
, >
assumes
(a?)
the form
GO
;
'
therefore
4>(x
<f>
+
(x)
0)
4>'(x
<j>
+ 0)
(x)
(b
(x)
But
(
N
the
first
member has
for
its
limits
-7-7-T
and
$ f + !) (x)
</>
i.e.
and *(? +
<f>
and the
(x)
unity as
is
indefinitely increased,
limit.
since
N
.
by hypothesis
1.
?= tends to that
ux
Thus
(12)
becomes
Substituting therefore in
functions
(B),
we
(-)
{&-)-}
76
SERIES.
[CH. V.
&c
n
l
(C),
1).
Pn+1 =
x (Pn -
The
results of the
above inquiry
may
be collected into
of
the function
Rule. u - 2tl
.
Determine
According as
is
convergent or divergent.
But if that
of whichsoever
first
is
limiting value he unity, seek the limiting values most convenient of the three systems of func-
According as, in the system chosen, the function whose limiting value is not unity, assumes a limiting value greater or less than unity, the series is convergent or divergent.
tions (A), (B), (C).
Ex.
5.
series
be
(
1+
Here
2f
+&c V^+-i 4* 3
3
13 )-
ux
= -^
x*~
therefore,
xx
(x+1)** 1
(x+1)
x+1
and x being
Now
have
applying the
criterion of the
system (A), we
ux
and the limiting value
criterion in (A),
is
x+
again unity.
we have
xux _ kf Ux ~ Ux
lx
xllx
'
ART.
8.]
SERIES.
77
is 0.
the limiting value of which found in the usual Hence the series is divergent.
way
Ex.
6.
of Ex. 4, viz.
,
g we
^(q + 1)
q(q+
2
l)(g
+ 2)
1,
Therefore
first criterion
_
which tends
unity.
If 6
(6
a)
x
'
+x
to the limit 6 -a. The series is therefore convergent or divergent according as 6 - a is greater or less than
-a
is
equal to unity,
we
have,
by the second
criterion
of (C),
&
H-')-}=M^-}
_ alx
a
+ x'
is 0,
since b
still
1.
The
limiting value
is
divergent.
It appears, therefore, 1st, that the series (14) is convergent or divergent according as t is less or greater than 1 ; 2ndly, that \i t l the series is convergent if b 1, divergent
if
= 6 a ^ 1.
The
is
a>
9.
Art. 3,
extension of the theory of degree referred to in involved in the demonstration of the above criteria.
78
SERIES.
[CH. V.
two functions of x are, in the ordinary sense of the of the same degree, i. e. when they respectively involve the same highest powers of x, they tend, x being indefinitely increased, to a ratio which is finite yet not equal to viz. to the ratio of the respective coefficients of that highest power. Now let the converse of this proposition be assumed as the definition of equality of degree, i.e. let any two functions of x be said to be of the same degree when the ratio between them tends, x being indefinitely increased, to a finite limit which is not equal to 0. Then are the several functions
term,
;
When
x
with which
(hc)
m
,
xlx
is
(llx)
m
,
&c,
or
<f>
(x)
tions
monstrations of the successive criteria, so many interposi1+a of degree between x and x however small a may For x being indefinitely increased, we have be.
,
hm
v
X
m
=oo,
Inn
,.
\J =
0,
xlx(llx)
xlxillxT
(lx)
is
intermediate in
degree
xlx
(llx)
(lx)
1+a
&c.
And
position of
m = 1,
is
new
function.
It may be noted in conclusion that the first criterion of the system (A) was originally demonstrated by Cauchy, and the Bertrand* first of the system (C) by Eaabe {Crelle,Yo\. IX.). to whom the comparison of the three systems is due, has demonstrated that if one of the criteria should fail from the absence of a definite limit, the succeeding criteria will also The possibility of their continued fail in the same way. failure through the continued reproduction of the definite limit 1, is a question which has indeed been noticed but has scarcely been discussed.
* Liouville's
p. 35.
EX.
1.]
EXERCISES.
79
EXEECISES.
1.
Find,
by an
Art. 4,
two
ad
in f-
a .=
1,
2.
The
series
a?
+ CC C
3?
-+
&c
is
convergent
if
is
if
is
greater than
4,
that
*(i)
4>{\)+m4> (m)
Del "
P osltlve
'
The hypergeometrical
ah
ca
series
,
a
c
1
,
+ 1) b (b + 1) (c + 1) a (a + 1)
(a
if
is
convergent
If
if
#<
divergent
x>
it is convergent a; = 1 c + d a b < 1.
if c
+ d a b>
4
a
1,
divergent
if
5.
The
1,
series
+ <
+ +-, + --^ -f
Z o
&c.
is
convergent
if
>
divergent
if b
1.
80
CHAPTER
VI.
SERIES.
It has been seen that the finite summation of series 1. depends upon our ability to express in finite algebraical terms, the result of the operation % performed upon the general term When such finite expression is beyond our of the series. powers, theorems of approximation must be employed. And
the constitution of the symbol
as expressed
by
the equation
2=
easy.
(6'
l -l)-...(l),
or,
under particular circumstances, in ascending powers of A. Hence two classes of theorems arise, viz. 1st, those which express the sum of a series whose general term is given, by a new and rapidly convergent series proceeding according to
the differential coefficients of the general term, or to the differential coefficients of some important factor of the general term 2ndly, those which differ from the above only in that differences take the place of differential coefficients. The former class of theorems is the more important, but examples of both and illustrations of their use will be given.
;
2.
Prop.
differential coefficients
To develope Xu x in a of ux
.
series
proceeding hy the
Since
%ux =
dx (e
l)~
-7-
uz
we must expand
(e
dx
l)~
in
ascending powers of
will
ART.
2.]
SERIES.
81
plicity
be determined by that of the function (e' 1)~\ For simwe will first deduce a few terms of the expansion and afterwards determine its general law. Now
(e'-l)-
t'
on actual division.
the subject u x
_*
,
Hence replacing
by
and restoring
we have
1
dV
dux
d 3ux
= C + JuJx--ux +
(2).
For most applications this would suffice, but we to determine the law of the series.
shall proceed
The development of the function (e* l)" cannot be directly obtained by Maclaurin's theorem, since, as appears from (1), it contains a negative index; but it may be obtained by
expanding t
the
result
by
t.
Eeferring to
(1)
we
-will
have
\t
t.
for its
power of
second term. It will now be shewn that term of the expansion which involves an odd Let
jrri
[
+*
all
Then
6
82
/_ e-i
,
and as
represented
by
B contain
the terms
Now
if for
the
moment we
6
1 by
4
0,
we have
e'-l
(a).
II.
A*0 n+2
I
&c,
in
which the
A"0 m
coefficient of f" is
-=m\
Hence the
coefficient
of
in (3) is
at
,
-j(T
- iAOm + J A
...
\\[
+ ^^ AwOml m+
J
(4),
since
Aw
vanishes
noted that
when m =
It is to
1.
be
The expression (4) determines in succession all the coeffi_1 f in ascending powers cients of the development of t(e 1) It must therefore, and it will, vanish when receives of t. any odd value greater than 1.
From
of
t
these results
l
we may conclude
(e
l)~
t
will
1
(e
-iy =
^+A +A
A m_
l
+ A/+A/ + &c
(4).
(5),
where in general
is
expressed by
ART.
3.]
SERIES.
83
development in the
^^ = 7-5 + ^-?V + f^
The
and
&c
<
quantities &c. are called Bernoulli's v 3 their general expression will evidently be
,
B B
r+1
numbers,
K-, =
Hence we
-O
(-
l)
{- 4A(T + JAW...
+^}
+ &o
(7).
find
r
^1 du n tu^G + JuJx-iux+
7
1
^-^^
T
B. d 3 U
(8).
Or, actually calculating a few of the coefficients by the table of the differences of given in Chap, n.,
means
of
du x 12 dx
1
d 3 uT 720 dx
1
z
+ 30240
3.
dx"
Attention has been directed (Differential Equations, p. 376) to the interrogative character of inverse forms such as
(^-irv
The
strictly speaking, to
object of a theorem of transformation like the above determine a function of x such that if
it
d dx
is,
we
perform upon
1) the result will be ux To the inquiry what that function is, a legitimate transformation will necessarily give a correct but not necessarily the most general answer. Thus G in the second member of (8) is, from the mode of its introduction, the constant of ordinary integration ; but for the most general expression of %ux C ought to be a
above case
this is e
62
84
SERIES.
[CH. VI.
periodical quantity, subject only to the condition of resuming the same value for values of x differing by unity. In the applications to which we shall proceed the values of x involved will be integral, so that it will suffice to regard C as Still it is important that the true relation a simple constant. of the two members of the equation (8) should be understood.
The
Bernoulli's
z?
following table contains the values of the numbers calculated from (7),
t>
first
ten of
_ 1_ 6'
3_ 30'
7?
5_ 42'
7?
'
=__
30'
Ti
9_ ___ 66'
b -i?L _
11
-I
'
2730
13
'
H15 ~ 3617
510
'
43867 B = ^m>
^ = ~23ir
1222277
(10)
It will be noted that they are ultimately divergent. It can seldom however be necessary to carry the series for *%ux further than is done in (9), and it will be shewn that the employment
of
its
convergent portion
is sufficient.
Applications.
4.
The
Sux
in
(9),
Art. 2, gives us
any
rational
A
and
entire function of x.
Ex.
1.
Thus making ux
= x we
,
have
J
/v = C+
*As
2
Ju
Jb
12
JO
dx
720
dx 3
5~2 + 3""30is
of chief importance
when
finite
sum-
impossible.
Ex.
2.
Thus making ux
xx
we
,
have
ART.
5.]
SERIES.
85
2*
=
The
~ 1 0_
1 *
x
I
aj
2x*
+ -i-f^)--_f 12 720
,
/-
2\
Va? /
J_
2x2
_L + _1
6x
3
30#
_& c
value of
72 2
+ ^o ~r O 2 3
2
' '
1
o 2
~r
(-l)
Now by
2
what precedes,
3
2
III r+ +
2
2
'*'
{x-lf~
first
-r--- x 2x
member
is
+ 30^
1
6*
Let x
= co
then the
equal to
7T
by a known
Hence
1
member
2
reduces to C.
1 1
11
and if x be member will
5.
suffice.
7T
When
to
the
sum
of the series
ad
inf. is
unknown,
determine
or is
known
be
infinite,
we may approximately
C by
giving to x some value which will enable us to compare the expression for %ux in which the constant is involved with the actual value of 2w x obtained from the given series by addition
of its terms.
Ex.
3.
series
be 14-2
+ -...+-.
3
u,
we have
= - + 2, X
X
1
l +-L + ftc. _I + d + ]o g -i ri ?
86
To
determine G, assume x
10,
then
2*30258,
we have
C=
u
Ex.
If
'577215
+ log 8e x +
e
2x
\2x
+ T-&C.
? 2
120a;
for 1
. . .
x.
. . .
x,
we have
. . .
log u
+ log x
But 2
loga?
- log X
1.2
&
'
.-.
logu=C+(x+-J logx ~ x +
a;
j^
&c
(I 1 )-
= C+
e
c -z
(x
+ -A\ogx-x,
(12),
2x+i
whence
1.2.3...x =
1.2.3... 2x
xxx+i
x
(2x)
=e
X
,
c "2x
(13).
4 6
.
...
2x
= 2 x ec~x x xx+i
(14).
AET.
6.]
SEEIES.
87
3.5.7
.
...
(2a;j-
= 2 x+l e~x xx
.
whence
2. 4. 6. ..2a;
... (2a;
3.5.7
-1)
eV = rs
2*
infinite,
__
(2a;
...
- 2)
(2a;-
V(2a?)
1)
/ /tt\
3., 3.5.7
\f \2J
.'.
(7=logV(27r).
this value in (11)
And
we
find
now, substituting
and determining
'
w,
W = V(27r)xa^*X""^"
"^ +&C
+&c
"
= V(27ra;)./.e"^'^
If
(15),
we
9
12 *
86to3
in descending
pow-
ers of
x we
aJ
2 .3...
= 7(2^).^(l +
l+
^-^
+ &c.)
(16).
Hence
for
we may assume
(17),
1.2.3
the ratio of the two
infinity.
..*x=**/(2irx) (-}
members tending to unity as x tends to speaking generally it is with the ratios, not the actual values of functions of large numbers, that we are
And
concerned.
Yet even in cases in which the value of x involved is 6. not only large but infinite, it may be necessary to take account of terms involving negative powers; just as in the
88
SERIES.
[CH. VI.
Differential Calculus
Ex.
5.
x being
x'
1
.
We have,
fied,
by
1.2.3...
a;
= V(2w)af *6-(l +
<
^...),
...
2x =
2x
x+i
>J-jrx"
H e~x (l + ^-)
...
+ ^)
3.5... {2x
- 1) =
2*+ *arV3
'
--.
j
= 2 '+l
^(
t-
Hence
2.4... 2s
3.5... (2^-1)
^ /ili^^
1
24s 7
=M
Ex.
6.
i-)
(18)
'
To sum
1
the series
1
"i
T^n
""
J_
o2/i
'
J_
/i
'in
~r
ART.
6.]
SERIES.
89
we have
~x
2n
'
^x2n
1
,
(2w
- 1) ar "-1
2
??
1
2a2"
2?? (2rc
2^
2n+1
1)
(2n
H*
+ 2)
'
720a*
For each particular value of n the constant C might be determined approximately as in Ex. 3, but its general expression may be found as follows.
Let x become
1
infinite,
^+
32^
+ &c ad f- =
-
0,
and
it
first
member.
Now
COt^ =
= V(-l)^7 + _ f V(-l)(l V
g2ev(-i)
1
v v
20 v(-i)
_1
&C
'j_
V(-i)
.
|2^V(-1)
1
2
2
1.2
*
in
i.2
is
5 ^-fiTi^- &c
7?
*n
which the
coefficient of
2n_1
2 2w
~\
1.2...
2rc
member and
dif-
90
~
20
6>V
26
20
2
(,
&
N_1
"^
!!
OoC.
2
1
+&c .),_|
6
(
+ I + Ji+&c
5
- 1+ ^V
in
- 2" 6
+ - + &c
a
6
+ &c,
which the
coefficient of
2n_1
is
1
Equating
element,
same
we have
1
+ 2 *n +
(27r)
2n
32, -h 48 .
+ &c
(iyj -
2 (1
...
2n)
Hence
W
1
1
o/io 2 (1 2
.
...
o\ 2w)
/o~ (2w
2n_1 l\ ^2n-l 1)
2^n
6ic
2n+1
+
the expression required.
2^ (2n
l)
(2n
2w+3
+ 2)
720*
+&C
(2 j '
The above investigation gives us also another expres7. sion for Bernoulli's numbers, viz.
2(1 2
"
2 ")
;:
(2 7r )
(1
while for the divided coefficients which appear in the actual expansion of %ux we have the expression
2n_ x
2
l
s
1.2. ..2n
(2tt)
+ -L +
^ + &c-)
(22).
Hence too it appears, First, that the series of Bernoulli's numbers is ultimately divergent, for, n becoming indefinitely great, we have
ART.
8.]
SERIES.
91
^=\
B^
\
(23),
%ux
Secondly, that the divided coefficients in the expansion of are ultimately convergent. For
f
Limit
B
.
2
(2ir)"
|l
...
(2
+ 2)
'
...
in)
M^
'
_A
(2ir)
2/i
L
47T
(24).
we have
+*=W- 1 >~l?5:|?
The second
and endeavour
line of this expression
to
^shall represent
its
we
by B,
value.
Now by
(22),
^V-i
1.2... 2r
2
(27r)
2rAw=1
^ w=0O J_
m*"
Therefore substituting,
_ 9 5*m=
Assume
^r=oo
J"J
to
92
And
rem
then,
making
= ee we are
,
led
by the
general theo,
for the
summation of
2
series
(Diff.
Equations
p.
431)*
d2n+1 n
is {Diff.
Equations, p. 383)
2n
(
'
l)
n
f
(2m7rY
n+l
\
or,
^ ^ axr j
+
1
,
2W+1
since
we have
to
which
sin (2imrx)
,
= 0,
(?<;
l)
r v2w +i J sm 2m7ric
^
7
2r.+l
o n+1
w^ dx.
Hence
^ Sin
2mTO
^ ^
->
ffain 2irx
2rt+1
sin 47ttc
2to+1
d2n+1 u
dx'
2
,_-x
il(27T
(4tt)
'
as
makes
d2n+1 w
2n+1
to vanish, the
upper limit
a?.
Hence
if
within
* The theorem which is of most frequent application is the following. u = up xp + up{r xP* r + up+ir xP+ ir + &c. ad inf., then supposing the law of the
efficients to
If
co-
be
um =<p
and making
(m)
um - T
or
um
cf)
(m) u m _
= 0,
x=e
s , the differential
equation for
will be
u-
*
<t>
{ d
\ r6
\ddr
P0 pe to.
The
student will easily deduce this from the more general theorem referred
ART.
8.]
SERIES.
93
the
dZn+l u
the limits of integration
'
n+t
retain
a constant sign,
value of
fn+1 u
+l
'
JW;
therefore,
8-
+1+
(47T)
aB+1,
"J^*
,
.
1,1
2,m
'
J
^'j
d**u
271+1
(27r)
(4tt)
dx2ni
therefore,
by
(22),
,-,
2tt1.2...
2ndxM
(27).
When
first
is
tion of form
between the
its
n- l
large this expression tends to a strict interpolalast term of the series given and the
term of
remainder,
d*
viz.,
R_
1.2...
it
2-l
2ndx' n
~1
iand
+ 2)S?^ 2) dx
;
(28) '
dx
(26)
is,
in the limit, a
mean
(
d? u f t^p=i an ^
is
n~x
d 2n+1 u
v
an+i
m
.
28 )-
And
tn i s interpolation of form
usually accompanied by interpolation of value, though without specifying the form of the function u we can never affirm that such will be the case.
The practical conclusion is that the summation of the convergent terms of the series for 2w affords a sufficient approximation, except when the first differential coefficient in the remainder changes sign within the limits of integration.
The
series for
S log
gent terms
x,
xm
-
'
^{ax +
b)
m1
94
%ux
Beside the general expression for *Zux given in Art. 2, 9. there are certain other forms which suppose a particular constitution of the function u x and are advantageous under particular circumstances.
,
Thus the
series
0(0)
th
x has for its x + 1 term ( l) (j>(x), and for the sum of its Applying the theorem of Art. 2, (x). x terms X ( l) we should obtain for this an expansion proceeding according It is obviously to the differential coefficients of ( 1)*0 (x). desirable to substitute for this an expansion proceeding ac(x). cording to the differential coefficients of
first
By a theorem employed in the demonstration of the fourth integrable form of Chap. IV. Art. 2, we have
s(-ir0H = (-ir(-/-i)Now
0(^)
= (-l)~*(<?+l)-*0(a>).
J__J_
t
-2
U_J|
1.2.3.4
=I_I+-5m
2
1.2
?!?+&C. T
^o^^-tAiC-^^
whence
rf
UJU
fZ
ART.
1.0.]
SERIES.
95
Therefore
2(-ir<^( a) =
a+(-ir{-^(
a;
)+ I^(2 -i)f(a;
(29).
(30),
(-1)*
<f>
[x) in
ascending
differences of
(x).
We have hy Art.
$(-iy<f>(x)
9,
i
= (-ir
(e
i +ir<i>(x)
= (-ir(2+Ar<M*)
= G+ (
ir i^p -*m +
*m _
inf.,
&c
(31).
Suppose
series
</>
it
(0)
</>
(1)
+ <
(2)
cj>
(3)
+ &c.
0.
ad
In
(31) let
x be
(0)
<j>
(1)
<j>
(2)
- &c. ad inf. =
(7.
96
SERIES.
[CH. VI.
(0)
<j)
(1)
+
4
<j)
(2)
= 1 (0)_Ai (0)
to the
- &c. ad A^(0)_
8
inf.
This theorem, which may be applied with great advantage summation of slowly convergent series whose terms are alternately positive and negative, may also be established as
follows,
<j>
(0)
(1)
(/>
(2)
- &c. =
(l
-D +D
1
-Z> 3 +
&c.)
< (0)
= =
(l+i>)(2
</)(0)
+ A)- </>(0) /l A A A
1
A,
"2
_<ft(0)
A0(O)
4
c/>(0)
* C '
The above results are virtually included in the two 11. x following more general theorems by which %<j)(x)t is expressed in series proceeding according to the differential coefficients and according to the differences of (j>(x), viz.
1st.
{x)f
= C-
- U(x) +
A^
s
where
A =
A0"
l-i!
2ndly.
2*(a:)f =
0- /- L(x) + ~^(x)
2
(i4l)
A^( + &cj
a!)
(34).
ART. 11.]
SERIES.
write,
97
iv.
To
p. 53,
we must
Chap.
= f{t(l+A)-l}-
cj>(x),
last
two members
and of
expansion being effected by the reciprocal form of Maclaurin's theorem, Chap. II. Art. 11.
EXAMPLES.
1.
Approximate
to the
T
2.
..
111 + + ....+
12
7T
1000
1
+ log 2
-t-
log 3
+log 1000
3.
,11
1
A
5
_6,
shew that
i =2+
(7
^-^ + - -&c 4
2
6
for
*'
C.
Find an approximation
3.5...
2
.
(2a;
+ 1)
2x
B. F. D.
98
;
EXERCISES.
[CH. VI.
supposing x large the first negative power of x which presents itself being retained.
5.
Transform the
,
series
1
The
series
ax x
x*
+ a x &c. may
3
he transformed
into
(XimX
.
where
Aan = a+1 - a.
99
CHAPTER
VII.
EQUATIONS OF DIFFERENCES.
ordinary equation of differences is an expressed between an independent variable x, a, dependent 2 variable ux and any successive differences of u x as Au x A w x n ...A u x The order of the equation is determined by the order its degree by the index of the power of its highest difference in which that highest difference is involved, supposing the equation rational and integral in form.
1.
,
An
.
relation
Equations of differences may also be presented in a form involving successive values, instead of successive differences, n of the dependent variable for A u x can be expressed in terms of ux ux+l ... ux+n Chap. II. Art. 10.
;
,
Equations of differences are said to be linear when, considered functionally, they are of the first degree with respect 2 to u x Au x u x &c; or, supposing successive values of the independent variable to be employed instead of successive differences, when they are of the first degree with respect
, ,
to
u x u x ^ v ux+2 &c.
,
The connexion
of the
two cases
is
obvious.
Equations of differences which do not belong to the ordinary species, viz. equations of partial differences and of mixed differences, will be defined in another chapter.
The genesis of equations of differences is analogous to 2. that of differential equations. From a complete primitive
F(x,u x ,c)=0
connecting a dependent variable u x with
(1),
an independent
72
100
variable
EQUATIONS OF DIFFERENCES.
[CH. VII.
x and an
arbitrary constant
c,
equation
AF{x,ux ,c)=0
we
obtain,
by
eliminating
c,
<l>{x,u x ,Au x )
=
in
(3).
ferences,
f{x,ux ,u x+1 ) =
Either of these
4 )-
may
if,
of differences of the
order.
In like manner
F{x,ux cv c#
...cn
>=0
(5),
and from n successive equations derived from it by successive performances of the operation denoted by A, we eliminate c n we obtain an equation which will assume the form cv c2
. . .
,
A n u x =0
)
(6),
(7),
according as successive differences or successive values are employed. Either of these forms is typical of equations of th In (6), u x may be replaced by u. differences of the n order.
Ex.
1.
Assuming
as
complete primitive u = cx +
2
,
we
Aw = c,
by which, eliminating
c,
there results
2
,
u = xAu + (Aw)
the corresponding equation of differences of the first order. Thus too for any complete primitive of the form u = cx +f(c) there will exist an equation of differences of the form
= xAu +/(Aw)
ux =ca x + cb x
(8).
Ex.
2.
Assuming
as complete primitive
j
ART.
3.]
EQUATIONS OF DIFFERENCES.
+1
101
we have
c'b \ tco ux+1 =car- + x+ ux+2 = ca *
x
x+ \ + cbx+2
Hence
ux+1 - au x =
c(b-
a) b
x
,
x+1
.
~ ob^ - b
(*Vi
- au x = 0,
)
ux+ -(a
+ b)ux+1 + ahux =
being
contained
is
(9).
Here two
arbitrary
the
constants equation
of differences
in the of the
The arbitrary constants in the complete primitive of an 3. equation of differences are properly speaking periodical functions of x of the kind whose nature has been explained, and whose analytical expression has been determined in Chap. IV. Art. 1. They are constant with reference only to the operation A, and as such, are subject only to the condition of resuming the same value for values of x differing by unity; a condition which however reduces them to absolute constants when x admits of integral values only. The subject will be more fully discussed in Arts. 4 and 11 of this chapter.
The proposition converse to that of Art. 2, viz. that a complete primitive of an equation of differences of the n th order involves n arbitrary constants, has already been established by general considerations (Biff. Equations, p. 187).
Linear Equations of
4.
the First
Order.
The
ux+1
where
-A u
x
= Bx
(1),
x and x are given functions of x. consider the case in which the second member
We
is 0.
shall first
To
^+i~A^=0.
(2),
102
EQUATIONS OF DIFFERENCES.
[CH. VII.
we have
whence, the equation being true
for all values of x,
1
Ux
==
A x_
Ux-v
U x-%1
= -^-x-2 ^x-i
Ur+1
~ -A
Ur
ux+i A x A xl A x_2
r being an assumed
initial
. . .
Au
r
(3)
value of x.
be the arbitrary value of ux corresponding to x = r, (arbitrary because it being fixed the succeeding values of uz corresponding to x = r + 1, x =r + 2, &c, are determined in succession by (2), while u r is itself left undetermined) then
Let
(3)
gives
u x+i
whence
and
ux
^^-x A x_ A = CA^A^.... A
x
. . .
r,
r,
(4),
While, for any particular system of values of x differing by successive unities, C is an arbitrary constant, for the aggregate of all possible systems it is a periodical function of x, whose cycle of change is completed while x varies continuously through unity. Thus, suppose the initial value of x to be 0, then whatever arbitrary value we assign to u the values of u v u2 u3 &c. are rigorously determined by the equation (2). Here then C, which represents the value of uQ is an arbitrary constant, and we have
,
Ux+1
(;A X
A x_
.
...
Suppose however the initial value of x to be ^-, and let Then, whatever arbitrary be the corresponding value of ux value we assign to JE, the system of values of u* us &c. will
,
be rigorously determined by
(2),
ART.
4.]
EQUATIONS OF DIFFERENCES.
103
The given equation of differences establishes however no connexion between G and E. The aggregate of possible solutions is therefore comprised in (4) supposing G therein to be an x
arbitrary periodical function ofx completing its changes while changes through unity, and therefore becoming a simple arbitrary constant for any system of values of x differing by
successive unities.
We may for convenience express (4) in the form u,= CPA^ (5), where P is a symbol of operation denoting the indefinite continued product of the successive values which the function of x which it precedes, assumes while x successively decreases
,
by
unity.
it
may
= e\
Then u x+1 =
t+A
and
(2)
becomes
e^-A^^O;
.-.
At =0 e^-A
,
whence
At
*
= log A x
G,
Therefore
ux
as before,
= ^'-<.-,+, =C,PA X_
Resuming the general equation (1) let us give to u x the form above determined, only replacing G by a variable parameter Cx and then, in analogy with the known method
,
for linear
differential
We have
ux
= GX PA X_^
104
EQUATIONS OF DIFFERENCES.
(i)
[CH.
whence
becomes
Am PAa^<mPAm C*+1 Gx PA X = Bx
,
(ACX )PA X = BX
whence
ac = pI>
*
*-*+*
..u, =
(6);
PAh* 1+ cX
- (x + 1)
ux
(7),
Ex.
1.
Given ux+1
=1
...
(x
is
+ 1).
apparent that
member
it
A x = x + 1, PA x_ =x(x-l)
1
... 1,
PA x
.'.
~l *
>
PA~ X
... 1
>
ux
= x {x 1)
x (x+ C).
b are constant.
Ex.
2.
Given ux+1 - au x =
h,
where a and
x
,
Here
A x a,
and
M
t
PA X = a
therefore
1-a + G a%
where
is
an arbitrary constant.
when
We may observe before dismissing the above example, that A x a the complete value of PA X is <f multiplied by
For
an indeterminate constant.
ART.
4.]
EQUATIONS OP DIFFERENCES.
105
PA X = A X A X_
=
a .a .a
...
,
...
xr
+1
times,
But were this value employed, the indeterminate constant would in one term of the general solution (7) disappear by division, and in the other merge into the arbitrary conActually we made use of the particular value correstant G. sponding to r 1, and this is what inmost cases it will be
a~
r+1
convenient to do.
Ex.
3.
Given
Aux +
2u x
,
= xl.
substituting in
(7)
and reducing
u x =c{-iy-l-\.
Ex.
4.
tt
^_^ = __
ax
~x
We find
t
1
T2
W^W +c
+'
1*
...+
H
,
When, as in the above example, the summation denoted by cannot be effected in finite terms, it is convenient to employ In so doing we have supas above an indeterminate series. posed the solution to have reference to positive and integral The more general form would be values of x.
x-l f 1
106
EQUATIONS OF DIFFERENCES.
Linear Equations with constant
[CH. VII.
Coefficients,
general form of a linear equation with constant successive values not increments of the independent variable, and having its second member equal to 0, is
5.
The
coefficients
when expressed by
ux+n
If
+A
,
we assume
u x+n _ x
'
. . .
and
division
by cm x the
auxiliary equation
~'
2
mn + A mn + A*"
x
A=
0.
the roots of this equation are all different, n particular solutions of the form assumed above, corresponding to the n particular determinations of m. It is also evident from the linear form of the given equation, that the general value of ux will be the sum of the n particular values thus obtained.
therefore,
There exist
when
+ 6ux =0
leads
to
the
m - 5m + 6=0,
and therefore admits of the two particular integrals
^=
^=
When
roots, or
Cl (2),
n-y(8)",
+ c2 (3).
the auxiliary equation has imaginary roots, or equal the given equation has a second member, the principles employed in the corresponding cases of differential equations may be adopted without essential change.
when
If
lents
Ex.
Given ux =
c x \a
u x+2 +a*u x =
0.
X
Here
V(- 1)}"+
c2
{-a V(2
1)}
= ax
cx jcos
| + V(- 1) sin |j + c
jcos
| -V(- 1) sin ||
= a*Ll cos(^)+2?BinpJ
ART.
If
6.]
EQUATIONS OF DIFFERENCES.
107
have equal roots the solution may be derived by the limits (assuming the principle of continuity) from the answering form of the solution when the roots are un-
method of
equal.
Ex.
7.
a ux
= 0.
The
solution
ux
b approaching to a.
= c ax +
x
c2b
x
,
Therefore
u =limof
= Ca
If the second
x
Cax
C'xa
x -X
a =a {C + C
b
x
x).
member be not equal to the form of the solube deduced from the particular form which it assumes when the second member is 0, by treating the constants as
tion
may
variable parameters.
But in the last case, and usually in the preceding one, it is simpler to proceed by the symbolical method of the following
section.
uitli
constant coefficients.
The
efficients is of
the form
uXn +
A^x^_ + A u x+n_
x
. . .
+ A nux = X
by
(1)
Hence
if
D be a symbol
2ty
(a)
of operation defined
= <(+!)
~2
(2),
Z>X + Ajr*um + A Dn
2
ux
. . .
+ A nu x = X,
u x = X,
or,
(Dn + A J)*-* +
and
this
A Dn~*
2
. . .
+ An
we
by
(3);
f{D)u = X
108
EQUATIONS OF DIFFERENCES.
u
[CH. VII.
whence
{f{B)\'
(4),
a form indeed differing in interpretation from (3) only in that it presents u as the object of quest (Diff. Equations, p. 375). Suppose the roots of the auxiliary equation f(m) = to be l all different, then, resolving {f(D))~ as if it were a rational fraction, we have a result of the form
u = {N
where
1
JV!
(a 1
N
2
...
n.
But
if a
particular
we have
corresponding to that
{M (D - a)- + M
1
(D *
a)~
. . .
+ Mr (D - a)~ X.
r
]
... (6),
where
M=
t
1.2... (r *) \a2/
(|Y~LM
(z
a) r
(*
= a).
solution of the proposed equation is therefore made to all cases on the performance of the operation denoted by (D a)~ X; and this, since we are permitted by the
The
depend in
symbolical form of Taylor's theorem to substitute may be referred to the known theorems
dx
for
D,
(
Then
0>
vrlf
-Km)*-.
<
vn-v
mT mX #Urf-X-^# B+ .)x...{l).
I I
LI
-rr
Ltf
first
* It is only while writing this work that I have become acquainted with the treatise of M. Lobatto, entitled Theorie des Caracteristiques, pubIt contains the theorem in the text, the analogous lished at Amsterdam in 1837. theorem in differential equations, and in one word the whole of the theory of rediscovered in England a year or two coefficients constant linear equations with
remarkable
first and second volumes of the Cambridge Mathematical Journal. Every English mathematician will rejoice to see justice done to M. Lobatto. It is proper to add that M. Lobatto's treatise does not contain any anticipation of the higher symbolical methods subsequently developed in this country.
ART.
6.]
EQUATIONS OP DIFFERENCES.
(Z>
109
_ a)-VX =
(e^ -
a)"V loga
^^{aD-aY'X
=
_1
r (i)-l)- X
f
since
(D-
1)
= A_1 = 2.
x
therefore
Xby
I by I,
and
(D-ay X=ax - t a
i i i
X
1
,
(8).
It is to
= c + c cc -f c x
x
2
. . .
+ c^x^
the above theorem, with the complementary function in second member, will take the form
its
(D - aY
X = aT 2V*X + a'
1
(c
+ c,a?
. . .
+ c^aT )
1
. . .
(9)
If the equation f(m) = have a pair of imaginary roots a ft \/( 1), that pair occurring only once, then, as appears from (5), the value of u will contain a pair of conjugate terms of the form
{jf+#v(-i)H+/v(-i)r 2[{+/3V(--i)p^
Now
{a {a
let
= p cos
I)}""
1
0,
fi
= p sin
0,
whence
+ V(1) sin
tan
-1
then
1) 0}
& v (-
= Px x j3 V(- l)p = p~
~l
{cos (a
{cos
- 1)
1) sin
(x
^0 + V(-
x6}.
Whence,
substituting and reducing, we obtain as the real expression of the portion of the value of u corresponding to the imaginary roots in question,
S
p
'
110
EQUATIONS OF DIFFERENCES.
function
is
[CH. VII.
The complementary
"
evidently
2/T
which
is
_+
'
B smx0).
shewn by
If the imaginary pair of roots be repeated r times, then, as is (6), there will exist in u a series of pairs of conju-
{M + N V(-
)}{*
[{a
+ V(- 1)1-1]
and
i receiving every integer value from 1 to r inclusive, and being different for each different value of *. Proceeding
as before, we find that the real expression of that portion of the value of u which corresponds to the imaginary roots in question will consist of a series of terms of the form
T {McoB(x-i)0-N&m(x-i)0}t p-*coax0X~\
t
2p
i
'
varying from
It is evident also that the complementary function introduced by summation will ultimately be
^^
'
A B
l9
l9
When the second member either consists of a series 7. x or is a rational and integral of exponentials of the form ha function of x, or is composed of terms resolvable into factors of either of these species, the process of solution may be simplified by methods analogous to those employed in the corresponding cases of differential equations ; and, though it will still be necessary to determine the roots of the auxiliary equation f(m) = 0, it will not be necessary to decompose the
,
l
.
ART.
7.]
EQUATIONS OF DIFFERENCES.
Ill
In each of these cases, representing the given equation in the form f(D) u = X, we are permitted to write
u={f(D)rX+{f(D)ro
the second term in the right-hand
(13),
representing the complementary function. And, as this function introduces the requisite number of constants, it suffices to deduce any particular value of the first term.
First, let
member
X consist of
Then
above described.
since
,dx\
(14),
X m (13)
may
be determined.
by combining with
Cases of failure from F(a) becoming infinite maybe treated the term (11) such a term derived from
the complementary function as will replace the infinite by an indeterminate form, and then applying the rule for vanishing
fractions.
Secondly,
1
Then we may
[f{D)}~
a rational and integral function of x. into 1 + A, and so develope in ascending powers of A, and finally perform the
let
X be
either convert
summation
,
or of difference,
1
on
X;
or,
converting
into e
,
d dx
and perform on
the resulting
operations of integration or differentiation. The solution is necessarily finite, since if be of the nth degree both differences and differentials of an order higher than the nth vanish.
Thirdly,
let
n
.
Now
F(D) a x
(a?)
= F{erx =e
xl0 * a
xloga
tj>
(x)
F{e^
lC&a
)<j>(x)
=a
F(aD)cj>(x).
112
EQUATIONS OF DIFFERENCES.
[CH. VII.
Hence
By
this
is
already considered.
To the above cases various others arising from the introduction of circular functions may be reduced but it is unnecessary to pursue the subject.
;
Lastly, it may be observed- that when, in the application of the general method, the final summations can in no other way be effected, we must revert to the definition of the symbol as in Ex. 3, and write down the series of terms which it
may happen
to indicate.
It will be more instructive to exemplify the principles 8. above explained than merely to apply the results which have been arrived at.
Ex.
8.
ux+2
ux
2
5ux+l +
_1
1
6ux
a*
Here
3-2(|-2-2(|
a\
x
,
,
(aV
we should proceed
ux
a?-5a + S
+ CS X +C'2 X
ART.
If
8.]
EQUATIONS OF DIFFERENCES.
2,
113
may
a 3 or write
Suppose a
3,
then
we
ux
= limit
of
2 - ha aa
+6
,
+ <?3* + '2*
Ex.
9.
Given
w x+2
Symbolically
we have
(D 2 -4D + 4)^
x
.*.
ux
x),
by
(11).
Now
{aD 2)"2
a?
=
Hence,
finally,
a;
2a
*
(a-2) 2 ~(a-2) 3
Ex.
10.
^2 + a ws = cos mx.
2
Symbolically
we have
(B2 + a2 )ux
1
= cos mx
.-.
m* + {B + a )'
0.
But
(Z>
+a )~
2
cosma;
2
1
= l(D + a )B. F. D.
mW( -
1)
-mxVrl)
}
114
EQUATIONS OF DIFFERENCES.
[CH. VII.
_1 =a
and,
|/ 2mV(-l)
2\-l
mW{-l)
_|_
-2mV(-l)
_j_
2\-l
g-mWHlJJ
cos
2 4
,.
:
5 2
on reduction
by Ex.
6,
(D2 + a2 )"
= ax (A cos
cos
+5
sin
_ /
.
.'.
uv
15
=a
11
cos ma;
-f
9 2
z 4
+ 2a cos2m + l
u x+2
2
m , 2)
(a?
+ a M- c os
V
7ra?
^ + 5sin 2/
.
irx\
Ex.
a ux = tan ma?.
ux
= (D - a )2
2
tan
1
mx +
(Z>
-a
2 -1
)
= J- {(> - a)" - (D + a) -1
ACL
tan maj
+ Cax + C" (- a) x
2a
(- a)*'
+ Oa +C"(-a) x
As the summation may write
_-,
we
Za x t?mmx =
,
__
(tan
\
m
-\
tan
2m *
tan (x
...
,..,
1)
m]
2 ( )
tan mx.
reducible
to
linear
equations
with
constant
There are certain forms of equations which a transfor9. mation enables us to reduce to the class of linear equations with constant coefficients above considered. Far less attention has indeed been paid to such reductions than to the corresponding ones in differential equations, and the number of known The following are the most important. cases is small.
ART.
9.]
EQUATIONS OF DIFFERENCES.
Equations of the form
115
1st.
+A<P
()
$ X(
1)
*V-2
(1),
+A
where
cj>{x)cl>(x-l)<l>{x-2)u x+n _3 3
+ &c. = X
may
A X 2 ... A n are constant, and <j> (x) a known function, be reduced to equations with constant coefficients by
ux
assuming
= (j>(x-n)(j)(x-n-l) ...0(1)^
(2).
For
</>
x)
</>
(a
!)
1)
</>
(a
- 2)
...
<j>
(1)
i>
x+n ,
Vi = *
and
so on
;
~2
</>
(!)
Vi
factor
(x)
(x
1) ...
<j>
(1),
we
get,
Vx+n
+i
Vl + A^
+ -2
+ &C = ^(^fy^x-l)
coefficients.
...<(1)
(3),
to
In effecting the above transformation we have supposed x admit of a system of positive integral values. The general transformation would obviously be
ux
r being
(j>
{x
n) <
{x
n 1)
. . .
j> (r),
as initial.
2ndly.
ux+n
+A
x a u XJrn _ x
+ Ayxux+n_ + &c. = X,
2
(x)
= ax
they
may
t
U* + n
+ Ai> 0) Ux+n- + ~~
66
<f>
()
<f>
(x
- 1) ^+n_ + &C. = X
2
(4).
82
116
Hence,
EQUATIONS OF DIFFERENCES.
to integrate
[CH. VII.
them
it is
(x n) (x
n+1)
a (5).
=
3rdly.
r~"v
ux+1 ux + ax ux+1 +b x ux
= cx
(6)
Assume
"
v*
Then
<
+l
vx
vx
Whence
v x+2
we
find
(7),
+(b x
the a linear equation whose coefficients will be constant functions b x a x+l and ax b x + cx are constant, and which again by the previous section may be reduced to an equation with constant coefficients if those functions are of the respective
if
forms
A(f> (x),
B<j>(x)<l>(x-i).
also be integrated
The above
condition c x
(7)
equation
may
is
satisfied.
becomes
{V>+(bx -ax
^D-a b }v
= 0.
Now
this
may
+ *.)(2>-cvK0,
ART.
9.]
EQUATIONS OF DIFFERENCES.
117
which,
of the
lastly,
first
may
order,
\
J
Wwx
,
We
As
can
now
determine in succession
vx and ux
upon the
the value of v x is in all these cases made to depend solution of an equation of differences of the second order, it will involve two arbitrary constants, but they will effectively be reduced to one in the final substitution of the derived value of ux in (6).
4thly. Some non-linear equations may be solved by means of the relations which connect the successive values of circular functions.
Ex.
12.
ux+1 u x
- ax (u x+1 - ux + 1 = 0.
)
Here we have
l+ux+1 ux
mation ux
ax
whence
vx
= C + Stan' ,
1
ux Ex.
13.
tan C + 2 tan"
(
(1
Given ux+1 ux
V{(1
- u*x+l
- u*)} = ax
Let ux
= cos vxi and we have ax = cos v x+1 cos vx + sin vx+l = cos(vx+1 -^) = cosAv
sin v x
a; ,
118
EQUATIONS OF DIFFERENCES.
finally
[CH. VII.
whence
ux
= cos C + 2
(
cos"
ax ) .
The method by which the linear equation (7) was integrated in the particular case of cx = suggests the problem of determining a priori the forms of equations of differences of the higher orders which may be resolved into, or in some way made to depend upon, equations of the first order. The following will serve as a particular illustration.
Since the solution of the symbolical equation
2
(TT
+ a7T + b)ux = X
(9),
where a and h are constant coefficients, depends in general upon the interpretation of an expression of the form
let
us
u^N^ir-a^X+NiiTr-pFX (10), give to ir the form D + (x) such being a form which
;
renders (10) interpretable by the solution of linear equations Then (9) becomes of differences of the first order.
[{D+<f>(x)Y + a{D
or, effecting
+ <l>(x)} + b]ux = X,
Dux
2
,
D ux byu
2
x+1
,ux+2
+2
+K+
+ [W>H} +
<><t>
0*)
+ &] *.-X
The solution of this equation then, whatever form we assign to <j> will depend upon that of equations of differences of the first order.
9
There are, further, various cases in which the solution of equations of differences may be effected by a process of sucshall discuss this subject in Chap. IX. cessive reduction.
We
ART.
tions
10.]
EQUATIONS OF DIFFERENCES.
119
may
*
be extended to linear
One
among
the least
we
Theorem.
We
equation of differences
=X
(1),
it
were
+
;
(2),
and
let
ux
vx x
t
becomes
-t>
Vx+ntz+n
+ ^xVx+n-i^x+n-i +
x
xV x+n-2^x+n-2
~f~ <
^C
= **=X
1
,
Or
vx+n I)\
. . .
D"' &c. in Eeplacing by 1 + A, and developing ascending powers of A, arrange the result according to ascending differences of tx There will ensue
,
Dn
Vx+n
...) tx
+ PAtx .+ QA%...+Z&%*=X.
being, like the coefficient of P, Q, ... &c. and of the original coefficients x
,
(3),
A Bx
by
x
tx ,
,
&c.
Atx
of
vanishes
(2),
whence, making
1 th
tx
= Xwx
/.
ux = vx twx
(4).
120
It
EQUATIONS OF DIFFERENCES.
hence follows that the linear equation
[CH. VII.
ux+2 + A x u x+1
+ Bxux = X.
if
(5)
we know a
particular in-
ux+2 + A x u x+i
+ Bxux =
(6).
To deduce the form of the complete integral, let, as before, vx be the particular value of u x which satisfies (6), and let ua = vx tx ; then, proceeding as above, we get
vx+2
therefore putting
A% + (2^ + A v
+2
x x+1 )
= X; =X
vx+
Aw
2V *+2
-40 w
>
'
Wx
V~ x+2
;
(6),
for
by
virtue of
.Ay1= B v
Bxvx
x x
Hence
Ultimately therefore
we have
"- P:
each summation introducing an arbitrary constant.
If
we make
7
B
V
~ m
ART. 11.]
EQUATIONS OF DIFFERENCES.
121
may
If
X=
we have simply
H,-.. (0+0, ST.).
equations
Equations of differences are connected with differenby more than mere analogy. If in an equation
d_
of differences symbolically expressed we substitute e for D, purpose here to init becomes a differential equation. quire to what form of the general solution of a linear equation of differences, with constant coefficients and with vanishing second member, this transformation leads.
dx
We
M (e"-arO
r
(1),
a, r,
and
know from the theory of differential equain Art. 6. tions that the complete interpretation of this term will consist
of terms of the form
mx
{A+Bx + Cx\..+Ex r
(2),
-a = Q
.(3),
and A, B,
C,
...
different values of
differ for
Now
the roots of
0.
122
EQUATIONS OF DIFFERENCES.
0,
[CH. VII.
Let i =
then
m = log a,
x
and
(2)
gives
r - ls
a {A
+ Bx+Cx\..+Ex
result in Art. 6.
),
Again, assigning to i any two corresponding positive and negative values, (2) assumes the form
8 and T being
A + Bx+Cx\..+Ex -\
r
But
(4)
may be
reduced to the
[8 [cos
2iirx
+ V( 1)
sin 2t7rx]
replacing
still
nomials
1 th
T) V(-
1),
by
M and N,
(poly-
efficients),
we have
x a (if cos 2vkx
+ iVsin 2iirx),
Hence, giving to i the successive values 0, 1, 2, &c, it is seen that the complete solution, so far as it depends on the exr pression {D a)~ 0, will be of the form
a (P1
+ P x + P x\.. + P x^)
2 3
(5),
where each
.
coefficient
P is
x
of the form
(B \B2
l
cos 2ttx
sin 2-irx
+G + C
cos iirx
sin krrx
' '
^ "
&c, being
arbitrary constants.
Analytically, (6) is the general expression for a periodical function of x constant for values of x differing by unity. And this confirms the a priori determination of the significance of the constants in the solution of equations of differences. Art. 3. See also Chap. IV. Art. 1.
EXERCISES.
123
it
is
due
to the
imaginary constituents of
the expression e a that the constants in the solution of an equation of differences are such only relatively and not absolutely, their true character being periodical.
EXERCISES.
1. Find the equations of differences to which the following complete primitives belong.
1st.
u=cx +
2
c\
2nd. u
\c (- \)
2
x)
-V
x --.
3rd.
u = ex + cax
w
(A/
4th.
u
""
= cax + c
a
/.
c th Olilt
_ O r
Z1 T^ O 1,
.
x
....
<A
11
\x
LI )
2.
kux 0, = mx
.
find
ux
3.
4.
5.
6. 7.
8.
ux+l pa ux = qaf.
x.
{sin
(fra;)}
ux 3
sin {Trx)ux_ x
V
PA X
x
0.
The
ux
A xux_ Bxy
by
.
be-
sides
Integrate ux+1 ux + (x
2) i^+1
+ ccw^. = 2 2# x\
In Art.
9,
124
first
EXERCISES.
[CH. VII.
order is made to depend upon that of a linear equation of the second order whose second member is by assuming
ux =
-^-a
And
it is remarked that the two constants which appear in the value of v x effectively produce only one in that of ux
.
Prove
13.
this.
The
equation
W*+2 * ( a*+1
Ux+i
+ ux =
first
may
order.
14.
Given that a
ux+2 - a
(<f
1)
u x+1
+ az+1 ux =
is
ux
= ca
2
,
The above equation may be solved without the pre15. vious knowledge of a particular integral.
16.
The
equation
UxU^Ux+t = a
(.
+ ux+1 + uxJ
.
may
Shew also that the general integral of the above equa17. tion is included in that of the equation ux+3 ux = Q, and hence deduce the former.
18.
Shew how
125
CHAPTER
VIII.
OF EQUATIONS OF DIFFERENCES OF THE FIRST ORDER, BUT NOT OF THE FIRST DEGREE.
The theory of equations of differences which are of a 1. degree higher than the first differs much from that of the corresponding class of differential equations, but it throws upon the latter so remarkable a light that for this end alone shall endeait would be deserving of attentive study. vour to keep the connexion of the two subjects in view
We
differences
of the
first
order
+ P (Au) n
1
-i
(1),
PP
1
... P n and Q being functions of the variables x and 2 and then by algebraic solution reducing it to the form
u,
(2),
component equations,
(3),
will be a complete primitive of the given equation (1), i.e. a And thus far there solution involving an arbitrary constant.
is
complete analogy with differential equations (Diff. EquaChap. vii. Art. 1). But here a first point of difference arises. The complete primitives of a differential equation of the first order, obtained by resolution of the equation with
tions,
dy
respect to
equations,
may
without loss of generality be replaced by a single complete (lb. Art. 3). primitive, Beferring to the demonstration of
126
EQUATIONS OF DIFFERENCES
[CH. VIII.
this, the reader will see that it depends mainly upon the fact that the differential coefficient with respect to x of any function of v ... n variables supposed dependent on x, will be 2 linear with respect to the differential coefficients of these de-
V V
pendent variables
But
remain
if
the operation
is
and
therefore the different complete primitives of an equation of differences cannot be replaced by a single complete primitive.
On the contrary, it may be shewn that out of the complete primitives corresponding to the component equations into which the given equation of differences is supposed to be resolvable, an infinite number of other complete primitives may be evolved corresponding, not to particular component equations, but to a system of such components succeeding each other according to a determinate law of alternation as the independent variable x passes through its successive values.
Ex.
1.
Thus suppose
(Aufwhich
is
+ x) Au + ax =
(4),
Au- a = 0, Au-x=0
And
suppose
it
(5).
required to obtain a complete primitive which by satisfying the first of the component equations (5) when x is an even integer and the second when x is an odd integer.
shall satisfy the given equation (4)
The
and
to
x when x
is
even,
1)*
'-
+x
(- \y 'i
+ (-!>*?.
the solution of which
is
ax
x(x
^) + (-ir(^-i) +
1)
c,
ART.
1.]
127
and
it will be found that this value of u satisfies the given equation in the manner prescribed. Moreover, it is a com-
plete primitive.
To extend
tion
and
as
Au
this method of solution to any proposed equaany proposed case, it is only necessary to express a linear function of the particular values which it is
to
intended that it should receive, each such value plied by a coefficient which has the property equal to unity for the values of x for which that for all other the equivalent of Aw, and to forms of the coefficients may be determined by
proposition.
being multiof
becoming
term becomes
values. The the following
Prop.
If
a, /S,
...
x being an
is
equal to
is
equal to
For
1,
fju,
if
yLt=cos
fju
f-V(
1) sin
the n roots
will
be
fj? ...
n ~\
Therefore,
7*...
x x _ 1 + fi + fj? ... +
{n ~ 1)x
fjL
a*+ ffMn
~n
1 fi
nx
fju
-I x 1
(6).
Now if x be equal to n, or to a multiple of n, the above becomes a vanishing fraction whose value determined in the usual way is unity. If x be not a multiple of n, then since n fL = 1, the numerator vanishes while the denominator does not, and the fraction is therefore equal to 0.
Hence,
if it
Au
as shall
be required to form such an expression for assume the particular values p v p 2 ... jpn in suc-
x=
x=2
...
x=
n,
again,
on,
for the
inf.,
it
2n,
2
and so
ad
assume
Au =
where
(7),
(8)>
128
a, /3,
EQUATIONS OF DIFFERENCES
[CH. VIII.
...
equation
It will
being as above the different n th roots of unity. (7) must then be integrated.
The
be observed that the same values of Au may recur Further illustration than is afforded by Ex. 1, is not needed. Indeed, what is of chief importance to be noted is not the method of solution, which might be varied, but the nature of the connexion of the derived complete primitives with the complete primitives of the component equations into which the given equation of differences is resolvable. It is seen that any one of those derived primitives would geometrically form a sort of connecting envelope of the loci of what may be termed its component primitives, i. e. the complete primitives of the component equations of the given
in
any
order.
equation of differences.
If
x be the
abscissa,
on a plane referred to rectangular axes, then any particular primitive of an equation of differences represents a system of such points, and a complete primitive represents an infinite number of such systems. Now let two consecutive points in any system be said to constitute an element of that system, then it is seen that the successive elements of any one of these systems of points representing the locus of a derived primitive (according to the definitions implied above) will be taken in a determinate cyclical order from the elements of systems corresponding to what we have termed its component
primitives.
It is possible also to deduce new complete primitives 2. from a single complete primitive, provided that in the latter the expression for u be of a higher degree than the first with
The method which conrespect to the arbitrary constant. sists in treating the constant as a variable parameter, and which leads to results of great interest from their connexion
with the theory of
differential equations, will
be exemplified
A given
the
first
V=f{x,c)
(1)>
ART.
2.]
129
but under the condition that Au shall admit of the in terms of x and c as if c were a constant. It is evident that if the value of c determined by this condition as a function of x be substituted in the given primitive (9) we shall obtain a new solution of the given equation of difThe process is analogous to that by which from ferences. the complete primitive of a differential equation we deduce the singular solution, but it differs as to the character of the
let c vary,
same expression
result.
The
solutions at
solutions, but
c is
tial
new complete
made
equations, to an algebraic equation for its discovery, but to an equation of differences the solution of which introduces
is
integral.
2.
The
equation u
= xAu +
= cx+ c
2
primitive
An
indirect integral
is
required.
Taking the
difference
c is constant,
we have
Au = c
and taking the difference of
(2)
unknown
function of x,
c is
an
Au = c +
(x
1)
Ac +
2c Ac
(Ac)
2
.
Ac(a?+l+2c + Ac)
Of Ac = 0,
the
to
=0
1
(3).
Ac + 2c + x +
= 0,
first determines c as an arbitrary constant, and leads back the given primitive (2); the second gives, on integra-
tion,
c
B. F. D.
=c(-ir-f-i
4 ),
130
EQUATIONS OF DIFFERENCES
an arbitrary constant, and this value of on reduction
[CH. VIII.
c substituted
C being
*=\0(-1)'-\\'-t
Now
this is
(5).
integral. see that the prindetermination rests is that upon which rests the deduction of the singular solutions of differential equations from their complete primitives. But in form the result is itself a complete primitive; and the reader will easily verify that it satisfies the given equation of differences without any particular determination of the constant C.
ciple
an indirect
its
We
on which
Again, as by the method of Art. 1 we can deduce from an infinite number of complete primitives determining c, we can, by the substitution of their values in (2) deduce an
(3)
infinite
number
ences given.
It is proper to observe that indirect integrals may be deduced from the equation of differences (provided that we can effect the requisite integrations) without the prior knowledge
of a complete primitive.
Ex.
3.
Thus assuming
u=xAu +
and taking the
(Au)
(6),
we have
;
.*.
which
is
resolvable into
AV = 0, A u + 2Au+x+l =
9
(7).
The former
Au = c,
and leads on substitution
plete primitive (2).
in the given equation to the
com-
The second
equation of
(7)
AM= c(-i)--|-i,
ART.
3.]
131
and substituting
we have on
reduction
^-D'-i}'-?'
which agrees with
(5).
The process by which from a given complete primitive we deduce an indirect integral admits of geometrical interpretation.
u f (x, c) may system of points situated in a plane and referred to rectangular co-ordinates the changing of c into c + Ac then represents a transition from one such system to another. If such change leave unchanged the values of u and of Au corresponding to a particular value of x, it indicates that there are two consecutive points, i. e. an element (Art. 1) of the system represented by u =f(x, c), the
be understood
to represent a
position of
which the transition does not affect. And the successive change of c as a function of x ever satisfying this condition indicates that each system of points formed in succession has one element
common with the system by which was preceded, and the next element common with the system by which it is followed. The system of points formed
it
elements is the so-called indithus seen to be a connecting envelope of the different systems of points represented by the given complete primitive. The difference between this case and the one considered before is that here all the elements of all possible indirect integrals are, virtually, contained in the one complete primitive given.
of these consecutive
rect integral,
common
which
is
It evidently might happen that both methods admitted of application in the same problem.
3.
Of
the different questions which the above theory sugperhaps the most important.
indirect integral being itself a complete primiit
An
what
the process of
2ndly. An indirect integral represents a system of the loci represented by the complete primitive from which it was
92
132
derived,
EQUATIONS OF DIFFERENCES
[CH. VIII.
and it is itself a complete primitive. In differential equations envelopes are represented usually by singular soluHow is this tions, and occasionally by particular primitives.
to be
explained?
to these questions are contained in the state-
The answers
ments of two laws, viz. the Law of Reciprocity, connecting the integrals of an equation of differences and the Law of Continuity, governing the transition from the integrals of
;
Law
of Reciprocity.
complete primitive of an equation of differences Prop. with the indirect integrals declucible from that primitive by
the variation of its arbitrary constant will together constitute a cycle of complete primitives, such that from any one of them all the others may be deduced by the variation of its
arbitrary constant.
=/(v)
Then
(1).
Au=f(x + l,c)-f(x,c)
and the elimination of
equation of differences,
c
(2),
to the
F(x,
Au)=0
(3).
Again, taking the difference of (1) on the assumption that is an unknown function of x, we have
Au =f (x +
1, c
in (2) on
f(x+l,
+ Ac)-f(x-t
(4),
an equation of differences for determining c. It is satisfied by the assumption Ac = 0; but this, determining c as a constant, only leads us back to the given complete primitive. Virtually (4) is reducible to the form
Ac(f>(x,
c,
Ac)=0
(5),
ART.
3.]
133
Ac =
and
it
0,
(x, c,
Ac)
(6),
the values of c which satisfy the second of these according to the theory developed in Art. 1, satisfy the two equations under some law of alternation, that lead on substitution in (1) to indirect integrals.
is
equations, or which,
If
D = + A,
1
the equation
(4)
may
1, e)
be expressed in the
form
f(x+ 1,
said,
Dc)
-f(x +
(7).
Let be
its integrals, in
c
a,
= a, = fa
{x,
x)
fa {x, a 2 )
&c.
a v a2
...
Then
are
u=f{x,
it
fa(x, a t )},
u=f\x, fa(x,a
&c
(9).
But, since c = cf) 1 (x, aj is by hypothesis an integral of must, together with the value of Be which it gives, viz.
(7),
Bc = fa(x+1, a ),
t
reduce
(7) to
an identity.
9
Substituting in
i
(7)
we have
fa(x ai )}^0...{10).
9
= <
(x,
a2)
we have
f{x+l, fa(x+l,
and so
by
tlie
a 2 )}-f{x
+ l,
fa(x, 2 )}
= 0...(11),
on.
Thus
of
(3)
<f> 2
,
ject to
system
Now
let
us assume the
first
u=f{x,
as a given complete primitive,
fa (x, aj]
(12),
integrals.
134
EQUATIONS OF DIFFERENCES
of the equation
[CH. VIII.
The form
f{x + 1,
to the
ft (x
1,
Da,)}
-f[x +
1,
ft
(a*
1,
a t )}
0.
f{x+l,
or, since
(x+1, Da?j}-f{x +
ax )
l,
ft
(x,
<)}
= 0,
Dft
= ft
(a?
+ 1,
Da,),
ft
(a>,
f{x+l, Dft
(a, a,)}
-/{x + 1,
aj}
...
(13);
an equation which is of the same form with respect to the function ft (x, aj as the equation (7) is with respect to c. Hence by (8) it will admit of the system of integrals,
ft(a>,fl0=^l,
ft (x, a x )
)
ft
(a?,
A)
ft
(*>
ai)
=cf> 2
{x,A 2 ),&c.y" K
14
n
'
^4, ^4 1? ^4 2 ...
being arbitrary constants. These integrals determine the values of a v which must be substituted in the complete primitive (12) in order to obtain the system of its derived primitives.
The
form
u=f(x,
which
is
A),
(1).
The
first
integral in the second line of (14) gives at to the assumed primitive (12).
Av
The second
to the
form
u=f{x,
(9).
ft
(a,
2 )},
And
(9)
system
indirect integral in the system the forms being general, every integral in the can thus be derived from the first of the series.
by a formation, both to the complete primitive from which it was derived, and to all other complete primitives having the same origin. The ensee therefore that
indirect integral leads us,
We
any
own
ART.
tire
3.]
135
system is seen to constitute a cycle. Theoretically, that cycle will be infinite, but practically, owing to the limitation of our powers of integration, it may be finite.
Ex.
for a
4.
The equation
of differences,
2
,
=xAu+ (Aw)
(2,
2
,
has
and hence
Ex.
2)
was
shew
u
that,
z
assuming
= cx + c
(a),
is
con-
AM =C(-1)-|_I
Taking again the
(b).
C is
variable,
we
A =C(-i r _|_I
tt
+2CAa(-ir+(AO) (-ir+^(-ir
2
w,
(b) if
we have
A (7=0
is
c (-
iy x
+ &c{-iy x + l{-iy=o;
136
or dividing
EQUATIONS OF DIFFERENCES
[CH. VIII.
by (
l)
2x
,
and transposing,
A(7 + 2C=-|(-1)-,
the integral of which
is
c-fo+fK-ir.
Substituting this in the given primitive,
we
get
1\
X*
-l) +
-l)~
>
or,
replacing c
- by
c,
is
arbitrary,
= ex + c
2
,
Ex.
5.
(1
+ u,/)
5
is satisfied
by
(a);
u x = (x + c)
By
(4),
Art.
3,
is
(x
+ Dcf-
{x
+ 1 + c) = 0,
3
which
is
resolvable into
Ac =
Dc-fjic=
Z>c- z^c=
/a
(5),
(aJ
{/ju-1)
(v
1)
(c),
-1)
(a?
+ l)
(?),
and
z>
0,
imaginary cube roots of unity. The equation (h) leads back to the given primitive, while (c) and (d) give on intei.
e.
gration
ART.
4.]
137
=x
j^l + w.
v
c= x
whence, substituting in
(a),
1 +c
X i/;
we have
s
(^-^i)
/-<
w-
(v*-^)
and these with
It will
(a)
be found that either of the equations (e), (/), asas complete primitive, leads to the other and to (a) as indirect integrals.
sumed
We
138
EQUATIONS OF DIFFERENCES
[CH. VIII.
by an equation of differences, and the corresponding law of the ordinates of the limiting curve to be expressed by a differential equation.
Now
polygon.
there
plete sense in
is a more complete and there is a less comwhich a curve may be said to be the limit of a
In the more complete sense not only does every angular point in the perimeter of the polygon approach in the transition to the limit indefinitely near to the curve, but every side of the polygon tends also indefinitely to coincidence with the curve. In virtue of this latter condition the value of
A?/
-~
in
Ax
is
diminished to that of
Oil
-4-
ax
in the curve.
if
It is evident
realized
the angles of the polygon in its state of transition and tend to it as their limit.
to be alternately salient and re-enwhile the sides of the polygon are indefinitely diminished, to continue to be such without tending to any
trant, and,
limit in
cease.
Here
evident that while every point in the circumference of the polygon approaches indefinitely to the curve, its linear elements do not tend to coincidence of direction with the
it is
curve.
limit to
An
polygon
first
If then the solutions of an equation of differences of the order be represented by geometrical loci, and if, as Ax
0,
these loci tend, some after the first, some above modes to continuous curves then such of those curves as have resulted from the former process and are limits of their generating polygons in respect of the ultimate direction of the linear elements as well as position of their extreme points, will alone represent the solutions of the differential equations into which the equation This is the geometrical of differences will have merged.
approaches to
ART.
4.]
139
The
y x+2h be
As-
suming h
as the indeterminate increment of x, let yx yx+M the ordinates of three consecutive points of the
polygon, let (f> be the angle which the straight line joining the first and second of these points makes with the axis of x, ^r the corresponding angle for the second and third of the points, and let i|r (, or 0, be called the angle of contingence of these sides.
Now,
tan
6=
tanV = y***
Vx+h
7Jx+h
7
Vx+ih
_ Vx+h
li
l/x
tan 6
h
i
_i_
i/ x + h
y* y^+^h
h
yx+h
h
have,
Now,
since h
= Ax, we
yx +, h
-^yx+h +yx = ^ yx
2
2
yx +2k-yx + A = tyx + A yx
Therefore replacing
yx by
y,
A tan#
Ax \Ax)
(1).
Ax Ax
Now the principle of continuity demands that in order that the solution of an equation of differences of the first order may merge into a solution of the limiting differential equation, the value which it gives to the above expression for tan 6 should, as Ax approaches to 0, tend to become
140
infinitesimal
;
EQUATIONS OF DIFFERENCES
[CH. VIII.
since in any continuous curve or continuous Again, that the portion of a curve tan 6 is infinitesimal. above expression for tan 6 should become infinitesimal, it is
clearly necessary
and
sufficient that
so.
Supapplication of this principle is obvious. we are in possession of any of the complete primitives of an equation of differences in which Ax is indeterminate, then if in one of those primitives, the value of
5.
The
posing that
Ax
-~
tends, inde-
pendently of the value of the arbitrary constant c, to become infinitesimal also, the complete primitive merges into a complete primitive of the limiting differential equation
;
but
y if --^LaX
tend to become infinitesimal with Ax only for a particular value of c, then only the particular integral corresponding to that value merges into a solution of the differential equation.
have seen that when an equation of differences of order has two complete primitives standing in the mutual relation of direct and indirect integrals, each of them represents in geometry a system of envelopes to the loci represented by the other. Now suppose that one of these primitives should, according to the above process, merge into a complete primitive of the limiting differential equation, while the other furnishes only a particular solution then the latter, not being included in the complete primitive of the differential equation, will be a singular solution, and retaining in the limit its geometrical character, will be a singular solution of the envelope species. Hence, the remarkable conclusion that those singular solutions of differential equations which are of the envelope species, originate from
the
first
;
We
particular primitives of equations of differences their isolation being due to the circumstance that the associated particular
;
primitives of the equation of differences, not possessing that character which is required by the principle of continuity, are unrepresented in the solutions of the differential equation.
In the following examples we shall confine ourselves to those indirect integrals which arise from the supposition that
ART.
5.]
141
is wholly rejected in the process of the equation Ac = The derivation from the given complete primitive, Art. 2. other indirect integrals will, as is obvious from the mode of their formation, generally be of the purely discontinuous
species.
Ex.
its
6.
The
differential equation
y = x-^
(-^-\
has for
complete primitive
y cx
and
for its singular solution,
+c
(a),
is
which
y==f
It is
(*)
required to trace these back to their origin in the of differences. 1st, Taking the difference of the complete primitive, Ax being indeterminate and c a mere constant, we have
solution of an equation
Ay c Ax.
Hence
j
= -^-, -r^~,
Ay
we have
y ~ X Ax
This
is
+(&' \Ax.
'
w-
Taking the
but
c
Ax
being
still
indeterminate
2
,
we have
(Ac)
which, equated with the previous value of Ay, gives on dividing by Ac Ac-f 2c = - (x + Ax),
an equation of differences for determining
c.
To solve the equation, it is desirable to reduce it to the ordinary form in which the increment of the independent variable is unity. Let then Ax = h, and let x = ht then
}
Ax = hAt;
.-.
A*=l.
142
EQUATIONS OF DIFFERENCES
[CH. VIII.
And we have
Ac +
the integral of which
is
2c
= -h
(t+1),
(-ir-s-l.
whence, substituting in
(a),
=a
It results
(d),
ha (-if
2-
x2
n~I
(a),
W(a)
then that (c) has for complete primitives h being equal to Ax.
and
2ndly.
To
we have
Ay = cAx, A y = 0,
2
whence, substituting in (1), we find tan = 0. Thus the complete primitive (a) merges without limitation into a complete primitive of the differential equation.
(d),
2
,
we have
'
2xh
+ -
Hence
A Ax
2
?/
v
'
h
2
Now
= 0,
x2
and
this agrees
with
(b).
ART.
5.]
143
Thus while the complete primitive of the differential equation comes without any limitation of the arbitrary constant from the first complete primitive of the equation of differences, the singular solution of the differential equation is only the limiting form of a particular primitive included under the second of the complete primitives (d) of the equaGeometrically, that complete primitive tion of differences. represents a system of waving or zigzag lines, each of which perpetually crosses and recrosses some one of the system of parabolas represented by the equation
As
li tends to 0, those lines deviate to less and less distances on either side from the curves but only one of these tends to ultimate coincidence with its limiting parabola.
;
When the given complete primitive of the differential equation is homogeneous with respect to x, y, and any constant other than of integration, it suffices to form the equation of differences on the hypothesis that Aa;=l, and examine the form which its solution assumes when the above quantities tend to become infinite, still retaining a finite ratio to each other. The following somewhat difficult problem illustrates this, and at the same time affords a valuable exercise in the treatment of functions of large numbers.
Ex.
7.
The
differential equation
dy dx
has for
its
m
(
dy
dx
complete primitive
y = cx +
c
(a),
and
= 4zinx.
(a)
is
The complete
to y, x,
primitive
it is
and m, which
therefore permitted to
make
infinite
144
EQUATIONS OF DIFFERENCES
[CH. VIII.
Eegarding together, while c remains finite. are led to the equation of differences
771
Ax
as
1,
we
To complete
have from
(a)
we
in the usual
way
x+
1
c(c
+ Ac)
To
member, and,
putting log c
= vx we
,
have
v x+l
+ vx = log
m
x+l
whence
r,
^Rr^'sTi+M'
Hence
lo
1~
g x~^l
g x~^~2 ~'
8j+(- 1
lo S
>
x be odd, negative
if
be even.
Hence, according as x
is
odd or even,
,
^=
C=
2.4
S
...
^TiT^>
2.4
...
(x-1) am
~
1.3... s
(a),
.
1.3...(a;-l)
(2.i...x)a
(*>
Whence, by
according as
(x
is
odd or even,
1
.
_
J
. . .
- 1 $m
)
.3
...
(a5-2)
2.4
...
x (x-1) a
. . .
'
ART.
5.]
145
tion of x,
for
y regarded
as a func-
^~2.4
1)
(2.
4. ..2a?)
am
...
(2a;
-2) a
_ ^2x+l ~1.3...(2x-l) +
(2. 4. ..2a;)
am
1)
(2.4...
2a;)
'
y*wNow
the
if
_ 1.3...
(2a;
+ 1)
(2.4... 2a?)a
2.4...
1
(2a;
+ 2) am
'
.3
...
(2a?+l)
we suppose x
illustrated in
/2a;*
large
method
Ex.
\ 1
5,
7r\
5tt*\
Hence
^tef i
V*
rf^
Tr
l
>
ma
'
y2z+2
Therefore
y^x+i
2xh
AV-=~ y^x+2 ~~
nr
l
%
,
2, + i
+ y^
(d).
ma
jr
i
2xh
^
xi a
Before
iion
;o
we can deduce any conclusions from this expreswe must determine of what order of magnitude a is
10
be considered.
B. F. D.
146
EQUATIONS OF DIFFERENCES
general
[CH. VIII.
The
expression for
y2x
in
the
new
primitive is
deduced from that of y in the original complete primitive by changing x into 2x, and making
1.3...
(2^-1)
2a?)
(2.4. ..
Therefore
1
a=
.3
...
(2sc-l)
. . .
(2.4
1
2x) c
(2ttx)
'
x tending
to infinity.
Hence, since
as r\j
.
c is finite,
is
of the
But
m is
of the
same order
Hence the terms of the right-hand member of (d) are of such an order as to be finite, and the condition that that member must itself be infinitesimal gives
it^ma
1
whence
Substituting this in
/(
\jnirj
(c)
^ = 2V(2ma;)
or
y, = 2VH;
for
whence, restoring y
yz
fanx.
2x+1
for
y2x
The following proposition, due to Poisson [Journal 6. de VEcole Poly technique, Tom. VI. p. 60), contains a theory of the singular solutions of equations of differences similar
ART.
6.]
147
to that of
for
differential
equations.
The
student
de-
must
monstration
Prop.
the form
order of
(i)
Ay-/foy)
be satisfied by y = u, and if on developing f(x, u + z) in ascending powers of z the index of z in the second term be less than unity, then y = u will be a singular solution. But if that index be equal to or greater than unity, y = u will be a
particular integral.
Poisson begins by laying down the hypothesis that if y = u be a particular integral obtained by making the arbitrary constant C = a in a complete primitive, that primitive will be expressible in the form
y^u+iC-aYX^iC-ayX^..,
a, /3, y ... being ascending, positive, and constant indices. He then proceeds to investigate a method by which the values of &c. can be these indices and the forms of the functions 2 t, determined, regarding the failure of such method as an indication that the supposed particular primitive is a singular
X X
solution.
(G
a)
by
c the
y = u + cX
a,
j3,
(2),
positive,
Hence, substituting in
2
or
assuming
(3).
Now
member
m
a
f(x, u)
+f
(x, u) z
+/ (*,
+ &C,
102
148
EQUATIONS OF DIFFEKENCES
[CH. VIII.
the indices m, n, &c. being positive constants, the values of which will be known from the constitution of the function f(x, u). Then replacing z by the series for which it stands, and observing that Au =f(x, u), (3), becomes
u)
{cX
+ cX
m
...r
2 ...)
+f (x,u)(cX + c*X
z
1
+ &C,
or,
member
u)
in ascending powers of
(x) c
c,
AX + cAX
t
. . .
=/ (x,
x
Xcm +
'
<f>
+ &c
(4)
Now
first let
m 1,
first
members, we have
AX^f {x,u)X^
x
This being an equation of differences for determining t found, we should have on equating the second terms of the
.
members
of
(4)
and so on.
let
Secondly,
be greater than
which determines
X
,
then making
AX = 0,
t
as a
constant,
and assuming a = w,
!
we
get
(a )I 4I,=/ 1
!l
which determines
lar integral
and so on.
Here then
can be completed.
be less than 1 the two members of (4) But, thirdly, if cannot be made to agree. The supposed primitive cannot be completed, and is shewn to be a singular solution.
Poisson illustrates the above theory in the equation
y_^jAyf
4
of
9
is
A?/
(a),
ART.
6.]
149
and
for
The
employment of Cardan's
rule for determining Ay as a function of x and y being comPoisson's rule at plicated, it is better to proceed as follows.
|(Ay)==c, orj^=0.
Applying
this to (a)
we
get
2
(4r(Ay)
-l = 0,
whence eliminating Ay by means of (a) we have the result in question. It must be ascertained by trial that it satisfies the
equation
If
(a).
deduce the above species of solutions we should have to investigate the singular solution of the equation for determining c so that the above process would still have to be employed.
to
we attempted
Poisson says nothing about the geometrical character of But it is clear from what has been shewn in these solutions. the foregoing articles that they are not in any peculiar sense the primal forms of those singular solutions of differential equations which are geometrically interpreted as envelopes. They are of extremely rare occurrence, and I should not have deemed it worth while to notice them but for the double object of directing attention to a subject which seems to need further investigation, and of shewing how in those cases in which Poisson's hypothesis is realized a particular solution of an equation of differences may be completed.
150
EXERCISES.
[CH. VIII.
EXEECISES.
1.
which
shall satisfy
it
by making Ay a
for even,
and Ay
The
equation
y~
is satisfied
2x+i
+ 2^+i;
2 2 y = cx + c
.
by
Shew
that
^ = f(-D"-f-T
may
thence be deduced.
3. Shew that a linear equation of differences admits of only one complete primitive. 4.
The
equation
\a
2
1/
Deduce another
5. In what sense are the complete primitives of an equation of differences of the first order said to be connected by a law of reciprocity ?
151
CHAPTER
IX.
The symbolical methods for the solution of differential 1. equations whether in finite terms or in series [Biff. Equations, Chap. XVII.) are equally applicable to the solution of equations Both classes of equations admit of the same of differences. symbolical form, the elementary symbols combining according And thus the only remaining to the same ultimate laws. difference is one of interpretation, and of processes founded It is that kind of difference which upon interpretation.
exists
and 2.
It has
if in
we assume x = e9
the equation
may
^ A^ A- ^
u
+ /
>
a),
U being
a function of
6.
-jp,
and
Kiy^Ki^i
And hence it has been shewn to be possible, 1st, to express the solution of (1) in series, 2ndly, to effect by general theorems the most important transformations upon which finite integration depends.
152
LINEAR EQUATIONS
- and
e
G
[CH. IX.
Now
-j-
and
x,
and
it
is
proposed to develope in this chapter the corresponding theory of equations of differences founded upon the analogous employ-
-r-
and
xD
supposing
Ax
arbitrary,
and
A<j> (x)
<j>(x
+ Ax)
(j)
(x),
D<j>{x)
= <j>(x + Ax).
ir
Prop.
1.
If the symbols
'
and p
P
7r
=X
Ax'
= xD
3 )>
/W? =//>"
the subject
'
of operation in
the
1st.
Let
Ax = r, and
pux =
'
first let
ofpXNow
xDux xux+r
pV = Pxu*+r =x(x + r) u
= xix -f r) = x(x + r)
=x(x+r)
. . .
x+2r
whence generally
m p ux
[x
+ (m - 1) r]
ux+mr
an
equation to
which we
may
...
p ux
If
{x+(m-l)r}Dmux
have
y
(5).
ux
= 1,
= 1, we
m
p
to
...
[x+(m-\)r}
which we
= x{x + l) ...{x+(m-l)r},
ART.
1.]
153
2ndly.
now
Uxi
p Ux
... IT
p Ux
Now
irp
ux
= xx{x + r) _ go
-
...
{x+ (m-1)
r]
ux+mr
+ (m-l)r}ux+mr
=x
...
{x
(to-
1) r]
xu-*'~^~ mr iw'
'
= pm xu^-{x-mr)uX
M{^
= pm (x W^~
U'
+ mu
x
!l
(x^ux +mu
= pm
Hence
(-n-
+ m)ux
m v*p u
and generally
n 7r
m p ux
= pm {7r + m) nux
m f{ir)p u
= pmf(7r + m)u
have
(6),
which
is
the
first
Again, supposing u
= 1 we
,
/Hp"i=p7(t+.)i
m
p
154
UNEAR EQUATIONS
2
[CH. IX.
Butwl=x-r-l = Ax
0,
tt 1
0,
&c.
Therefore
/Wp-i = P-/Hi.
Or, omitting but leaving understood the subject unity,
fMpm =fHpn
Prop.
2.
(7).
Adopting
the previous
definitions
of
it
and
p,
/.(*)
The above
cular value of
proposition
is
portance to
Ax, but the only cases which consider are those in which Ax =
of
any im1
and
Xu
Q
x +n
Here
sions of
it is
it
most and p.
+ X ux+n_ ...+Xnux = (j)(x) convenient to assume Ax = 1 in Now multiplying each side of x (x + 1) ... (x + n 1),
1
1
(9).
the expres(9)
by
(5)
xu*+x = P ux>
x(x +
l)
we
shall
x ) u* +
0i () pux
...
+
p
<f>
n (x)
ux
(j>
{x)... (10)
But
since
Ax= 1,
7T
= xA,
= xD = xA + x.
/3,
Hence
X
and therefore
O (
7T +
0i
(
x)
= 0o (~ * + P)
x ) = 0i (~
^ + P) &c
i
These must be expressed in ascending powers of p, regard being paid to the law expressed by the first equation of (4).
ART.
1.]
155
applica-
The
(Tr)
P+
F (w)^
a
f 'w ru +fa
where F^ir),
(11) '
(7r),
.1
The
(8).
X u +X u
x
l
x_ 1
(12).
1
Here
it is
pression of
it
assume
Ax =
(x
in the ex-
by x
(x
1) ...
n+1),
,
and ob-
x(x-l) u x_ = p
pux
u x &c,
(x)
u x + fa
is
(x)
...
fa
(x)
p ux
= X,
easily seen
we have
X
whence, developing the
= 7T + p,
by the
theo-
coefficients, if necessary,
rem
(Tr
+ p)=F (Tr)+F
,
(Tr)p
+ F,(Tr)^ + &c....(l3),
where as before
JW=JL.,W-iC1 (T-i),
we have
again on reduction an equation of the form
(8).
156
LINEAR EQUATIONS
[CH. IX.
2. It is not always necessary in applying the above methods of reduction to multiply the given equation by a
(x
+ 1)
...
(x
+ n
1),
or
(x
1)
...
[x
n + 1),
X X
,
It
may
...
or the requisite
may
Thus resuming
Xu + Xu
m
t
a_ i
...+Xmu
M=
(14),
assume u
We find
*=T^7x'
...
X v + X xv
as
x_1
...+Xnx(x-1)
77=33
(x- n +
1)
vx_ n =
0... (15).
Hence assuming
aP p=xD
'
where
Ax
we have
vx
X
and
it
(16),
p for
coefficients
by
(13).
3.
consists in
preliminary transformation which is often useful assuming ux = /j,x vx This converts the equation
.
X^ + X^
1
x_ 1
...+X
...
into
fXfi. + /x-'X^.,
A =0 XA =
_n . is
(17)
(18),
When
expressed
XA M + XA
tt
n-1
w...+Xww =
(19),
it
may
ART.
4.]
157
Theorem.
7T
=x
xD,
then
(x
(7T
1) ... (ir
n + 1)
u=x
+Ax)
...
(20).
we
therefore
whence
Now reversing the order of the factors tt, it 1 it n + 1 in the first member of (20) and applying the above theorem to each factor separately, we have
,
. . .
(tt
=p"0>-V)*.
r
v
'
Ax
Ax
Ax'
...
+ r) ... {# + (n - 1) r} D% whence (7T-n+l)(TT-n + 2)...'7ru = x (x + r)...{x+ (n- l)r} f^J which, since r = Aa?, agrees with (20).
But pn w = x
(x
w,
When
Aa?
= 1,
Tr{7r-l)...{7T-n
Hence, resuming
x (x+
(x
+ n
1),
158
LINEAR EQUATIONS
result of the
[CH. IX.
form
%(x)
7T
{tt
1) ... {it
n +
(x) tt
1)
u
... (tt
+ <k
It only
coefficients,
(tt - 1)
- n + 2) u + &c. = 0.
remains then to substitute x = tt + p, develope the and effect the proper reductions.
0, let
(22),
and assume u
the
first
= ^am p
a m p"'
m
.
equation of
{/
(4),
attending to
+f
x
a,p- ...+/
(4),
a,p"' )
= 0,
2
in
{/ ()
p" +/, (
+ 1) ,Pm+1 ...+/.(+ )
coefficient of
m+
"}
= o,
m
p
equated to
gives
(23)
f (m) am + (m)
.
This, then, is the relation connecting the successive values The lowest value of m, corresponding to which am is of a m arbitrary, will be determined by the equation
/oM = o,
and there will thus be as many values of u expressed in
as the equation has roots.
series
If in the expression of
since
it
and p we assume
Ax= 1,
then
= x{x+l)
... (<c
+ m-l)
(24),
ART.
5.]
159
the series %a m p will be expressed in ascending factorials of the above form. But if in expressing it and p we assume Ax = 1, then since
m p
= x{x-l)
...
(x-m+1)
(25),
Ex.
(x
1.
Given
(2x
a)ux
I)
u x_
(1
2
)
(x
1)
ux _ 2
it
=x
-r-
= xD, = 0,
where
Ax =
x
1,
we have
(x
a) ux
(2x
7r
a 1) +p
pux +
(1
<f
p ux
whence, substituting
reducing,
for x,
developing by
(11),
and
Tr(ir-a)ux -tfp*u x =0
(a).
Hence
u x = ta mp m
= and = a the initial values of am corresponding to being arbitrary, and the succeeding ones determined by the
law
m(m-a)
Thus we have
for the
a m -q*am _ 2 =0.
complete solution
2 (2
a)
(2
a)
(4
- a)
might be
be observed that the above equation of differences so prepared that the complete solution should admit For assuming ux /ixvx and of expression in finite series.
It
may
we find
160
fj?Tr{TT-a)v x
LINEAR EQUATIONS
[CH. IX.
{p?-p)
{2tt
-a-
1)
pvx
(c),
+ {(f*-iy-q*}p\ =
which becomes binomial
7r (7r
if /a
=1+
(27r
q,
thus giving
a) vx +
for either
a
/a,
1)
pvx
= 0.
Hence we have
value of
ux = p,x $ampm
= p,xtamx(x-l)
...
(aj-w +
all
1)
(d),
or a,
and
succeeding values
m
It follows
(m
a)
am
(2m
-a- 1) am_ =
x
(e).
from this that the series in which the initial value terminates when a is a positive odd number, and the is a terminates when a is series in which the initial value of a negative odd number. Inasmuch however as there are two values of jjl, either series, by giving to /jl both values in succession, puts us in possession of the complete integral.
of
m is
case in which a
is
a positive odd
+q
+
(1
a
2
)
(1- a)
(3
-a) of
(l+ ?
1.2(l-)(2-a)
^j
g
"*"
(l-q)(3- a)^'
1
.
(1
+ qf
(1
- a)
(3
- a) +
' }...(*).
The above
results
may
Differential Equations.
AKT.
6.]
161
The simplest case which presents itself is when the 6. symbolical equation (8) is monomial, i. e. of the form
/.to*-- *
(26).
We have thus
u
{fMPX
(27).
fraction, the
Eesolving then {fiir)}' 1 as if it were a rational algebraic complete value of u will be presented in a series of terms of the form (it- a)- X.
But by
(4)
we have
(tt--a)~*X=. p (try p-*X
a
(28).
It will suffice to
in the expression of
examine in it and p.
which
Ax = 1
To
member
...
of (28)
we have
<j)(x
then
<j>{x)
=x (x + 1)
(
(x
+ a 1)
+ a),
<PW
TT'^ix)
l)(x
i
+ 2)...{x + a)'
(xA)-
cf)(x)
= 2-2-...
X
,
</>(*); r '
2-
by x and subsequent
Should
express
it
it
suffices to
x(x +
l) t
x(x +
l)
(a?
+ 2),
&c.
B. F. D.
11
162
LINEAR EQUATIONS
2
[CH. IX.
s
,
(/oWrv =i/oW}>
=
that
l/o
m
(
MP *
i
+ 1)
... (a?
+ to - 1)
...
(29).
function
it is
0^p
1
if' 0.
Hence
in particular if
(ir
i 1, we
1
find
- a)""
+ l)
...
{x
+ a-1)
(30).
(x
+ 1)
...
(x
+ n-1) Anu + A x
x
(x
1) ...
... (a?
(31).
leads to the
(7T-71+
1)
+ A^ (TT - 1)
...
...
{7r-n + 2) ...+A n }u =
X
(x
(32).
Ex.
2.
Given
l)
x{x +
A u - 2xAu + 2u = x (x + 1)
2
+ 2).
of this equation
is
u-27ru + 2u = x(x+l)
2
(x
+ 2)
(a),
(tt
- 3tt +
2
2)
=p
1
5
.
Hence
^ + ^,
2
ART.
7.]
163
it
ir
3-7T + 2
are
ir
and
1. Thus
,,.
(
we have
t*
+ 2) = x(x + l)(x + o
^
~
(*
+ !) +
^
~
J )-
Binomial Equations.
7.
u+
in
(/>
(tt)
p w
*7
(33),
known, u the unknown and sought function depend upon the (ir) and its theory will consist of two form of the function parts, the first relating to the conditions under which the
which
Z7 is a
of x.
The
equation is directly resolvable into equations of the first order, the second to the laws of the transformations by which equations not obeying those conditions may when possible be reduced to equations obeying those conditions.
As
it
may
be observed that
if
the equa-
tion be
u+
it
^h>p
u~
w>
will, on reduction to the ordinary form, be integrable as an equation of the first order.
Again,
if in (33)
<j)
we have
. . .
(tt)
yfr (tt
n + 1),
in
which
ty
(tt)
+o
sl system of equations of the first order. the general theorem that the equation
+a
t (f>
(it)
pu
+ a$ (tt) + aw
(/>
cf>
(tt
1)
p u
. .
- 1)
... <
(tt- n
+ 1) Pnu = U
112
164
LINEAR EQUATIONS
[CH. IX.
may
q
n ~1
cf)
(it)
pu
U,
+a
+a
n~2
2
...
+ aw =
0.
Upon the same principle of formal analogy the propositions upon which the transformation of differential equations depends ( i7>.pp. 408-9) might be adopted here with the mere substitution But we prefer to investigate what of 7r and p for D and e e may perhaps be considered as the most general forms of the theorems upon which these propositions rest.
.
From
we have
{l+<P(Tr)p"}u=U,
u
and
{l
+ 4>(Tr) Pr
U,
= F{<t>(ir)pn U
(35).
known
the unknown function u is to be determined from the function by the performance of a particular operation of which the general type is
Thus
Now
process into a
suppose the given equations transformed by some new but integrable binomial form,
v
+ yfr (ir) pn v =
V,
We
V={l+f{7T)pn }which
is
V,
a particular case of F{ty (tt) p ] V, supposing F(t) to denote a function developable by Maclaurin's theorem. It is
ART.
7.]
165
apparent therefore that the theory of this transformation must depend upon the theory of the connexion of the forms,
Let then the following inquiry be proposed. Given the forms of </> (tt) and ty (it), is it possible to determine an su h tnat we shall have generally operation x
^Wc"!xWJ=xW%Wp"}-S:
irrespectively of the form of
(36),
X?
to satisfy
Supposing F{t)
<t>
t,
we have
(*)
p"%
W x= X W f W
equation of
(4),
p"
X
p-x,
(37).
Hence by the
<},
first
()
x (* - ) p"X = t
w%w
r
to satisfy
we must have
fW.xW ^Wx('
ss
- B)i
x to -ijp$ *<*-*
or,
replacing P.
^^ X by X
w
1?
and therefore
X by |p ^U X
'
x ,
166
If for brevity
suffix
LINEAR EQUATIONS
[CH. IX.
we
represent
Pn ~X~ by
n
from
we have
X = P^(tt) p
n
(tt)
P~ X.
Hence
therefore
n
2
}
{</>
(tt)
X= P^
process,
P~l P^
(tt)
p^X
m
)
X=P[f
n
(tt)
m
}
P'
Supposing therefore
able
n
i^(^) to
}X= PF{f
P"X.
The symbols
it
and p combining
f(7r
in subjection
to
law
f(w)p'X=p
the
+ m)X,
equation
members of
the
following
are
symbolically
equivalent, viz.
F{<l>
w -?.$$' i+ W rt* J$
P'\
(^)-
A. From this theorem it follows, in particular, that we can always convert the equation
u4
into
</>
(tt)
p u
Z7,
+ t(7r)
^ = P ig^
B
by assuming u
= Pn ^j-^v.
ART.
8.]
167
+ ^(7T)pT
?7
whence
since
v
it
{l
+ f(7r)p
}- 1 V,
follows that
we must have
it
is
of course necessary
functions
<j>
(tt)
and
yjr (ir)
by
Pn
should be
finite.
The
(7T)
have been stated with sufficient fulness elsewhere, (Differential Equations, Chap. XVII. Art. 4).
B. The reader will easily demonstrate also the following theorem, viz: n m n p~ m X, F{cf> (tt) p ] p F{<f> (tt + m) p ]
X=
u -f
<f)
(it)
p u
U,
may be
converted into
+ $ (tt + m) pn v = p~ m U,
by assuming u = pm v.
These theorems are in the following sections applied to 8. the solution, or rather to the discovery of the conditions of finite solution, of certain classes of equations of considerable generality. In the first example the second member of the given equation is supposed to be any function of x. In the But the conditions of two others it is supposed to be 0. finite solution, if by this be meant the reduction of the discovery of the unknown quantity to the performance of a finite
168
LINEAR EQUATIONS
[CH. IX.
of the kind denoted by 2, will be the It is however to be as in the other. observed, that when the second member is 0, a finite integral may be frequently obtained by the process for solutions in series developed in Art. 5, while if the second member be X, it is almost always necessary to have recourse to the transformations of Art. 7.
+ b) ux +
first
{ex
+ e)
ux_ x + (fx
+ g) ux_ = X.
2
(a).
Consider
(ax
the equation
(ex
+ b) ux +
,
+ e) ux_
-\-f(x
1)
it
x_ 2
=X
(b).
+ b)vx + p,
(ex
+ e) vx_ +f(x-l)
x
vx_ 2 = /T
X+2
X.
which
it
= x -r
2
= xD,
in
Ax =
1,
then
2
p? (ax
+ bx)
vx
+ fi
it
(ex
whence, substituting
cients,
+p
x and developing
the coeffi-
we
2
find
2
/jl
(air
(c),
p,
so as to reduce this
Let
pu
+ cpu +f= 0,
(b
then making
2apu -\-c
= A,
a) pu + e = B,
ART.
8.]
169
we have
-afiTT (it
-J
vx
or
V
or,
a,
member
F to be any particular value of the second obtained by Art. 6, for it is not necessary at this stage to introduce an arbitrary constant,
supposing
B B A' A
h_
is
an integer. '43
w*+
afx
Tpwx =
(e),
(A).
B
vx =
P ^-wM
x
W^^ZV.
**A
(/).
we should assume
as the transformed
w* +
wh pw =w
*
W'
170
LINEAR EQUATIONS
[CH. IX.
K+-T
*.-*,
W+The value
stituted in
(e)
fm
ww =p
Ik
TT
+-
(h).
TT
+j
obtained from (/) or (h) is to be subwx then found by integration, and vx determined by (/) or (h). One arbitrary constant will be introduced in the integration for w x and the other will be due either to the previous process for determining x or to the subsequent one for determining vx
of or
x
(g),
Thus
ger,
which
-? is a positive inte-
we
should have
1
+)
/
|_
1 )...(
7r+1)
J"
0)
7r
+ -7
B\~
J
found to be
Substituting in
(e)
we
have,
ft
(ax+h)
wx +
and
wx
we have
for the
complete integral.
fi
(c)
(b a)
Jul
+ c = 0, + e = 0,
+ (b-a)c = 0.
ART.
8.]
171
Supposing
we obtain, on making fi =
JLCL
or,
member
by
yrA-F,
if
where
an equation which
is
integrable
and determine
first
and
lastly v x
by A,
To found upon these results the conditions of solution of the general equation (a), viz.
(ax
+ b)ux +
(ex
assume
Then
lax'-\- b
aJ^jtgt
(b) we see that it is only necessary in the expression of the conditions already deduced to change
,
.
,
b into b
a (1+ q) v J!
'
. ,
e into e
c(l v
+ ,y/ g)
.
.
172
LINEAR EQUATIONS
Solution of the above equation
[CH. IX.
when
X=
(a) in
by
definite
integrals.
9.
If representing ux
by u we express
- dx
the form
(ax +
or
-2-
b)u+(cx + e)e
dx dx
)
u+(fx+g)e
dx
= 0,
u=
x
its
(a
+ ce
+fe~*
u+{b + ee~ dx + ge
may
dx
)
0,
method
xc
is
of the form
u
the limits equation
=G
<p(t)
rr;
4>(t)
dt,
any
roots
of the
<t>(t)
=0.
The above
solution
xt
is
u=ffF(t)dt,
which
from the above only in that logt takes the and of course leads to equivalent results (Theorie Analytique des Probability, pp. 121, 135). And he employs this method with a view not so much to the solution of diffidiffers
t
place of
cult equations as to the expression of solutions in forms convenient for calculation when functions of large numbers are
involved.
ART.
9.]
173
Thus taking
**-(
and assuming ux = ffF (t)
x+l
+ 1)^ = 0,
we have
l)ft F(t)
x
dt,
ft
F(t)
dt-(x +
{t)
dt=0
1)
(i).
But
(x
1) ft?
dt
= fF(t) = F(t)t
(x
x+1
f dt
+1
-ff F'
(t)dt.
So that
(i)
becomes on substitution
ff
+1
{F(t)
= 0,
F'(t)+F(t)
which gives
F(t)
= 0>
Q,
F{t)f +1 =
the
first
of
Ce->,
to the
form
and
= co
x+
1 is positive.
Thus we have
finally
= C
fV fdL
J
peculiar method the well known expression for V (x + 1). of integration is then applied to convert the above definite integral into a rapidly convergent series.
+bx + c)ux +
,
(ex
(a).
Let u x =
(ax
2
1 .2
Vx
...
x'
then
xvx_ x
+gx(x-l)
vx_ 2
= 0.
174
LINEAR EQUATIONS
it
[CH. IX.
Whence, assuming
have
/a
2
= x
= #Z>,
where
Ax = 1, we
(aar
+505
+ 0)^ + ^
7r
(ex for
Therefore substituting
/j?
+p
and developing by
(b
(13),
(air
+ b ir + c) + fi
{(2a/j,
+ e) ir +
2
a)
jjl
+/}
/owx
+
First, let
//,
(A + ^e+^)p ^=0
+ g = 0,
ir
(J).
be determined so as
afjf+efjL
then
fi (air
2
+ lnr + c)vx +
5,
{(2a^
+ e)
(b
- a) fi +/}
pi? x
= 0.
Whence, by Art.
vx
=Sam x (x1)
2
{
...
(x m+1),
(am
+ bm + c) a m +
am=Z
(2afi
fi
+ffj] am_ 1
= 0,
(2afju
e)
m + (b a) + f a
+ bm + c)
/jL(am
~ ^'
and
let
am2 + bm + c =
be a and
Vx
j3,
then,
1) tf
(a+1)
=C {ajW -f(a +
=x
(x
1)
...
(x
p + 1).
ART. 11.]
175
One
+ e)m+(b-a)
2
fi
+/= 0,
exceeds
by an
am + bm + c =
The
solution
0.
may then
be completed as in the
last
example.
Secondly, let //, be determined if possible so as to cause the second term of (b) to vanish. This gives
2a fi
+ e = 0,
whence, eliminating
/ju,
2af+ (a-b)e = 0.
This being
satisfied,
/ju
assumed equal
to
(b)
becomes
2
(7T
+ l-K + C)VX J=
e'- 4
^) pV, = 0.
Or putting
^M,
e
j
2
7T -f
IT
+a
P\=0,
and
is
differ
by an odd number.
Discussion of the equation
(ax
2
+ bx + c) A wx +
2
(esc
+/)
A^ + #wx = 0.
equation
is
reduci-
a)
(x
.... (a).
176
LINEAR EQUATIONS
let
[CH. IX.
Now
a (x
x a = x + 1 and ux = vx
(x
-,
then
we have
Aty +#ty
+ 1)
+ a - fi +
1)
A ty + 6 (x + a - 7 +
2
1)
= 0,
or,
a (x
+ 1)
(x
+ a-0+1) A\ + e (x + a-y +
1)
If from the solution of this equation v x be obtained, the value of u x will thence be deduced by merely changing x
into
x a 1.
Now
multiply
(h)
by
x,
and assume
A#
where
r
(20),
'
A# =
1.
Then, since by
X (x+1)
A\ =
vx
7T (lT
1) 1^,
we have
a (x + a /5 +
1) tt (tt
1)
+ e (a + a - 7 +
But x
1) 7T ty
+^7, = 0.
= it + p,
therefore substituting,
coefficients
7r
we have on
reduction
{a
(it
a + 13 1) (tt 1) + e (ir a + 7 1) +#} vx - {a {it -I) {tt-%) +e{w -1) + g] pvx = 0...(c).
whose
(x
And
of the form
vx
= %am x
(m
{x
+ 1)
...
+ m-1),
=0...(d),
m{a (m + f3 l)
is an arbitrary constant, corresponding to while all succeeding values of a m are determined by the law
which value a m
'
ART.
12.]
177
Hence the
when
equal
to,
or exceeds
by an
a particular root of the latter equation is 0, a particular may therefore always be obtained when (e) is satisfied either by a vanishing or by a positive integral value of m.
finite solution
As
by
(38)
admits of the
The ground
is
of this extension is that the symbol 7r, which here newly introduced under F, combines with the same
it
symbol
Pn
,. { ^(tt)
Pn
(/i(tt)
external to F, as if
were algebraic.
And
/
Thus
pu +/,
W u +/,
()
<f>
()
()
(- 1) p
tf
will
/,w+j;w+wp.+/,()+w*(*-i)A-i l
by
the assumption
^
12
P.W,.
While those transformations and reductions which 13. depend upon the fundamental laws connecting it and p, and are
expressed by (4) are common in their application to differential equations and to equations of differences, a marked difference exists between the two classes of equations as respects In differential equations the conditions of finite solution.
,
where
2-rable
^ = ~Tny
for
= e
->
forms B.F.D.
178
'
EXERCISES.
.
[CH. IX.
avr
err
+b n +er
TT
U+
,
a(7r
TT
n) +b
2
7T--
n P U=
(tt-|J (ir-n)
air
+b
primary in the sense implied by the fact that every binomial equation, whatsoever its order, which admits of finite solution, is reducible to some one of the above forms by the transformations of Art. 7, founded upon the formal laws connecting In equations of differences but one primary inteit and p. grable form for binomial equations is at present known, viz.
=-
ou
U,
and this is but a particular case of the first of the above forms for differential equations. General considerations like these may serve to indicate the path of future inquiry.
EXERCISES.
1.
Of what theorem
4,
Art.
2.
constitute a generalization ?
x(x +
3.
l)
A u + xAu - n\i = 0.
2
ferences of
Solve by the methods of Art. 7 the equation of difEx. 1, Art. 5, supposing a to be a positive odd
number.
4. Solve by the same methods the same equation supposing a to be a negative odd number.
179
CHAPTER
X.
y,
then
Ax
Ay
These
are,
x v
'
Ax
(!)
ux y
"
Ay
properly speaking, the coefficients of partial difBut on the assumption uXtV that Ax and Ay are each equal to unity, an assumption which we can always legitimate, Chap. I. Art. 2, the above are the partial differences of the first order of u XtV
ferences of the first order of
.
On
ference of
uXtV
is
A m+n
(Ax)
(A?j)
nU*> y
A \m / A \n
\Ay)
r
>
\Ax')
>
When
preted
the form of uXtV is given, this expression is to be interthe successive operations indicated, each elementary operation being of the kind indicate J. in (1).
by performing
Thus we
shall find
Ax Av Ux V ~ Ux+2Ax AV
_
'
'
2
.
U x+b.x,y+&.u
~t~
ux,y+a&y
in combination
and
are commutative,
122
180
OF EQUATIONS OF PARTIAL
[CH.
IjkX
in terms of -jCtX
as
in which
A =e
dx
ax~
it
follows that
Ax
is
an absolute constant,
A\ n
v
1.2
Ax/
(Ax)'
and therefore
// x+(w_i)Aa;,y
^f^ ^
)
+i . 2)
a - &c.j - (Ax)"
(4).
-r )
w^
it
would be necessary
-r
-r
to
symbolical expansions by the binomial theorem, perform the final operations on the subject func-
Though
in
variable will be
it
will
still
be
nition, e.g.
Ax A y
,
These things premised, we may define an equation of paran equation expressing an algebraic relation between any partial differences of a function uXttttK ^j the function itself, and the independent variables x, y, z Or instead of the partial differences of the dependent function, its
tial differences as
. . .
ment
successive values corresponding to successive states of increof the independent variables may be involved.
ART.
1.]
181
Tims
and
uXiV + y
x y
,
= 0,
0,
are, on the hypothesis of Ax and Ay being each equal to unity, different but equivalent forms of the same equation of partial differences.
the form -
Ax' Ay'
,
-^-,
dx' dy'
-r-,
singly or in succession.
Thus
A d x IT- ux y V T" U x y v + J v =
Ax
'
dy
'
is
an equation of mixed differences. Upon the obvious subordinate distinction of equations of ordinary mixed differences
and equations of
to enter.
partial
mixed
differences
it is
unnecessary
classes of equations a
Before engaging in the discussion of any of the above remark must be made upon a question of analogy and of order.
The theory of partial differential equations is essentially dependent upon that of simultaneous differential equations of the ordinary kind, and therefore follows it. But it may be affirmed that in the present state of the Calculus of Finite Differences any such order of dependence between equations of partial differences and simultaneous equations of differences is of a quite subordinate importance. In nearly all the cases
in which the finite solution of equations of partial differences is as yet possible it will be found to depend, directly or virtually, upon a symbolical reduction by which the proposed equation is presented as an ordinary equation of differences, the chief, or only, difficulty of the solution consisting in a difficulty of interpretation. And the same observation is equally applicable to equations of mixed differences, no general theory determining a priori the nature of the solution of equations of either species, partial or mixed, even when but
182
OF EQUATIONS OF PARTIAL
[CH. X.
of the first order, as yet existing. At the same time whatever belongs to the solution of simultaneous equations of differences and specially whatever relates to the process of their reduction to single equations is of common application, and of equivalent import, whether those equations are of the ordinary, of the partial, or of the mixed species. therefore think it right to treat of the subjects of this chapter in the order in which they have been enumerated, i. e. we shall treat first of equations of partial differences secondly, of equations of mixed differences; lastly, of simultaneous equations of all species.
We
.F(AA> = 0, F(AD)u = 0,
Now the symbol of operation relating to x, viz. A x or x combines with that relating to y, viz. A^ or y as a constant with a constant. Hence a symbolical solution will be obtained by replacing one of the symbols by a constant quantity a, integrating the ordinary equation of differences which results, replacing a by the symbol in whose place it stands, and the arbitrary constant by an arbitrary function of the independent variable to which that symbol has reference. This arbitrary function must follow the expression which contains the symbol corresponding to a.
The condition last mentioned is founded upon the inter-i pretation of (D a) X, upon which the solution of ordinary equations of differences with constant coefficients is ultimately dependent. For (Chap. VII. Art. 6)
(D-apX^a^X'a^X,
ART.
2.]
183
whence
{D-aY = ar t
i
=a
x-i
(c
+c x...+ cn_ xn
1 l
-1
),
The difficulty of the solution is thus reduced to the culty of interpreting the symbolical result.
Ex. 1. Thus the solution of the equation ux+1 aux which the symbolical form is
diffi-
0,
of
D ux -au = 0,
x x
being
ux =Cax
the solution of the equation
symbolic form
will be
is
..,=
W(y).
To
interpret this
we
B e
y
dy
we have
Ex.
2.
0.
in
D A u-u =
x
Now
is
replacing
by
a,
aAxu u =
u=(l+aT
therefore the solution of (1) is
x
)
C,
=(i+-0,T*(y)
(2),
184
where
OF EQUATIONS OF PARTIAL
[CH. X.
is an arbitrary function of y. <f> (y) the binomial, and applying the theorem
Now, developing
*V**(yJ-*(y-).
we
find
=</>(y)
+ #( 2/-l)+^=^^(y-2)+&c....(3),
when x
(2)
is
which
is finite
an integer.
Or, expressing
in the form
y,
and
inter-
we have
u
cj>
(y
x) + x(j> {y x -f
1)
+ ^2l^-*+J+&o
(4)>
Or, treating the given equation as an ordinary equation of differences in which y is the independent variable, we find as the solution
u=(Ax)-Vcb(x)
Any
of these three forms quirements of the problem.
(5).
may
Thus
the form
(2)
if it
e
my
,
0,
u should assume
or to revert to
(3)
which gives
{y)
=e
mj/
,
whence
l
u=(l+D y
x
)
mv
(l
+e
Ty x
)
my
{l
+ e-m
x
)
mv
(6).
We
3. There is another method of integrating this class of equations with constant coefficients which deserves attention. shall illustrate it by the last example.
ART.
3.]
185
tion
ab-b -1 =
Hence
a
0.
= 1+b
2 G (1 + b)'P
As
the
sible values
summation denoted by 2 has reference to all posof b, and C may vary in a perfectly arbitrary
manner
for different values of b, we shall best express the character of the solution by making C an arbitrary function of b and changing the summation into an integration ex-
tended from
oo
to oo
Thus we have
As <j)(b) may be discontinuous, we may practically make the limits of integration what we please by supposing cj> (b) to vanish when these limits are exceeded.
If
we
b,
we
have
uXtV =j
by
cj){b)db+xf
by
<j>{b)db
+
Now
X(k
*~ 1)
2
V"V (&)<& + fa
= ylr(6),
(?)
b 9 <l>{b)db
/.
>
(4).
186
OF EQUATIONS OF PARTIAL
[CH. X.
to
Although it is usually much the more convenient course employ the symbolical method of Art. 2, yet cases may arise in which the expression of the solution by means of a definite integral will be attended with advantage and the connexion of the methods is at least interesting.
;
Ex.
3.
Given
^
Ax
2
ic x _
1>y
= b?y uXty _
Replacing ux
by
u,
we have
or
(a:dv -a;dx)u = o.
But
therefore
or
= 0.
This
is
{DJ>y -i)u=0,
(Dx -Dv )u
1
= 0.
The
first
gives
D u-D - u = 0,
x
y
is
= (/>(?/ -a).
The second
gives,
by Ex.
u
1,
= ^r(x + y).
integral is
<j>
u=
(y
x) +
ifr
(y + x).
Upon the result of this example an argument has 4. been founded for the discontinuity of the arbitrary functions which occur in the solution of the partial differential
equation
72 a u
72 a u_
ART.
5.]
187
d2u
~d?
me, after what has been said in Chap. VIII. Arts. 4 and 5, to add that I regard the argument Analytically such questions depend upon the as unsound. following, viz. whether in the proper sense of the term limit, we can regard sin x and cos x as tending to the limit 0, when
It is perhaps needless for
x tends
to
become
infinite.
5. When together with A^ and A y one only of the independent variables, e.g. x, is involved, or when the equation contains both the independent variables, but only one of the operative symbols A x A y the same principle of solution is applicable. symbolic solution of the equation
, ,
F(x, A.,
will be found
A>=0
Ay
for
by
substituting
in the solu-
F(x,
Ax
a)
= 0.
And
F(x,y,
will be obtained
A x)=0
by integrating as if y were a constant, and replacing the arbitrary constant, as before, by an arbitrary function of y. But if x, y, A x and A y are involved together, this principle is no longer applicable. For although y and Av are constant relatively to x and A x they are not so with In such cases we must endeavour by respect to each other. a change of variables, or by some tentative hypothesis as to the form of the solution, to reduce the problem to easier conditions.
,
The
second
member
4.
extension of the method to the case in which the is not equal to involves no difficulty.
Ex.
Given
u v . xu v
= 0.
188
OF EQUATIONS OF PARTIAL
for
[CH. X.
Writing u form
u xy the equation
l l
may
be expressed in the
(1).
u-xD- Dy - u =
Now
is
replacing
~l
by
x
a,
the solution of
u ax~D~ u = or Gx (x 1)
Wherefore, changing a into
ux
1
.
axux_ =
x
. . .
x
.
D ~\
y
the solution of
(1) is
ii={D,-*)*x{x-\)...
1.0 (y)
Laplace has shewn how to solve any linear equation in 6. the successive terms of which the progression of differences is the same with respect to one independent variable as with respect to the other.
The given
A-x t
equation being
,
,
J/-2
+ ^C =
*
x y
,
A xy Bx
,
&c, being
functions of
x and
y,
let y x k
,
loj v x
the equation
Xv X X
,
...
X being functions of x.
x y, and the x y.
is
is
replaced by functions of
arbitrary constants
by
arbitrary
that the progression of difsuch as to leave x y unaffected, for when x and y change by equal differences Hence if x y is represented by k and we is unchanged. take x and k for the new variables, the differences now having reference to x only, we can integrate as if k were constant.
is
xy
Applying
this
method
to the last
example,
we have
ART.
7.]
189
= CX (x
l) ... 1,
result.
tion of
The method may be generalized. Should any linear funcx and y, e.g. x + y, be invariable, we may by assum;
ing it as one of the independent variables, so to speak reduce the equation to an ordinary equation of differences but arbitrary functions of the element in question must take the
place of arbitrary constants.
Ex.
5.
Given ux>y
-pux+l>y_ x
(1
0.
Here x
+y
is
invariable.
Now
the integral of
is
Vx =:c
+ c'(
is
u*,v
<
(*
+ y) + (~Vy ir x + J)( 1
Equations of partial differences are of frequent oc7. The following is currence in the theory of games of chance. an example of the kind of problems in which they present themselves.
engage in a game, each step of which them winning a counter from the other. At the commencement, A has x counters and B has y counters,
Ex.
6.
and
consists in one of
and in each successive step the probability of ^4's winning a counter from B is p, and therefore of B's winning a counter from A, 1 p. The game is to terminate when either of the two has n counters. What is the probability of A's winning it ? Let u XiV be the probability that values being assigned to x and y.
A will
win
it,
any positive
190
OF EQUATIONS OF PARTIAL
1
[CH. X.
winning the game may be resolved into two 1st, His winning the first step, and afterwards winning the game. 2ndly, His losing the first step, and afterwards winning the game.
a
Now A
alternatives, viz.
The
probability of the
first
1
first
alternative
1
is
As
J
pux+
Xi
y_ v
for after
which
y_ x
(1
is
in like
manner
Hence, the probability of any event being the sum of the which it is composed, we have as the equation of the problem
is,
by the
last
example,
lt
i>
x + y)
The number
game.
of counters
it
Represent
x+y by m, then
2J
Now
of
A" 8 success is certain if he should ever be in possession n counters. Hence, if x = n, uxy = 1. Therefore
Again,
loses the
game
B will
mn
o=*w+y^w
ART.
8.]
191
The
last
P 1
1
>
V
,
<j>
W = pn _
= w* *#
_ ~pm-n
pm-n
'f
'W = pn _ p m-n
-I
whence
jyzn-x-y
(2),
which
is
will
Symmetry
win the game
B will
(i- P )"- P
~
to
be unity.
The problem of the duration of play' in which it is proposed to find the probability that the game conditioned as above will terminate at a particular step, suppose the rth depends on the same equation of partial differences, but it involves great difficulty. very complete solution, rich in its analytical consequences, will be found in a memoir by the late Mr Leslie Ellis [Cambridge Mathematical Journal, Vol.
,
IV. p. 182).
by the method
Laplace usually solves problems of the above class of generating functions, the most complete statement of which is contained in the following theorem.
8.
then making x
= e e y = e 6 &c, we
'
have
d
(i).
^(^(^...)v?,,,!^'"'
192
OF EQUATIONS OF PARTIAL
[CH. X.
Here, while 2 denotes summation with respect to the terms of the development of u, S denotes summation with
respect to the operations which would constitute the first member a member of a linear differential equation, and the bracketed portion of the second member a member of an
equation of differences.
Hence
it
follows that
if
we have
^(m,...) Mm_^,. =
(2),
the equation (1) would give for the general determination of the generating function u the linear differential equation
d_
d_
(3),
But if there be given certain initial values of umn which the equation of differences does not determine, then, corresponding to such initial values, terms will arise in the second member of (1) so that the differential equation will assume the form
$.&..)<**'-'{,.:)
If the equation of differences
differential equation
W-
have constant coefficients the merges into an algebraic one, and the generating function will be a rational fraction. This is the case in most, if not all, of Laplace's examples.
It must be borne in mind that the discovery of the generating function is but a step toward the solution of the equation of differences, and that the next step, viz. the discovery of the general term of its development by some independent process, is usually far more difficult than the direct solution of the original equation of differences wouh As I think that in the present state of analysis the be. interest which belongs to this application of generating functions is chiefly historical, I refrain from adding examples.
ART.
9.]
193
an equation of mixed differences admits of re9. solution into an equation of simple differences and a differential equation, the process of solution is obvious.
When
Ex.
7.
A -?
du dx
A a&u
7 o -j-
du , + abu = ax
Jj-)(A-J)-0,
the complete value of u will evidently be the values given by the resolved equations
sum
of the
-j
au
= 0,
t\u
bu = 0.
Hence
w=
where
e^ is
Cl
6-+c2 (i+5r,
an absolute,
c2 a periodical constant.
Ex.
8.
Ay = x -j- Ay +
dz *_+y,
fdz\
2
-7-
Ay) being
we
have, on integration,
z
ex + c
2
,
v y = ^z =
where
c is
ex (x iA
1) ;
C, +cx+ ^
2
an absolute, and
C a periodical
constant.
B. F. D.
13
194
OF EQUATIONS OF PARTIAL
[CH. X.
Equations of mixed differences are reducible to differential equations of an exponential form by substituting for x or
Ax
their
differential
expressions
e**,
d ax
1.
becomes
Ex.
9.
Au -j- =
and
its
the values of
m being the
e
-l-m=0.
10.
differential equations {Biff. Equations, p. 440) has been extended by Poisson to the solution of equations of mixed dif-
* + Z*fe + w+ JR^F
Jfi
(1),
where L,
MN
y
V are functions
,
of x.
in the
Writing u
form
for
it is
easily
shewn that
it is
ax
Hence Lee
if
we have
(2)
:
N-LM-L'=0
ART. 10.]
195
(+*)
which mixed
is
(z>
+ z,._f;
resolvable
by
differences
if
and a
But
(D
+ L)u = v
(3),
we have
(J^+M\v + {N-LM-L')u=V,
whence
u
which
is
N-LM-L'
.
to,
u = A x -^ + Bx v+Cx
r>
/-i
dv
(3)
we have
dv
4 x^ +
which, on division
Mx + B^Dv
is
+ {LB-l)vt =-Cx-LC
by
A x+V
of the form
The
into
altered coefficients,
tion, if
original form of the equation is thus reproduced with and the equation is resolvable as before
differential equa-
N-L^-L^O
is satisfied.
(5)
is to
be repeated.
132
196
OF EQUATIONS OF PARTIAL
[CH. X.
An
are employed in the above process leads to another reduction similar in its general character.
(B + L)( + M. t) u
where mixed
+ (N- LMJ
u=V
an equation of
M_ = D~ M,
X
its
differences
and a
the condition
N-LM_=0
If this equation be not satisfied, assume
(6).
{ + *+)
u-v
>
and proceeding as before a new equation similar in form to the original one will be obtained to which a similar test, or, that test failing, a similar reduction may again be applied.
Ex.
This
first
10.
Given
^ -a -^ + (x n) u
x+1
axux = 0.
Taking
L a M=xn, N= ax.
t
(2)
is
not satisfied.
But
(3)
and
(4)
dv
dX
whence
+{x ~ n)v
an
&- an
-n
ART. 11.]
or,
197
on reducing,
dv^
dx
>+!
dv,
a iz
dx
+ xi
-1
) 1
Vr axv*=
0.
Comparing this with the given equation, we see that n reductions similar to the above will result in an equation of the form
dw dw ==* a -=
x ,.
dx
dx
V xw-.. x+1
axw 0, x
'
(| +
is
*)(D-),= 0,
unmixed
character.
Poisson's second reduction applies when the upper sign is taken in the equation given and thus the equation is seen to be integrable whenever n is an integer positive or nega;
tive.
Its actual
solution deduced
by another method
will
be
coefficients
involved only in the first degree admit of a symbolical solution founded upon the theorem
{^(i^X^W^X
[Differential Equations, p. 445.)
(1).
The
Since
following
is
'
the second
member
-y-
operate on
only,
and
-y-
on
w,
we
193
OF EQUATIONS OF PARTIAL
[CH. X.
d\
d\
Let
yfr
i-^-)u
fd
= v,
then
-W"i- +I ^'
or if
i|r
(-=j
be replaced by
'
.">-*<*>.-{.$}.
Inverting the operations on both sides, which involves the inverting of the order as well as of the character of successive operations, we have
10,
which we
shall' express
in the
n being
Now
putting u for ua
u ~ {J -a) +
x
ne
dx
\
[dx
+ x (edx - a) u = 0.
Let
then
(e
dx
a)u = z,
we have
J" + ne*
^ _ a)-
1
]-
+ xz = 0.
ART. 11.]
Or,
199
d
dx
ne
X
Tx
^ _J
Hence,
and therefore by
(1),
Si
J:
_ J \Tx) =
e
*
-l
""1 (di)
-Mlog (e^-a)
_a
to
n
)
e*
Uj aT e"4 [dx)
(e
dx
- a)'n
(5)
It
is
desirable
By
(1),
we have
(s +
bolically equivalent,
aj
=e "
G&C
the form
,:=(<--fl)--T(iL)
,T (*-_)-<)
(o).
Now
d (e *
a)
z,
therefore substituting,
and replacing
e- by B,
w = (Z>_a) w
-1
6
'"2^J
e^(D-a)-n
(.4).
Two
200
First, let
OF EQUATIONS OF PARTIAL n be a
positive integer
(c
;
[CH. X.
then since
...
+cx
x
+ c^oT ),
1
,
(D-a)
we have
B-1
=(A + l-a)
{0+/e
2
B-1
w=(A + l-a)
.
w -1
e~ 2
ax
(c
+c
x...
cn _ 1 x
n- 1
)dx]
(*)>
ing
For
integrat-
x r x Se^a x dx^6^a x^
and
in
r-1
dx + {r--l)f^axxr 2 dx
#2
x
x2
/e 2 a
xdx=
e2
a*+loga/e 2
a*6&c.
These theorems enable us, r being a positive integer, to reduce the above general integral to a linear function of
elementary integrals fe 2 ax dx, and of certain algebraic x m terms of the form e 2 a x where m is an integer less than r.
the
,
Now
For
if
we
(d), it
will
(A
+ 1 - a)
71
'1
e^
(e*
x a x
m
)
=
=
(A +
- a) n_1 axxm
0,
is
since
is less
than
r,
1,
(A
+ 1 - a)
71
'1
e~T{C
+C
/e? ax dx)
ART. 11.]
201
and here
constant,
introduced by ordinary integration is an absolute while Cx introduced by the performance of the operation 2 is a periodical constant.
A
fluity
superfluity
among
which does not affect their arbitrariness, is always to be presumed when the inverse operations by which they are introduced are at a subsequent stage of the process of solution followed by the corresponding direct operations. The particular observations of Chap. xvn. Art. 4. {Differential Equations) on this subject admit of a wider application.
Secondly,
let
n be
or a negative integer.
It is here desirable to change the sign of the given equation in the form
so as to express
du -dx
t
du a-^ +
.
dx
ix v
n) u, axu = A 0,
v
'
while
its
= (D- a)-n- eT f)
1
e^ {D
- a) n 0.
And
in both
is
or a positive integer
Now
since
(B - a) n
= 0,
'1
-x~
and
'd_
l-j-
dx
(7,
we have
ip- a)-
11
Ce-r
-x*
2
C(D -a)"'*-l
+ (2>--a)-^0
X*
2
Ca
x
-l
+:
aT e
-X2
-\-a
x ~ n-l
Vn +1
(c
is
arbitrary, the
+ 1 periodical constants c
cx
. . .
202
OF EQUATIONS OF PARTIAL
[CH. X.
The general expression of these relations is somewhat complex; but in any particular case they may be determined without difficulty.
Thus
if
a=
1,
n=l,
1
it
will
be found that
=0 XVrt T+
If a
(l-x).
= 1, n 2, we
shall
have
verified,
though not
by
12. The same principles of solution are applicable to equations of mixed partial differences as to equations of partial
differences.
If A^. and
-y-
are
the
symbols of pure
operation involved, and if, replacing one of these by a constant m, the equation becomes either a pure differential equation or a pure equation of differences with respect to the other, then it is only necessary to replace in the solution of that equation by the symbol for which it stands, to effect the corresponding change in the arbitrary constant, and then to interpret the result.
Ex.
11.
Lxu-a~
-j-
dy
=0.
integrating,
(1
Replacing
by m, and
u
we have
=c
+ am)
x
.
(J)^)>
ART.
12.]
203
d\ x
ty (y) being
an arbitrary function of y.
Ex.
12.
VXtV
Treating
is
-*- (4)"*
having reference
to x.
(1)^... +(!)"*
No
constants need to be introduced
by
[-7-
Ex.
13.
Given u x+2 - Sx
...
^
,
+ 2x
(x
1)
= 0.
Let ux =l .2
(x
2)vx
then
rf
r-M'Ty + ^ty)
{
D'-i){ D*dy)
x
dy
whence by
resolution
and integration
^-i....(.-){(|)Vw + r(*)V}
Ex.
14.
-3 wx+2 +i
^ + Vf = ^ dy dy
2
where
7 is
a function
of
x and y
204
OF EQUATIONS OF PARTIAL
[CH. X.
Here we have
K- ^| +2 (IJK =F;
3
= \[Dd f n dy\ x
dy
dy
dyj
^-iT"
dy)
~
dy\ dy)
\
dy\dy)
\dy)
The complementary
the performance of
u introduced by
will evidently
be
tfr(y).
''|)"ttjr)
But
may
,
be avoided.
have, as a particular solu-
Thus, representing
tion
V by Vx we
y.
which terminates if Vx is rational and integral with respect The complement must then be added.
t(
when
V=F(x)+y,
is
ART.
13.]
205
Simultaneous Equations,
Whether the equation be one of ordinary differences, 13. or of mixed differences, the principle of the method of solution
is
the same.
We must,
by
Ax
and
-=-
enable us by elimination to deduce a final equation involving only one of the dependent variables with its differences or The integrations of this will give the general differentials. value of that variable, and the equations employed in the process of elimination will enable us to express each other dependent variable by means of it. If the coefficients are constant we may simply separate the symbols and effect the eliminations as if those symbols were algebraic.
Ex.
15.
= 0,
v x+l
xux = 0.
From
the
we have
ux+i -a2
{x
1)
^ = 0.
+1
u x+2 - a2x
the solution of which
is
(x
l)ux =0,
Then by
the
first
equation
U^ = 1.2...(a;-l) c M+ _ a y^ 0{ a ax
l {
Ex.
16.
206
EXERCISES.
[CH. X.
- 4) 2 ux =
a2 - 2
/
(a
- 2)a* - 2a-*is
1
.
The complete
integral of
which
m
^X
4-(c
(a
^ -4) 2 ~
a
aT*
a(a" 2 -4) 3
a; .
Moreover by the
first
EXEECISES.
1.
umtl vm = 2m(x+l)
vx+1
2.
(1),
(2).
-ux = -2m{x + l)
+(l) x vx = 0,
^+i+(-l)X=0.
3.
4.
Integrate
Aa,v ~ a J~ u*.v = -
6.
7-
W" M
*.i'
+i
= + y-
^1,^-^=^.
EXERCISES.
8.
207
d2
where
affects
x only
shew that
208
CHAPTER XL
OF THE CALCULUS OF FUNCTIONS.
The Calculus of Functions in its purest form is dis1. tinguished by this, viz. that it recognises no other operations than those which are termed functional. In the state to which it has been brought, more especially by the labors of Mr Babbage, it is much too extensive a branch of analysis to permit of our attempting here to give more than a general view of its objects and its methods. But it is proper that it should be noticed, 1st, because the Calculus of Finite Differences is but a particular form of the Calculus of Functions 2ndly, because the methods of the more general Calculus are in part an application, in part an extension of those of the
particular one.
In the notation of the Calculus of Functions, <j>{ty(x)} is usually expressed in the form <j>y]rx, brackets being omitted except when their use is indispensable. The expressions <jxfix, (jxjxftx are, by the adoption of indices, abbreviated into 2 s ip x, <j> x, &c. As a consequence of this notation we have The inverse form (j)x = x independently of the form of </>. cj)~ is, it must be remembered, defined by the equation
l
(jxfT
_1
=x
(1).
Hence < may have different forms corresponding Thus if same form of
</>.
to the
cj)X
= x2 + ax,
we
=
(/>
x
and
</f *
ART.
2.]
209
The problems
kinds, viz.
1st.
two
Those in which it is required to determine a funcform equivalent to some known combination of known forms e. g. from the form of yfrx to determine that of nx.
tional
;
2ndly. Those which involve the solution of functional equations, i. e. the determination of an unknown function from the conditions to which it is subject, not as in the previous case from the known mode of its composition.
We
may
inverse.
properly distinguish these problems as direct and Problems will of course present themselves in which
Direct Problems,
2.
yfrx,
required that of
n
yjr
x.
There are cases in which this problem can be solved successive substitution.
by
Ex.
1.
Thus,
if
tyx
and generally
n
ifr
= xa\
2
3
Again, if on determining ty x, ty x as far as convenient it should appear that some one of these assumes the particular form x, all succeeding forms will be determined.
Ex.
2.
Thus
if
tyx
= l x, we have ^x = 1 (1 x) = x.
or
Hence
Ex.
3.
ty
x=
x
=
1
7
x according
,
as
is
odd or even.
If
tlrx
T
x we
X
find
^x
B. F. D.
-tfx T
= X.
14
210
[CH. XT.
Hence
n by
ty
= x,
1
j.
X
or
1
x
according as on dividing
3 the remainder
is 0, 1
or 2.
Functions of the above class are called periodic, and are distinguished in order according to the number of distinct n forms to which ^\r x gives rise for integer values of n. The function in Ex. 2 is of the second, that in Ex. 3 of the third,
order.
made
Theoretically the solution of the general problem may be to depend upon that of an equation of differences of the
first order.
For assume
fx=
+1 Then, since iy x
t,
+, f" * =
+1
(2).
= ^jn^'x, we
have
(3)-
4- + W
The
be determined by the condition
condition
t
o
arbitrary constant in the solution of this equation may = y\rx, or by the still prior t x
= -fx = x
(4).
It will be more in analogy with the notation of the other chapters of this work if we present the problem in the form Given yfrt, required ijr% thus making x the independent variable of the equation of differences.
:
Ex.
4.
Given
x
yjr t
yfrt
= a + bt,
have
required
x
yfr t.
Assuming
= ux we
is
+i
= + bux
ux
= cb x + ^j
a
Now
yfr
t,
therefore
'= c+
r--r
ART.
2.]
211
Hence determining
we
find
on substitution
(5),
ux = a^= + Vt
the expression for
x
yjr t
required.
Ex.
5.
Given
x
tyt
=j
required
x
-*jr
t.
Assuming
ty
ux we
have
or
uxux+1 + bux+1 = a.
This belongs
to the third of the
Art. 9.
Assume
therefore
then
vx+2 - bv x+1
is
- avx = 0,
x
2
vx
a and
ft
= c a?+c
1
(3
m bm a 0.
2
HenCC
or,
-=
\a* + cl{3
putting
for ->
ci
and a + /9
for 6,
and reducing,
1
<**-'
"--P-VTW
Now
u
+ C/3*-
(6)
'
= yjrt
t,
therefore
= ~ al3 -LTc-
P + Ca 1 + C
'
142
212
[CH. XI.
whence
and, substituting in
G
t
+a
(6),
ux U
= - P aft
, +1
_ /gx+1 + (ax_^^ =y
O
V),
x
t
required
a
,
ifrt
we
have,
by
direct
~t~ t
a
b
a
,
+ sfrt
+b
+ t'
and continuing the process and expressing the result in the usual notation of continued fractions,
a a
a a
i'
v^fcF+b + Zb +
the number of simple fractions being x. Of the value of this continued fraction the right-hand member of (7) is therefore the finite expression. And the method employed shews how the calculus of finite differences may be applied to the finite evaluation of various other functions involving definite repetitions of given functional operations.
Ex.
6.
Given
tyt
=
C -p
it
required ty
x
t.
Assuming
differences
as before yp"t
uxi we
eux ux+1
+ cux+1 -bux -a =
same method as
before,
(8),
and applying
to this the
we
find
cr + qy"
*
c
e
{y)i
6?+Cp
a and
/3
eV-(5+c) em + bc-ae = Q
(10);
AET.
2.]
213
and
t,
7(t^)T~c
We
(11) -
When a and f3 are imaginary, the exponential forms must may, however, so be replaced by trigonometrical ones.
integrate the equation (8) as to arrive directly at the trigonometrical solution.
For
let that
c\ (
b\
bc
ae^
Then assuming ux =
tx
we have
or
,.
,
tjx+l
(12),
2
/1oN
Hence
or,
U^^_ = _
v
1
fJ,
tx tx+1
assuming
tx
vsx
+ VW
I
'
f*
is
= tan C x tan'
tx
But
tx
= vsx
and ux =
+ fi,
where
Hence
ux =
z>tan (
Q-x tan
-)
/*'
(15).
214
[CH. XI.
Now
t gives t = v tan C + p.
have, finally,
Hence determining
C we
1
^=
v tan ftan"
-^x
yjr t.
a;
tan" -}
/*'
(16),
This expression
is
A + Bt
the coefficients A, B,
Beverting to the exponential form of yjr t given in appears from (10) that it is real if the function
(b
(9), it
+ c)
bc
ae
is positive. But this is the same as 4zA The trigonometrical solution therefore applies when the expression repre2 sented by v is positive, the exponential one when it is negative.
In the case of v =
tf-w
+ Pfom-O^O,
tx
tx+1
~ fi>
is
-+
^x-^ + ^t
'
fJLX
fJU
Xt
x
'
a result which
solution
by
may also be deduced from the trigonometrical the method proper to indeterminate functions.
ART.
3.]
215
Periodical Functions.
It is thus seen, and it is indeed evident a priori, that 3. x in the above cases the form of "^r t is similar to that of yfrt,
but with altered constants. The only functions which are known to possess this property are
a
c
+& + et
j * and at.
On this account they are of great importance in connexion with the general problem of the determination of the possible forms of periodical functions, particular examples of which will now be given.
Ex.
7.
tion of the
conditions
is
+ bt
By Ex.
we have
f t=a^j- + b%
and
this,
for the
t.
particular value
of
in question,
must
reduce to
Hence
a
b
-\ =0 fZi
x
>
^=1
>
equations which require that b should be any except 1 when a is not equal to 0, and any when a is equal to 0.
xth xth
Hence
if
we
riodic forms of
confine ourselves to real forms the only pea + bt are t and a t, the former being of
latter of
Ex.
8.
+ et
th
order.
= 0,
In the following investigation we exclude the supposition of which merely leads to the case last considered.
216
[CH. XI.
Making then
t
=
1
t,
we have
1
fi'
+ ^taiWtan" -^^--cctan~
-j
(18),
or
V
= tan
(tan
\
-1
x tan"
V
fJbJ
an equation which, with the exception of a particular case to be noted presently, is satisfied by the assumption
x tan - = ^7^,
i being an integer.
Hence we have
-=tan
fJb
(19),
or,
jjl
their values
H
from
(13),
whence we
find
tf-Zbccos^ + c2
4a cos
(20).
case of exception above referred to is that in which in which therefore, as is seen from (19), i is a mulFor the assumption v = makes the expression for tiple of x. t given in (18) indeterminate, the last term assuming the form If the true limiting value of that term be found in x oo the usual way, we shall find for t the same expression as was But that expression obtained in (17) by direct integration. would lead merely to x = as the condition of periodicity, a condition which however is satisfied by all functions whatv
The
= 0,
and
(j)t
t.
The solution (9) expressed in exponential forms does not lead to any condition of periodicity when a, b, c, e are real
quantities.
ART.
4.]
217
We
not of the form + Bt, is a periodical function of the x order, are expressed by (20), * being any integer which is not a multiple of x*.
ih
+ et'
when
4.
of others
From any given periodical function an infinite number may be deduced by means of the following theorem.
(fcfifi^t
Theorem, lift be a periodical function, then a periodical function of the same order.
For
then
let
is
also
#/>~^ =
yfrH
^K = tff-^fft
And
f = 4>f
n
t
(j)- t.
Now,
if
t,
and
r^t^^t.
Hence
Therefore
n
ty
= (jxjfH =
n
th
t.
order.
t
Thus, it being given that 1 t is a periodic function of the second order, other such functions are required.
Eepresent
1
of
hj ft.
t
2
,
Then
If 6t
if cj>t=t
#/>-**
=(i-V0
2
.
= *Jt,
number might be
indefinitely multiplied.
The system
(fcffy'H
of functions
included in the
general form
has been
218
[CH. XI.
Functional Equations.
The most general definition of a functional equation that it expresses a relation arising from the forms of functions; a relation therefore which is independent of the particular values of the subject variable. The object of the solution of a functional equation is the discovery of an un5.
is
its
which
known.
nature of functional equations
their genesis.
c,
The
is
which con(j>x,
by
then
<px=f(x,
and changing x
into
c),
yfrx,
Eliminating
of the form
=0
cf>,
(1).
This is a functional equation, the object of the solution of and yjr which would be the discovery of the form those of
being given.
above process nor its result were a function of x which did not change its form when x was changed Thus if we assume as a primitive equation into tyx.
It is evident that neither the
would be
and change x
into
(x)
cx + have
(a)
x, we
<l>{-x)
= -cx+-.
Eliminating
we
have, on reduction,
, -{*H)} -%
{*()}
ART.
5.]
219
interpret c as an arbitrary even function of x, the only condition to which it is subject being that it shall not change on changing x into x. Thus we should have as particular solutions
<j)
In that primitive we
may however
(x)
= x cos x +
=3? + ^,
c
COS
<j>(x)
by assuming
= cos x
and x2 respectively.
Equations of differences are a particular species of funcchange being that of x into x + 1. And the most general method of solving functional equations of all species, consists in reducing them to equations of differences. Laplace has given such a method, which we shall exemplify upon the equation
tional equations, the elementary functional
the forms of
y]r
(2),
that of
cj>
sought.
But
though we shall consider the above equation under its general form, we may remark that it is reducible to the simpler form (1). -1 For, the form of yjr being known, that of i/r may be presumed to be known also. Hence if we put ^x = z and x~^ z ^\z we have
->
F(f~%
and
this,
<j>z,
<f>f 1
z)=Q,
is
since
-1
-v/r
and
are
known,
reducible to the
general form
(1).
Now
resuming
(2) let
$y\rX
= Vn
vt
<p
Xx = V t+1
relation
(4),
= <j>u
will be determined if
we can
express v t as a
220
[CH. XI.
Now
nating
x an equation
(5),
the solution of which will determine u therefore tyx, therefore, by inversion, a? as a function of t. This result, together with the last two equations of the system (3) will convert the given equation (2) into an equation of differences of the first order between t and v t the solution of which will determine v as a function of t, therefore as a function of u t since the form of u has already been determined. But this determination of vt as a function of u is equivalent, as has been seen, to the determination of the form of <.
,
Ex.
9.
<j>
(mx)
acf>
(x)
= 0.
Then assuming
x
<j>
(x)
= un =v
t
(/>
mx = ut+1\ (mx) = v J
t
'
we have from
the
first
two u t+1 -
mu = 0,
t
is
u
Again, by the
last
Cm*
(a)
(b).
two equations of
v t+1
becomes
~ av = 0,
t
whence
v^C'a*
Eliminating
t
(c).
between
v
(b)
and
(c),
we have
.
log u -log C
t
C'a
x, v t
logw
lose
Hence replacing u by
t
by
l
<#,
and C'a
l0 %
m by
C we
l9
have
og*
(d).
(f)X=
G ay^ m
x
ART.
5.]
221
And here Gt must be interpreted as any function of does not change on changing x into mx.
If
x which
we
an
+a
t x
iogmj
+a
2 i
cos
47r
iogmj
~~
+ &c.
+K
a
,
sin
jfsNr
iogmj
But it suffices to Oji h iy &c. being absolute constants. adopt the simpler definition given above, and such a course we shall follow in the remaining examples.
Ex.
10.
Given
<
(\^) - <*$ () = 0.
_ ~~ Ut
l
>
Assuming
+ ^_ U( x ~ +v
\TZTxJ
$
we have
()
= v $
=v
l+u
or
u ut+1
t
- ut+1 + u +l=0.
t
The
solution of
which
u
t
is
= tan
0+-H.
t
Again we have
v t+1
-av = 0,
whence
Hence replacing u by
t
a?,
by
4
(j>
(x),
and eliminating
t,
- tan-'x
(x)
CX
222
[CH. XI.
ing
+x
X
.
into
L ~~
6.
^x + a^-'x + a^-'x
For
let it
...
+a n
<j>
(x)
=X
(6),
may
be reduced to
be a symbol which operating on any function effect of converting it into <^r (x) Then the <f> above equation becomes
(x)
has the
n 7r < (x)
+ ay- ^ (#)...+ an $
1
(x)
= X,
(7).
or
(>n
+ ay-
...
+ an)<l>(x)=X
{u
+ v) = ITU + 7TV,
and that
it is
= airu.
reduce
(7)
Hence we
manner,
viz.
are permitted to
in the following
<$>(x)
(Tr'>
+ ay-
...
+ an)2
X
(8),
= [N
^- my + N. ^-my...}X
1 1
m m
.
...
mn + wn~
...+a =
(9),
NN
as in the analogous
Now
if (it
my
X=<f)
(7T
(x),
we have
m)(j) (x) = X,
ART.
or
to
7.]
223
(x)
m<j)
(x)
X, = xn
Ex.
11.
Given
<j>
(m 2 x)
ft
+ a<j> (mx) +
b(f>
(x)
Eepresenting by a and
solution
is
the roots of
log
+ ax + h = 0,
logfl
the
w
C and
into
= m + am +b + c
*-
xM
^ m+G x ^>
'
C being
functions of
of
mx.
Here we may notice that just as in linear differential equations and in linear equations of differences, and for the same reason, viz. the distributive character of the symbol it, the complete value of <j> (x) consists of two portions, viz. of any particular value of <p (x) together with what would be its = 0. This is seen in the above complete value were
example.
There are some cases in which particular solutions of 7. functional equations, more especially if the known functions involved in the equations are periodical, may be obtained with great ease. The principle of their solution is as
follows.
to
be
(10),
F(x, (j)x,^x)=0
and
let
tyx be a periodical function of the second order. Then changing x into ^x, and observing that ^jr z x = x, we have
F(fx,<f^rx
Eliminating
cfryjrx
^x)=0
(11).
<f>x
as a function of
known,
If
tyx, as a function of x.
is
x and
^x
is
supposed
^x
it
would
be necessary to effect the substitution twice in succession, and then to eliminate (frtyx, and (j>^x and so on according to
;
224 Ex.
12.
[CH. XI.
Given
1
(c/^)
$ *
J.
"T"
= X
a*x.
The
then
function
J.
X
"r~
is
Change
into
1
+x
and we have
4>x = a Y 1+xJ Y
1
,l
x\
~x
.
1+x
Hence, eliminating
X
,
<j>
j.
~\~
we
find
<px= ai
as a particular solution.
l+x^
Equations, p.
7).
fails if
yield a
the process of substitution does not sufficient to enable us to effect the elimination. Thus, supposing yfrx a periodical function of the second order, it fails for equations of the
This method
form
F((j)x,<t)fx)=0,
if
we must
symmetrical with respect to <px and cf>yjrx. In such cases either, with Mr Babbage, treat the given equation
as a particular case of some more general equation which is unsymmetrical, or we must endeavour to solve it by some more general method like that of Laplace.
Ex.
13.
Given
This
is
{(px)
+ m -U (^ - xj = 1 + nXx,
I
respectively,
and n being constants which must be made equal to and %x being an arbitrary function of x.
and
ART.
7.]
225
Changing x
x, we have
rrr
{* (f
Eliminating
" x)} + m
MX W =
ft
+nx (i "
*)
<(--#]
2
(1
ra
2
)
{< (a?)}
= 1 on +
jx#
m%
Therefore
Now
fraction
if
become
5
and n become
0,
independently, the
may
be replaced
by an
arbitrary constant
Thus we have
( )
;
{<
x )Y
+ cx x - c%(| - x)
whence, merging
*(*)
= +*(*) -%(f-*)}
(12).
that,
is
in effect Mr Babbage's solution, excepting and n dependent, he finds a particular value which in the above solution becomes an arbi-
\(j)
Let us now solve the equation by Laplace's method. (x)Y = tyx, and we have
Let
f{x)+^(j-x) = l
Hence assuming
7r
B. F. D.
15
226
OF THE CALCULUS OF FU
[CIT.
XI.
we have
u t+1 + u := 2'
t
7T
**H
+ % = 1.
The
solutions of
which are
u
t
=c
(-
v+5.
ir+i.
%=C
Hence
(-
Iljl= -* =
7T
c.
Therefore
i+fS
or
*(z)=i + c(*-).
Therefore
,
^+
C, a,
-l)}'
a;
in
which
C must
be interpreted as a function of
is
which does
not change
when x
changed into -
IT
x.
~ x.
It is in fact
an
ofx and
The
is
included in this.
cj>
(x)
with
a view to
ART.
8.]
227
c=
x () - x (f IT
x xM x^X
TT
IT
~I n~ x -i
IT
which
8.
is
x and
TT
x.
siderable importance, which involve at once two independent variables in such functional connexion that by differentiation and elimination of one or more of the functional terms, the
solution will be made ultimately to differential equation.
Ex. 14. Representing by P<p (x) the unknown magnitude of the resultant of two forces, each equal to P, acting in one plane and inclined to each other at an angle 2x, it is shewn by Poisson (Mecanique, Tom. I. p. 47) that on certain assumed principles, viz. the principle that the order in which forces the principle of are combined into resultants is indifferent (so-called) sufficient reason, &c, the following functional equation will exist independently of the particular values of
x and
y, viz.
<
+ y) + < x - y) =
(
0*0
<M^)-
Now,
we have
(v+y)
<t>
(y)
And
to y,
differentiating the
<t>"{+y)
Hence
^ = (y)
<P(x)
tj>(y)
152
228
[CH. XI.
of
is
"being
an arbitrary constant.
The
(x)
tion is
(x)
<f>
= Aemz + Be~mx
or
<j>
and B,
we
find
<j)(x)
= mx +e~mx
or
2cosmx.
assuming, on the afore-named principle of sufficient reason, that three equal forces, each of which is inclined to the two others at angles of 120, produce equilibrium, it fol-
Now
lows that
<j)
( )
</>
= 1.
(x)
second form of
Thus
<j)(x)
= 2 cos x.
be taken, and that m be made equal to 1. And hence the known law of compo-
Ex. 15. ball is dropped upon a plane with the intention that it shall fall upon a given point, through which two perpendicular axes x and y are drawn. Let cj* (x) dx be the probability that the ball will fall at a distance between x and x -f dx from the axis y, and </> (y) dy the probability that it will fall at a distance between y and y + dy from the axis x. Assuming that the tendencies to deviate from the respective axes are independent, what must be the form of the function (x) in order that the probability of falling upon any particular point of the plane may be independent of the position of the rectangular axes? (Herschel's Essays).
The
functional equation
is
easily found to be
ART.
8.]
229
x and with
respect to y,
we
have
Therefore
iM = jM.
x<p(x)
y$(y)
= 2m,
X(p (x)
^{x)
= Cemx\
The condition that <f> (x) must diminish as the absolute value of x increases shews that must be negative. Thus
we have
<
(x)
= Ce^\
EXERCISES.
1.
If
<j){x)
<2x
j.
~
2
25
determine
n
<j>
(x).
2.
If
cj)
(x)
=2x -l,
determine
n
(j>
{x).
3.
If ty
(t)
=
c
+ et
and ^(t)
f
x
(t)
means
A = E = C-B
a
e
230
4.
EXERCISES.
[CH. XI.
Shew hence
that
x
yfr
(t)
may
bx
a + hxt -h + c + et'
Ex.
5,
Art.
2,
may
hence be deduced.
5.
=/(# + #)
6.
to
2
1
&c.
7.
What
is
deducible
by the method
f(x)+mf(^-) = a +
n(f>(x).
Required the equation of that class of curves in which 8. the product of any two ordinates, equidistant from a certain ordinate whose abscissa a is given, is equal to the square of that abscissa.
th If 7tx be a periodical function of x of the n degree, will exist a particular value of f(ir) x expressible in the form
9.
+ ajrx + a 7r x
2
. . .
+ an_jrn
~x
x,
a v aq
, . .
an_ t
EX.]
10.
EXEECISES.
231
Shew hence
will
be
, 6 r (x) =
,
a3
1
/
5
[x
\
11 + a; +al x
ax + a
2~
~3
x-\ rr +
11.
obtained
The complete solution of the above equation will be by adding to the particular value of x the comple4tan~'a:
mentary function Ca
12.
232
CHAPTER
XII.
GEOMETRICAL APPLICATIONS.
The determination of a curve from some property con1. necting points separated by finite intervals usually involves the solution of an equation of differences, pure or mixed, or more generally of a functional equation.
The particular species of this equation will depend upon the law of succession of the points under consideration, and upon the nature of the elements involved in the expression of the given connecting property.
Thus if the abscissas of the given points increase by a constant difference, and if the connecting property consist merely in some relation between the successive ordinates, the determination of the curve will depend on the integration of a pure equation of differences. But if, the abscissas still increasing by a constant difference, the connecting property consist in a relation involving such elements as the tangent, the normal, the radius of curvature, &c, the determining equation will be one of mixed differences.
If, instead of the abscissa, some other element of the curve is supposed to increase by a constant difference, it is necessary to assume that element as the independent variable. But when no obvious element of the curve increases by a constant difference, it becomes necessary to assume as independent variable the index of that operation by which we pass from point to point of the curve, i. e. some number which is supposed to measure the frequency of the operation, and which increases by unity as we pass from any point to Then we must endeavour to form two the succeeding point. equations of differences, pure or mixed, one from the law of succession of the points, the other from their connecting property ; and from the integrals eliminate the new variable.
ART.
2.]
GEOMETRICAL APPLICATIONS.
233
There are problems in the expression of which we are led what may be termed functional differential equations, i. e. equations in which the operation of differentiation and an unknown functional operation seem inseparably involved. In some such cases a procedure similar to that employed in the
to
The
one,
may
suffice.
To find a curve such that, if a system of n right 1. originating in a fixed point and terminating in the curve, revolve about that point making always equal angles with each other, their sum shall be invariable. (Herschel's
Ex.
lines,
Examples,
p. 115).
The
angles
made by
some fixed
2(M n
1),r
.
line
may
be represented by
M+Hzr,
n
fl
+ -,.... n
Hence, if r = </> (0) be the polar equation of the curve, the given point being pole, we have
na,
and
let d> T
=u
ti
then
we have
is
cos h (7 2
Attz
...
n +C n
, x
cos
(2n v
2)
n
irz
234
GEOMETRICAL APPLICATIONS.
find
1
[CH. XII.
Hence we
r
= a+C cos0+
(72
{n
1) 0,
being
C=A+B +E
i
cos
sin
+B n6 + i?
n6
cos 2w0
A,
Bv Ev
=a+b
and the curve is seen to possess the property that " if a system of any number of radii terminating in the curve and making equal angles with each other be made to revolve round the origin of co-ordinates their sum will be invariable."
Ex. 2. Required the curve in which, the abscissas increasing by a constant value unity, the subnormals increase in a constant ratio 1 a.
:
scissa
Representing by y x the ordinate corresponding to the abx we shall have the equation of mixed differences
9
dy x_
y * dx
<Wx-x
dx
(!)
Let#
fl '
dx
U*> then
.- '
aux_ i=0;
ux = Ca%
whence
y dx
x
dy x
'
Ca x
(2).
Hence
integrating
we
find
y*=
s/(Cx c f
+ c)
(3),
Cx being a periodical constant which does not vary changes to x -f 1, and c an absolute constant.
when x
ART.
2.]
GEOMETRICAL APPLICATIONS.
235
Ex. 3. Eequired a curve such that a ray of light proceeding from a given point in its plane shall after two reflections by the curve return to the given point.
solution as given
The above problem has been discussed by Biot, whose by Lacroix {Biff, and Int. Calc. Tom. ill.
Assume the given radiant point as origin let x, y be the co-ordinates of the first point of incidence on the curve, and
;
Also
let
p, -~~ =p'.
shewn that twice the angle which the normal makes with the axis of x is equal to the sum of the angles which the incident, and the corresponding reflected ray at that point make with the same
at
any point
of the curve
axis.
Now
the point x,
axis of
is
The tangent
is
_2
p
2p
f
Hence the tangent of the angle which the ray x, y makes with the axis of x is
%P
reflected
from
\-f
i
y x
y
2
2p
1
j? X
Again, by the conditions of the problem a ray incident from the origin upon the point x, y would be reflected in the same
236
straight
GEOMETRICAL APPLICATIONS.
[CH. XII.
line, only in an opposite direction. But the two expressions for the tangent of inclination of the reflected ray
being equal,
-p =Q x (1 -p + 2yp
2xp
2
-y
(1
)
(2):
x (1
-p + 2yp
2
)
Now, regarding x and y as functions of an independent variable z which changes to z + 1 in passing from the first point of incidence to the second, the above equations become
'
17
^(l-/)+2?/p
The
first
2xp-y (l-p
2
2
.
cc(l-p )+2yp
whence by
substitution
; '
Ay = CAx.
Therefore
y=Cx+C.
Here G and are primarily periodic functions of z which do not change when z becomes z + L. Biot observes that, if C be such a function, ( (7), in which the form of < is arbitrary, will also be such, and that we may therefore assume (7'=</>((7),
(/>
whence
V =Cx+<j>(C),
and, restoring to
(4),
its
given in
we
shall
have
ART.
2.]
GEOMETRICAL APPLICATIONS.
237
*
This
is
{}
'
Although Lacroix does not point out any restriction on the form of the function <f>, it is clear that it cannot be quite arbitrary. For if C=ty{z) we should have
y
C"
= W(z),
to
inverse, there
<f>
which
yfr
is
result
= z,
would change when z was changed into z + \. From so that the general form of periodic constants, Chap. II., it is evident that a rational function of such a constant possesses the same character. Thus the differential equation (5) is applicable when cf> indicates a rational function, and generally when it denotes a functional operation which while periodical itself does not affect the periodical character of its subject.
If
we make
0,
we have on
reduction
tf-u?)p + xy{l-f) = 0,
the integral of which
is
x* + f = r ^
denoting a
circle.
* It is only while writing this Chapter that a general interpretation of this equation has occurred to me. Its complete primitive denotes a family of curves defined by the following property, viz. that the caustic into which each of these curves would reflect rays issuing from the origin would be identical with the envelope of the system of straight lines defined by the equation y = cx+<p(c), c being a variable parameter. This interpretation, which is quite irrespective of the form of the function <p, confirms the observation in the text as to the necessity of restricting the form of that function in the problem there discussed. I regret that I have not leisure to pursue the inquiry. I have also ascertained that the differential equation always admits of the fol-
viz.
(y-A) 2 + (x-B) 2 = 0,
and
238
If
2a,
GEOMETRICAL APPLICATIONS.
[CH. XII.
we make we find on
{a?
and equal
to
(y-a) B +
cV=
The following once famous problem engaged in suc3. But the cession the attention of Euler, Biot, and Poisson. subjoined solution, which alone is characterized by unity and completeness, is due to the late Mr Ellis, Cambridge Journal, It will be seen that the problem leads to Vol. in. p. 131. a functional differential equation.
Ex. 4. Determine the class of curves in which the square any normal exceeds the square of the ordinate erected at
of
its foot
by a
constant quantity a.
If
= yfr (x)
,
will be
-fy
(x)
+
-J
The
fw + {^F-4 + ^}=
Differentiating,
a),
we have
which
is
1+^1=0
fW'+fi' + ^l-O
(2),
(3).
ART.
3.]
GEOMETRICAL APPLICATIONS.
of these gives on integration
Tfr(x)
239
The
first
+x =ax + /3
2
9
(4).
find as
hence deduced, in
(1),
we
a
and, supposing this satisfied,
2 2
= 0,
(4)
gives
+ x = ax + /3,
the equation of a circle whose centre is on the axis of x. It is evident that this is a solution of the problem, supposing
= 0.
assume
To
x+
and there
results
(
(x)
=%(),
xyx -2 X (x)+x =
(5).
To
and we have
t+2
t+1
whence
u
t
=C+ C't,
P
P
C and
t
C being
t
into
+ 1.
If
functions which do not change on changing (t) 9 we (t) and we represent them by x
have
(t),
=x
and u t+1
= % (x) x + \ tf (x)
we have
x = P(t)
+ tPx {t),
W(x)^P
(t) 9
240
X16I1C6
GEOMETRICAL APPLICATIONS.
[CH. XII.
f(x)dx = P
^(x) by
(t)
{F()
+ -P.(0
1
+-P/ ()}*.
+tp;(t)}dt.
is
=jplt {P(t) +p
yjr
(t)
Replacing therefore
the two equations,
(x)
by y 2
l
the solution
expressed
x = P(t)
l
+ tP
(t)
{h
assigned,
(t)
a,
(t)
= ft,
thus
making them
constant,
and substituting
e.
e for c
a/3,
f = /3x +
Substituting this in
(1),
we
find
-0"
a = .
real,
negative.
Thus, in order that the solution should be Let a = A 2 then /3 = 2h, and
,
a must be
3/
=2^ + e
This indicates two parabolas.
(7),
a=
0,
to the axis of x.
241
EXEECISES.
1. Find the general equation of curves in which the diameter through the origin is constant in value.
Find the general equation of the curve in which the 2. product of two segments of a straight line drawn through a fixed point in its plane to meet the curve shall be invariable.
3. If in Ex. 4 of the above Chapter the radiant point be supposed infinitely distant, shew that the equation of the reflecting curve will be of the form
-&+(?)
<j>
to.
If a curve be such that a straight line cutting it 4. perpendicularly at one point shall also cut it perpendicularly at another, prove that the differential equation of the curve
will be
y
<j)
= -x +0(
,
tion,
integral of the above differential equais unrestricted, may be interpreted by the system of involutes to the curve which is the envelope of the system of straight lines defined by the equation
(/>
y = mx
<f>
(in),
B. F. D.
16
242
MISCELLANEOUS EXAMPLES.
Selected chiefly from the Senate-House
Examination Papers.
1.
Prove
(1)
/ (A)0- = (. + l) I
/(A)<r" =
(l
^^
2x
O.
(2)
+ A)/(A)0\
shew that
or
2.
If /(*)
= f(\) + J.
f(D)
2x+1
C,
= 0,
f{D)
is
= 0,
taken.
Shew
that
x f(D) a
cos
where
r cos 6
m),
rsm6=f(D)asm(0.m),
in
which
4.
D operates on 0.
Sr = l + 2 +
r r
. . .
If
+x
2rc
r
,
shew
that
2W
1 <fc
0zn+v
n being a positive
5.
integer.
2
. . .
If a
+ a x + a cc +
x
cessive values of
. . .
a ru r x
. .
= I E {flT^ (a*)]
when 2
+2
[a^'H] A^+
-..],
denotes summation with respect to values of a which are roots of the equation o. n -1=0.
MISCELLANEOUS EXAMPLES.
6.
243
If u x
= a + kx,
and z v z 2
...
zm
equation
a-
_ ^- + pJf" _
= Pjr +
2
.
. . .
+ (_ l Tpm = 0,
o
prove that
*w*
+*2
...
ux+Zm
2
{p m _
2
,
a
3
+ Pm_ a
+ Pm _ a
{
+ Fm_ a
3
+ ...)
^
2
. . .
...)
ujr*
r+1
^7 '"^
2
&~
+
7.
If .Bp -53
1
...
1.2
8.
1.2.3.4
= U.
Assuming that
1 1
2
19
log
(1
+ a)
f ?/
12
24
720
shew that
/*
^
</>
(a?)
dx = Ax |- + u
+ w2 +
2
. .
+ 2V* + ^ "
19
8
?/
12
^^ ~ A
)
" (,)
where
(j>
(A
...},
{3 a = nAx,
w1?
. . .
(x) at
=a
x = /3.
9.
Shew
dxn
'"
(for
satisfies the
equation
dz
d
u being any
-z n Az n
function of x.
244
MISCELLANEOUS EXAMPLES.
is inscribed in a fixed circle, the variable arc between one of its angles and a fixed point in the circumference, and z n the ratio, multiplied by a certain constant, of the distances from the centre of the feet of perpendiculars drawn from the nth and (n l) th angles, counting from -4, on the diameter through the fixed point, prove that z n is a function which satisfies the equation.
be moveable, and
x denote
10.
If
<j)(z)
=<j)(x)
(j>(y)
where z
is
f (x)f (y),
245
e(l+t+t + ^A.
2
III.
1.
2*3263359, which
is
- 3v + 10^+ 5v 4 - 2v 5
2
- 2v + 5v +
x
10i? 4
- 3v5
10
'
10
3.
a:
-a:+4.
1
1.
IY.
385.
&c+16
6 (x
2.
9
4' +I
2) (x
3) (x
4)
2
3.
+ -=3
aj-1 x + 2'
"
1
5.
-acosO +
a*"
cos (x
1) 6
cos
xB
7.
Assume
for the
(A +
Bx
J!
...
+Mx
.
n~
)s*
'
uu
J.-+1
.,
...
ux+m
cot
3-
cot (2*
_1
0).
1 1.
Refer to
p. 62.
smm(2sc 1)
2 sin
cos
.,
cosra(2a?)
(2 sin
f/
sinwi(2aj + l) A2#/
(2 sin
rv
'
m)
^v
'
m)
.
(a?)
w + -j^-4
.
sin ra(2a;+ 3) 4 7 A4
(2 sin
m)
r^-
<
(x) 7
+ &c.
246
V.
7T
-
ANSWERS TO EXERCISES.
1.
7T
- and
a 2
o a \2
(
- tan -1 -
aJ
which,
if
a =
1,
reduce to
and-.
VI.
74854709.
1.
2.
2567-6046.
5
-
27F
37r
472 i
+&C
VII.
1.
R
<
2nd.
The same.
1/
3rd.
TV
= >
Am
2
+cAw_- ux -
0.
4th.
0.
5th.
The same.
2.
3.
Ux = Cl
(-iy + ca (iy +
_
3m
qa {x
1
~ 1)2
- ap
K
5.
mx =
cosw(aj~ V
1) ;
acoswoj =
n+
1
a 2 2a
cos
Ca'.
6.
^=(o +
Note that
oia!)(-2)- +
|-A.
7.
this
may
a?
ux
The
3 sin
(ttx)
(x
1)}
ux _ 2 =
0.
final integral is
ux =
+c
(2)"}.
ANSWERS TO EXERCISES.
n
,
247
,
1W/
x(xl)(x 2)
1X
10
3 w_= (m + n2) 2 v
{ c,
cos
a?
tan"
mj
+ c2
sin
# tan" M
1
n.
13.
w,
2*-c
it first
Express
(Z>
- or*)
(Z)
-<*) =
<).
14.
ux =a
{c+ c'%a
it
2
}.
15.
Express
a){D
sin
a~
x
)
0.
16.
ux = Ja tan
I
1
cos
+ C
17.
unity,
C are
A s + B3 + C3 =3A(a+JBC).
18.
last
example.
VIII.
primitive
is
1.
y = ~first
x~^^-(-iy+C.
a 2*+
3.
The complete
of the
4.
y = C2 +
2.
(-a) 0? +a
1
v '
(l+a) 2
'
IX.
X.
2.
Yes.
1.
ux = c + c 2 (-l) x +mx,
1
vx
= c -c 2 (~
1
l)
-m (x +
1).
M. = Co + Cl (-l)w,,
* = ,-,, (-If.
~
.
_
3.
=^ +
^cos
7r^\
- x_
^-
248
ANSWERS TO EXERCISES.
(
wx = A - B cos
I
7r(x+l)
'-
(7 sin
^ (- i^ 1)*+ (05+1)1,
7T
' \
(1 v
m - 2n) x
2
^ = ^ + (^cos"^3
+2>
+ CBin
ff
))(-l)- +
^ + ^3
a?
.
5.
mx> v =
a" (
roots of
6.
j ^
b
(y)
+ 0*
f^j
cj>
(y),
where a and
/5
are
am +
0.
w fJ ,= o;(y
+ a;-l) + ^(y+a;).
2.
n
<j>
(x)
=\
{(x
Voj
- If + (a: - Jx^^lf}.
(m = 2 n ).
5.
f(x) = cx.
/() = f +
2/
6.
y +
^
x.
7.
+ ()-*).
,
8.
ce*
(x_a)
<(#)
12.
cj>
(x)
= tan mx.
= sec mx.
if/
(x)
MISCELLANEOUS EXAMPLES.
7. For some remarks on this problem see Quarterly Journal 8. See De Morgan's Calculus, of Mathematics, Vol. ill. p. 262. m 10. Ch. xiii. Arts. 61, 67. 4 (x) = {f(x)}
.
CAMBRIDGE: PRINTED BY
C. J.
23,
W.C.
an& Cambridge.
MACMILLAN AND
CO.'S.
DESCRIPTIVE CATALOGUE OP
The
Works in
this
Series
of Cambridge
Class-Books for
which
have
been
the use of
Schools and
Colleges,
are intended
the
of Education,
from
to
most
Elementary
most Advanced,
and
Of
large
and
A SERIES
Catalogue.
is
in course of
this
publication, a
6000.
16
64.
Kingdom and
2.
An
By H.
J. ROBY, M.A., Under Master of Dulwich College Upper School, late Fellow and Classical Lecturer of St. John's College,
Cambridge.
3.
An
Elementary History of
St. Catharine's College,
the
Book of Common
Prayer.
Cambridge
4.
the
Church,
Holy
of the Collection and Reception of the Scriptures in the Christian Churches. 18mo. 4s. 6d.
5.
of the Ancients, prepared to be rendered into Latin Verse, for Schools. By F. C. HODGSON, B.D., late Provost of Eton College. New Edition. Revised by F. C. HODGSON, M.A., Fellow of King's College, Cambridge. 18mo. 3s.
6.
By
7.
18mo.
2s. 6d.
A Key
to this
work
By
8.
The School Class Book of Arithmetic, BARNARD SMITH, M.A., late Fellow of St. Peter's
College,
Cambridge.
The Bible Word-Book. A Glossary of old English Bible Words with Illustrations. By J. EASTWOOD, M.A., St. John's College, Cambridge, and Incumbent of Hope in Hanley, Stafford, and W. ALDIS WRIGHT.
M.A., Trinity College, Cambridge.
[In the Press.
9.
For the Brief Biographical Dictionary, Schools and General Readers. By the Rev. CHARLES HOLE.
1
8mo.
***
BARNARD SMITH,
Cambridge.
M.A.
1.
2.
Arithmetic
the
Arithmetic
&
Algebra
For
Use
of Schools.
Edition (1862) 348 pp. Crown 8vo. strongly bound in cloth, 4s. 6d. Answers to all the Questions.
New
Ninth Edit. 696 pp. (1863). Cr. 8vo. 10s. (id. strongly bound in cloth.
The first edition of this work was published in 1854. It was primarily intended for
3.
Key
Crown
8vo.
8s. %d.
To meet a widely expressed wish, the the use of students at the Universities, and for Schools -which prepare for the ARITHMETIC was published separately the larger work in 1854, with so from found however been Universities. It has to meet the requirements of a much much alteration as was necessary to make independent of the ALGEBRA. It quite it used larger class, and is now extensively in Schools and Colleges both at home and has now a very large sale in all classes of at home and in the Colonies. A Schools found It has also been in the Colonies. of great service for students preparing copious collection of Examples, under rule, has been embodied in the work each for the Middle-Class and Civil and Military Service Examinations, from in a systematic order, and a Collection of Miscellaneous Papers in all branches of the care that has been taken to elucidate the principles of all the Rules. Testi- Arithmetic is appended to the book. mony of its excellence has been borne by some of the highest practical and theoof which the follow- 4. retical authorities ing from the late DEAN PEACOCK may metic. 104 pp. Cr. 8vo. (1860) be taken as a specimen 2s. Or with Answers, 2s. %d.
;
Exercises in Arith-
"Mr. Smith's Work is a most useful publication. The Rules are stated with great clearness. The Examples are well selected and worked out with just sufficient detail without being encumbered by too minute explanations ; and there prevails throughout it that just proportion of theory and practice, which is the crowning excellence of an elementary work."
each.
Answers,
6d.
These
EXERCISES
The greater
carefully compiled
by ISAAC
TODEUNTER, M.A.
St.
F.R.S.
Algebra.
Colleges
3.
Schools.
Spherical Trigonometry.
the Use of Colleges
and
For
and
Schools.
Third Edition. 542 pp. (1862). Strongly bound in cloth. 7*. Gd.
This work contains all the propositions which are usually included in elementary treatises on Algebra, and a large number of Examples for Exercise. The author
132
pp.
Crown
8vo.
Second
4*.
Edit, enlarged.
(1863).
6d.
has sought to render tbe work easily intelligible to students without impairing the
accuracy of the demonstrations, or contracting the limits of the subject. The Examples have been selected with a view to illustrate every part of the subject, and as the number of them is about Sixteen hundred and fifty, it is hoped they will supply ample exercise for the student. Each set of Examples has been carefully arranged, commencing with very simple exercises, and proceeding gradually to
those which are less obvious.
This work is constructed on the same plan as the Treatise on Plane Trigonometry, to which it is intended as a sequel. Considerable labour has been expended on the text in order to render it comprehensive and accurate, and the Examples, which have been chiefly selected from University and College Papers, have all been
carefully verified.
Comprising the First Six Books and Portions of the Eleventh and Twklfth Books, with Notes, Appendix, and Exercises.
2.
Plane Trigonometry
For Schools and
Colleges.
384 pp. 18mo. bound. (1862). 3s. 6d. As the Elements of Euclid are usually placed in the hands of young students, it is important to exhibit the work in such a form as will assist them in overcoming
the difficulties which they experience on
their first introduction to processes of continuous argument. No method appears to be so useful as that of breaking up the demonstrations into their constituent parts, Strongly bound in cloth. 5s. and this plan has been adopted in the The design of this work has been to ren- present edition. Each distinct assertion in der the subject intelligible to beginners, the argument begins a new bine ; and at and at the same time to afford the student the end of the lines are placed the necessary the opportunity of obtaining all the infor- references to the preceding principles on mation wbich he will require on this branch which the assertions depend. The longer of Mathematics. Each chapter is followed propositions are distributed into subordiby a set of Examples ; those which are nate parts, which are distinguished by entitled Miscellaneous Examples, together breaks at the beginning of the lines. The with a few in some of the other sets, may Notes are intended to indicate and explain oe advantageously reserved by the student the principal difficulties, and to supply the for exercise after he has made some pro- most important inferences which can be gress in the subject. As the Text and Ex- drawn from the propositions. The work amples have been tested by considerable finishes with a collection of Six hundred experience in teaching, the hope is enter- and twenty-five Exercises, which have been tained that they will be suitable for impart- selected principally from Cambridge Exing a sound and comprehensive knowledge amination papers and have been tested by of Plane Trigonometry, together with long experience. As far as possible they readiness in the application of this know- are arranged in order of difficulty. The ledge to the solution of problems. In the Figures will be found to be large and disSocond Edition the hints for the solution tinct, and have been repeated when necesof the Examples have been considerably sary, so that they always occur in immediincreased. ate connexion with the corresponding text.
2ndEdit.
279pp. (1860).
Crn. 8vo.
TODHUNTER,
M.A., F.R.S
-continued.
Applications.
Crown
This work is intended to exhibit a comprehensive view of the Differential Calcu10s. %d. Crown 8vo. cloth. lus on the method of Limits. In the more In writing the present Treatise on the elementary portions, explanations have Integral Calculus, the object has been to been given in considerable detail, with produce a work at once elementary and the hope that a reader who is without the complete adapted for the use of beginners, assistance of a tutor may be enabled to acand sufficient for the wants of advanced quire a competent acquaintance with the students. In the selection of the propo- subject. More than one investigation of sitions, and in the mode of establishing a theorem has been frequently given, them, the author has endeavoured to ex- because it is believed that the student dehibit fully and clearly the principles of rives advantage from viewing the same the subject, and to illustrate all their most proposition under different aspects, and important results. In order that the stu- that in order to succeed in the examinadent may find in the volume all that he tions which he may have to undergo, he requires, a large collection of Examples should be prepared' for a considerable vafor exercise has been appended to the riety in the order of arranging the several branches of the subject, and for a corresdifferent chapters. ponding variety in the mode of demonstra6.
Analytical Statics.
With numerous Examples.
330 pp. (1858).
10s. 6d.
tion.
Second Edition.
9.
Crown
8vo. cloth.
Geometry
AS APPLIED TO THE STRAIGHT LINE
In this work will be found all the propositions which usually appear in treatises on Theoretical Statics. To the different chapters Examples are appended, which have been selected principally from the University and College Examination Papers these will furnish ample exercise in
;
Cm.
8vo.
cl.
7s. 6d.
76 pp. (1858).
Cm.
8vo. cloth.
4s.
examples in illustration of Analytical Geometry of Three Dimensions has long been required both by students and teachers, and the present work is published with the view of supplying the want.
collection of
By
J.
H. PRATT, M,A.
Theory of Equations.
With a
Collection
A
La
of Examples.
(1861).
Treatise on Attractions,
and
the Figure
Crown
8vo. cloth.
279 pp.
Place's Functions,
7s. 6d.
of the Earth.
Second Edition. Crown 8vo. 126 pp. all the proposiCloth. 6s. 6d. (1861). which are usually included in eleof Equathe Theory mentary treatises on In the present Treatise the author has tions, together with a collection of Ex- endeavoured to supply the want of a work amples for exercise. This work may in on a subject of great importance and high fact be regarded as a sequel to that on interest La Place's Coefficients and FuncAlgebra by the same writer, and accord- tions and the calculation of the Figure of ingly the student has occasionally been the Earth by means of his remarkable anareferred to the treatise on Algebra for pre- lysis. No student of the higher branches liminary information on some topics here of Physical Astronomy should be ignorant The work includes three of La Place's analysis and its result "a discussed. chapters on Determinants. calculus," says Airy, "the most singular in its nature and the most powerful in its
This treatise contains
tions
11.
Theory of Errors of
Observations
And
the Combination of Observations.
Calculus of Variations
During
8vo. cloth.
By
G. B.
AIRY, M.A.
12s.
Astronomer Royal.
It is of importance that those who wish cultivate any subject may be able to ascertain what results have already been obtained, and thus reserve their strength
to
Crown
In order to spare astronomers and observers in natural philosophy the confusion and loss of time which are produced by for difficulties which have not yet been referring to the ordinary treatises emconquered. The Author has endeavoured bracing both branches of Probabilities, the in this work to ascertain distinctly what author has thought it desirable to draw has been effected in the Progress of the up this work, relating only to Errors of Calculus, and to form some estimate of Observation, and to the rules derivable the manner in which it has been effected. from the consideration of these Errors, for the Combination of the Results of Observations. The Author has thus also the advantage of entering somewhat more fully into several points of interest to the observer, than can possibly be done in a 12. General Theory of Probabilities.
Algebra
for
Beginners.
18mo.
this
AN ELEMENTARY TREATISE ON
2s. 6d.
cloth,
work
intelligible to
by the use
of simple language
H. H.
St.
CHEYNE,
B.A.
6s. 6d.
Scholar of
By
W.
J.
STEELE, B.A.
Differential Equations
468 pp. (1859). Cm. 8vo. cloth. The Author has endeavoured in
145.
this
treatise to convey as complete an account of the present state of knowledge on the subject of Differential Equations as was consistent with the idea of a work intended, primarily, for elementary instruction. The object has been first of all to meet the wants of those who had no previous acquaintance with the subject, and also not quite to disappoint others who might seek for more advanced information. The earlier sections of each chapter contain that kind of matter which has usually been thought suitable for the beginner, while the latter ones are devoted either to an account of recent discovery, or to the discussion of such deeper questions of principle as are Likely to present themselves to the reflective student in connection with the methods and processes of his previous course.
Dynamics
of a Particle.
cl.
In this Treatise will be found all the ordinary propositions connected with the Dynamics of Particles which can be conveniently deduced without the use of D'Alembert's Principles. Throughout the book will be found a number of illustrative Examples introduced in the text, and for the most part completely worked out others, with occasional solutions or
;
each Chapter.
By
the Rev. G. F.
Mathematical
South
2.
of Finite Differences.
Crown
8vo. doth.
10s. 6d.
The Calculus
its
In this work particular attention has been paid to the connexion of the methods with those of the Differential Calculus connexion which in some instances involves far more than a merely formal analogy. The work is in some measure designed as a sequel to the Author's Treatise on Differential Equations, and it has been composed on the same plan.
indicates, object is to develope peculiarities in the Ellipsoid; and further, to establish analogous properties in unlimited congeneric series of which this remarkable surface is a constituent.
By
J.
B.
PHEAR, M.A.
Hydrostatics
Elementary
Elementary
By
the Rev.
Solutions.
Third Edition.
Crown
8vo. cloth.
;
the descriptions
(150 pp.) 1860. Cm. 8vo. cl. is. 6d. hance its value." English Journal of This work is published under the au- Education. thority of H. M. Secretary of State for This Edition contains 147 Examples, and India for use in the Government Schools solutions to all these examples are given and Colleges in India. at the end of the book.
and explanations are sufficiently full and intelligible ; the investigations are simple and scientific. The examples greatly en-
PARKINSON,
B.D.
By
R. D.
BEASLEY, M.A.
of
Head Master
Grantham
School.
AN ELEMENTARY TREATISE ON
With a
of
Collection
of Examples.
Examples.
Crown
8vo. cloth.
9s. 6d.
The Author has endeavoured to render This Treatise is specially intended for the present volume suitable as a Manual use in Schools. The choice of matter has for the junior classes in Universities and been chiefly guided by the requirements the higher classes in Schools. With this of the three days' Examination at Camobject there have been included in it those portions of theoretical Mechanics which can be conveniently investigated without the Differential Calculus, and with one or two short exceptions the student is not presumed to require a knowledge of any branches of Mathematics beyond the elements of Algebra, Geometry, and Trigonometry. A collection of Problems and
bridge, with the exception of proportional parts in logarithms, which have been omitted. About Four hundred Examples have been added, mainly collected from the Examination Papers of the last ten years, and great pains bave been taken to exclude from the body of the work any which might dishearten a beginner by their difficulty.
Examples has been added, chiefly taken from the Senate-House and College Examination Papers which will be found
useful as an exercise for the student. In the Second Edition several additional propositions have been incorporated in the work for the purpose of rendering it more complete, and the Collection of Examples and Problems has been largely increased.
By
St.
J.
BROOK SMITE,
M.A.
Arithmetic in Theory
and Practice.
2.
A Treatise on Optics
10s. 6d.
Part
I.
Crown
8vo. cloth.
first
3s. 6d.
A collection of Examples and Problems has been appended to this work which are sufficiently numerous and varied
in character to for the student
:
afford useful exercise for the greater part of to the Exset in the University and
last
part of a Treatise on Arithmetic, in which the Author has endeavoured, from very simple principles, to explain in a full and satisfactory manner all the important processes in that
subject.
twenty
Subjoined to the copious Table of Contents the author has ventured to indicate an elementary course of reading not unsuitable for the requirements of the First Three Days in the Cambridge Senate
The proofs have in all cases been given form entirely arithmetical: for the author does not think that recourse ought to be had to Algebra until the arithmetical proof has become hopelessly long and perin a
plexing.
At the end of every chapter several examples have been worked out at length,
in
House Examinations.
which the best practical methods of operation have been carefully pointed out.
PUCKLE, M.A.
Windermere
College.
The
Principal of
arranged.
Second Edition.
Crown
8vo.
This book has been written with special reference to those difficulties and misapprehensions which commonly beset the
student
when he commences.
With
this
object in view, the earlier part of the subject has been dwelt on at length, and geometrical and numerical illustrations of the analysis have been introduced. The Examples appended to each section are mostly of an elementary description. The work will, it is hoped, be found to contain all that is required by the upper classes of schools and by the generality of students at the Universities.
The specific features of this annual publication will be obvious at a glance, and its value to teachers engaged in preparing students for, and to parents who are sending their sons to, the University, and to the public generally, will be clear. 1. The whole mode of proceeding in entering a student at the University and at any particular College is stated. 2. The course of the studies as regulated
by the University examinations, the manner of these examinations, and the specific subjects and times for the year 1863, are
given.
3. complete account of all Scholarships and Exhibitions at the several Coland the means by which they are gained. brief summary of all Graces of the 4. Senate, Degrees conferred during the year 1861, and University news generally are given.
By
M.A.
St. Peter's
Dynamics of a System
of Rigid Bodies.
With numerous Examples.
336 pp. (1860).
Contents
5. The Regulations for the Locat, Examination of those who are not members of the University, to be held this year, with the names of the books on which the Examination will be based, and the date on which the Examination will be held.
By N. M. FERRERS, M.A.
Fellow and Mathematical Lecturer of Gonville and Caius College, Cambridge.
Crown
8vo. cloth.
10*. 6d.
AN ELEMENTARY TREATISE ON
Chap. I. Of Moments of Inertia. II. D'Alembert's Principle. III. Motion about a Fixed Axis. IV. Motion in Two Dimensions. V. Motion of a Rigid Body in Three Dimensions. VI. Motion of a Flexible String. VII. Motion of a System of Rigid Bodies. VIII. Of Impulsive Forces. IX. Miscellaneous Examples.
Trilinear Co-Ordinates
The Method of Reciprocal Polars, and the Theory of Projections.
154 pp. (1161).
Cr. 8vo.
cl.
6s. 6d.
The numerous Examples which will be found at the end of each chapter have been chiefly selected from the Examination Papers set in the University and Colleges of Cambridge during the last few
years.
The object of the Author in writing on this subject has mainly been to place it on a basis altogether independent of the
ordinary Cartesian System, instead of regarding it as only a special form of abridged Notation. A short chapter on Determinants has been introduced.
10
SNOWBALL, M.A.
St.
Late Fellow of
8vo.
8vo.
M.
By Jameson.
and
By Walton
Mackenzie.
8vo.
Tenth Edition.
240
pp.
7*. 6d.
(1863).
10s. 6d.
Crown
In preparing a
8vo.
1857.
8*.
new
1860.
of some of the more important propositions have been rendered more strict and general and a considerable addition of more than Two hundred Examples, taken principally from the questions in the Examinations of Colleges and the University, has been made to the collection of Examples
;
8vo.
The above books contain Problems and Examples which have been set in the
Cambridge Senate-house Examinations at various periods during the last twelve years, together with Solutions of the same. The Solutions are in all cases given by the Examiners themselves or under their
sanction.
and Problems
for practice.
By W.
EC.
DREW,
M.A.
By
EC.
A.
MORGAN,
M.A.
Collection of Mathe-
Examples.
With Answers.
190 pp. (1858).
Second Edition.
Crown
8vo. cloth.
4s. 6d.
Crown
8vo.
6s. 6d.
In this work the subject of Conic Sections has been placed before the student in such a form that, it is hoped, after mastering the elements of Euclid, he may find it an easy and interesting continuation of his geometrical studies. With a view also of rendering the work a complete Manual of what is required at the Universities, there have been either embodied into the text, or inserted among the examples, every book work question, problem, and rider, which has been proposed in the Cambridge examinations up to the present time.
This book contains a number of problems, chiefly elementary, in the Mathematical subjects usually read at Cambridge. They have been selected from the papers set during late years at Jesus College. Very few of them are to be met with in other collections, and by far the larger number are due to some of the most distinguished Mathematicians in the University.
By
C.
TAYLOR,
B.A.
Geometrical Conies.
Solutions to the Problems in Drew's Conic Sections.
Crown
8vo. cloth,
is. 6d.
222 pp. 1863. Crown 8vo. Is. 6d. This work contains elementary proofs of
the principal properties of Conic Sections, together with chapters on Projection and Anharmonic Ratio.
11
M.A.
St.
Geometry.
M.A., 1863.
John's
Juvenal.
10s. 6d.
JOSEPH WOLSTENHOLME,
Christ's Coll. Cambridge.
With English Notes and an Index. 464 pp. (1854). Crown 8vo. cloth.
"
cloth. 18s.
The authors have endeavoured to present before students as comprehensive a view of the subject as possible. Intending as they miliar knowledge of Continental critihave done to make the subject accessible, cism, ancient and modern, is unsurpassed, at least in the earlier portion, to all classes we do not say among English School-books, of students, they have endeavoured to ex- but among English editions generally." plain fully all tbe processes which are Edinburgh Review. most useful in dealing with ordinary theorems and problems, thus directing the 2. student to the selection of methods which are best adapted to the exigencies of each problem. In the more difficult portions of With English Notes. the subject, they have considered themselves to be addressing a higher class of 168 pp. (1861). Fcp. 8vo. cloth. 5s. students; there they have tried to lay The Text is that of Halm's 2nd edition, a good foundation on which to build, if (Leipzig, Weidmann, 1858), with some any reader should wish to pursue the corrections from Madvig's 4th Edition science beyond the limits to which the (Copenhagen, 1858). Halm's Introduction work extends. has been closely translated, with some additions. His notes have been curtailed, omitted, or enlarged, at discretion; passages to which he gives a bare reference, are for the most part printed at length for the Greek extracts an English version has been substituted. A large body of
for really ripe scholarship, extensive acquaintance with Latin literature, and fa-
Cicero's
Second Philippic.
Mythology
for Versification.
and
historical,
Brief Sketch of the Fables of the Ancients, prepared to be rendered into Latin Verse for Schools.
By
F. O.
HODGSON,
B.D.,
has been added from various sources. A list of books useful to the student of Cicero, a copious Argument, and an Index to the introduction and notes, complete the book.
New
of
18mo.
3s.
The Author here offers to those who are engaged in Classical Education a further help to the composition of Latin Verse, combined with a brief introduction to an essential part of the study of the Classics. The Author has made it as easy as he could so that a boy may get rapidly through these preparatory exercises and thus having mastered the
:
HALE PUCKLE,
of
M.A.
College.
Head Master
Windermere
Is.
Fcp. 8vo.
first
difficulties,
he
composition.
12
An
Grammar.
By H.
Under -Master
J.
ROBY, M.A.,
Dulwich College Upper School, late Fellow and Classical Lecturer of St. John's College, Cambridge.
of
18mo.
2s.
M.
The Author's experience in practical teaching has induced an attempt to treat Latin Grammar in a more precise and intelligible way than has been usual in school books. The facts have been derived from the best authorities, especially Madvig's Grammar and other works. The
works also of Lachmann, Ritschl, Key, and others have been consulted on special points. The accidence and prosody have
been simplified and restricted to what is really required by boys. In the Syntax an analysis of sentences has been given, and the uses of the different cases, tenses and moods briefly but carefully described. Particular attention has been paid to a classification of the uses of the subjunctive mood, to the prepositions, the oratio obliqua, and such sentences as are intro-
Second Edition. To which is prefixed iEsCHLNES AGAINST CtESIphon. With English Notes.
287 pp. (1860). The first edition
edition of
Fcap. 8vo.
cl.
5s.
met with considerable acceptance in various Schools, and a new edition being called for, the Oration of iEschines against Ctesiphon, in accordance with the wishes of many teachers, has been appended with useful notes by a competent scholar.
2.
.ZEJschyli
Eumenides
7*. 6d.
With English
8vo.
lar is
duced by the English that.' Appendices treat of the Latin forms of Greek nouns, abbreviations, dates, money, &c. The
'
known and
Grammar is
written in English.
of the
him
Book
For
the
of
Common
Prayer.
Use of Schools and popular
reading.
flowing and melodious, elegant and scholarlike. The Greek Text is well printed the notes are clear and useful." Guardian.
4
By
MERIVALE,
Sallust.
With English Notes.
to give a popular abridgement of his larger work in a form which should be suited for use in Schools and for general readers, has attempted in this book to trace the History of the Prayer-Book, and to supply to the English reader the general results which in the larger work are accompanied by elaborate discussions and references to authorities indispensable to the student. It is hoped that this book may form a useful manual to assist people generally to a more intelligent use of the Forms of our Common Prayer.
urged
18mo. cloth. 2s. 6d. Second Edition. 172 pp. (1858). The Author having been frequently 4*. 6d. Fcap. 8vo.
" This School edition of Sallust
cisely
is
pre-
what the School edition of a Latin author ought to be. No useless words are spent in it, and no words that could be of use are spared. The text has been carefully collated with the best editions. With the work is given a full current of extremely well-selected annotations."
The Examiner.
The " Catilina" and " Jugurtha" may had separately, price 2s. 6d. each,
cloth.
13
WRIGHT, M.A.
By
EDWARD
of
THR1NG, M.A.
Uppingham
School.
Head Master
Head Master
Crown
Demy
2.
136 18mo.
pp.
2s.
(1860).
Never was there a hetter aid offered alike to teacher and scholar in that arduThe style is at once familiar ous pass. and strikingly simple and lucid ; and the explanations precisely hit the difficulties,
and thoroughly explain them." English
New
Journal of Education.
2.
Demy
Hellenica.
A FIRST GREEK READING BOOK. sions into the byeways and eccentricities With Vocabulary. of language. This Work took its rise Second Edit. from questionings in National Schools, Fcap. 8vo. cl. 3s. 6d. and the whole of the first part is merely
In the last twenty chapters of this volume, Thucydides sketches the rise and progress of the Athenian Empire in so clear a style and in such simple language, that the author doubts whether any easier or more instructive passages can be selected for the use of the pupil who is commencing Greek. 3.
the writing out in order the answers to questions which have been used already with success. Its success, not only in National Schools, from practical work in which it took its rise, but also in classical
schools,
is full
The Author's effort in these two books has been to point out the broad, beaten, every-day path, carefully avoiding digres-
of encouragement.
3.
School Songs.
SCHOOLS.
THRING
7s. 6d.
and
Third Edit.
3s.
H. RICCIUS.
This work is intended to supply the pupil with an easy Construing-book, which may, at the same time, be made the vehicle for instructing him in the rules of grammar and principles of composition. Here Livy tells his own pleasant stories in his own pleasant words. Let Livy be the master to teach a boy Latin, not some English collector of sentences, and he will not be found a dull one.
Music
Size.
By
5.
EDWARD TURING,
M.A.
Latin Gradual.
First Latin Construing Book for Beginners. 167 pp. 1864. 18mo. cloth. 2s. 0>d.
this little
work has
Home."
2s. 6d.
:
%*
The Vocabulary and Exercises may also be had bound tip with " The Seven Kings of Rome." 5s. cloth.
led to considerable changes in the way of working this out. The intention is to supply by easy steps a knowledge of Grammar, combined with a good vocabulary ; in a word, a book which will not require to be forgotten again as the learner advances. A short practical manual of common Mood constructions, with their English equivalents form the second part.
14
By
4.
EDWARD
THRING, M.A.
On
By
Confirmation.
struing' Book.
"With Suitable Prayers.
This Construing Book is drawn up on the same sort of graduated scale as the Passages Author's English Grammar. out of the best Latin Poets are gradually
C. J.
YAVGKAN,
Is. 6d.
D.D.
5th Edition.
70 pp. (1862).
8vo.
Fcp.
up into their perfect shape. The few words altered, or inserted as the passbuilt
ages go on, are printed in Italics. It is this plan that the learner, whilst acquiring the rudiments of language, may 6tore his mind with good poetry and a good vocabulary.
hoped by
This work, originally prepared for the use of Harrow School, is published in the belief that it may assist the labours of those who are engaged in preparing candidates for Confirmation, and who find it
hand upon any one book of suitable instruction at once suffidifficult to lay their
By
Vicar
of
C.
J.
YATJGBAN, D.D.
in
St.
ciently full to furnish a synopsis of the subject, and sufficiently elastic to give free scope to the individual judgment in the use of it. It will also be found a handbook for those who are being pi-epared, as presenting in a compact form the very points whieh a lecturer would wish hi* hearers to remember.
8vo. cloth.
By dedicating this work to his elder Pupils at Harrow, the Author hopes that
sufficiently indicates what is and what He desires is not to be looked for in it. to record his impression, derived from the
Hand-Book
Is. Gd.
he
experience of many years, that the Epistles of the New Testament, no less than the Gospels, are capable of furnishing useful and solid instruction to the highest If they are classes of our Public Schools. taught accurately, not controversially positively, not negatively; authoritataught with tively, yet not dogmatically close and constant reference to their literal meaning, to the connexion of their parts, to the sequence of their argument, as well
;
History
of
the
Christian
Church during- the First Three Centuries, and the Reformation in England. By W. SIMPSON, M.A. Fourth
Edition.
3s. Gd.
as to their moral and spiritual instruction ; they will interest, they will inform, they will elevate ; they will inspire a reverence for Scripture never to be discarded, they will awaken a desire to drink more deeply of the Word of God, certain hereafter to be gratified and fulfilled.
18mo.
3s. Gd.
15
Manuals
1.
Church during the Middle Ages. By Archdeacon HARDWICK. Second Edition. 482 pp.
(1861).
cloth.
8vo.
has been reopened with great learning and accurate research, and it is mainly with the view of epitomizing their extensive publications, and correcting by their help the errors and misconceptions which had obtained currency, that the present volume has been put together.
This Volume
claims to be regarded as
treatise
4.
an
integral
and independent
on
the Mediaeval Church. The History commences with the time of Gregory the Great, to the year 1520, the year when Luther,
History of the Canon of the New Testament during the First Four Centuries.
By
having been extruded from those Churches 594 pp. (1855). COTT, M.A. that adhered to the Communion of the Crown 8vo. cloth. 12s. 6d. Pope, established a provisional form of government and opened afresh era in the The Author has endeavoured to connect history of Europe. the history of the New Testament Canon with the growth and consolidation of the Church, and to point out the relation 2. History of the Christian existing between the amount of evidence Church during the Refor- for the authenticity of its component parts and the whole mass of Christian literature. mation. By Archdn. of inquiry will convey both WICK. 459 pp. (1856). Crown Such a method the truest notion of the connexion of the 8vo. cloth. 10s. 6d. written Word with the living Body of This Work forms a Sequel to the Au- Christ, and the surest conviction of its thor's Book on The Middle Ages. The divine authority. Author's wish has been to give the reader a trustworthy version of those stirring 5. Introduction to the Study incidents which mark the Reformation of the
HARD-
period.
GOSPELS. ByBROOKE
458
cloth.
mon Prayer.
of
its
pp.
10s.
(1860).
Crown
8vo.
With
a Rationale
Offices.
By
FRANCIS
Fifth Edition.
cloth.
U.
PROCTER, M. A.
10s. %d.
In the course of the last twenty years the whole question of liturgical knowledge
This book is intended to be an Introduction to the Study of the Gospels. In a subject which involves so vast a literature much must have been overlooked ; but the Author has made it a point at least to study the researches of the great writers, and consciously to neglect none.
with
T.
Vignette
Titles
by
f^^^^=:^^S^ W(^^/U^Y^3J
TO"^^V*^>^^m1
Woolner,
J.
W. Ilohnan
P lain
7s
-
6d
Hunt,
V'^KvrtW'
Morocco exVolume.
by F. T.
PALGRAVE.
a Vignette by T. "Woolner.
POETS.
by
COVENTRY PATMORE.
crown 8vo.
morocco,
10s. 6d.
by
ROUNDELL PALMER.
T.
Woolner.
BACON'S ESSAYS
With Notes and
And
Glossarial Index,
<&ambribcje.
t#j