Kadomtsev Petviashvili - On The Stability of Solitary Waves - KP - 1970

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

SOV I E T PHYSICS - DOKLA D Y VOL . 1 5 , NO.

6 D ECEMBE R , 1 970

FLUID MECHANICS

ON THE STABILITY OF SOLITARY WAVES


IN WEAKLY DISPERSING MEDIA
B. B. Kadomtsev• and V. l. Petviashvili
Translated from Doklady Akademi! Nauk SSSR, Vol. 1 92 , No. 4 ,
pp. 753-756 , June , 1970
Or iginal a rticle submitted Febr uary 9, 1970

A wide r ange of one-dimensi.onal, no nlinear 1m =3(chV2) -ª. (4)


waves in weakly dispers ing m edia (e.g. , waves in
shallow water, ionoacoustic and magnetoacous- A solit ary wa ve is a oce-di.mensional nonlin-
tic waves i.n plasmas, etc.), a s . is known from [l -
ear wa ve which b y v irtue o f the one-di.mensional i-
3l. are described by the Korteweg and de Vr ies ty is fully s t a ble . However , a question ar ises
equatio n wheth e r this . s tability is preserved for a soli tary
wave with a we ak beoding distortíon whose amplí -
éiu i at + u au. J ax + f>3u 1éir' = O. (1) tude a and phase ~ are slowly varyio.g functions.
of the coordinate y, measured transversely w ith
This equation d escribes the development of a wa ve
res pect to the s o litary wave •s p ropagat ion direc-
in a syst em of coor dlnates that moves with the
tion. The possibilíty of developi.o.g a self-focussing
propagation velocity e of a long- wavelength pe r-
type of instability in periodíc nonlinear waves in-
turbation (fo r c larity w e shall call this veloc i.ty the
di cates the necessity of studying the analogous ef-
velocity of sound). F o r wa ves in shallow water,
fect for tbe c ase of a solit ary wave.
or in pla s ma, u can l"epresent a pe rturbation of
For c ases that are not strictly e ne-dimension-

-
.;
;:;~~.,:.
· -·•,,, •. :·
...: .
veiocity o r p r es s ure (for a simple wave they are
uniquely interdepe nde nt) .
Equation (1) c an b e equally we ll applled to m e-
dia with ne gative o r pos itive dis per sion. ln the
al, Eq. (1) no longer applies ; however, if the y-
coordinate d epende nce is weak, Eq. (1) can be
corrected by adding a small quantity o<p / &y:
forrner case the phase velocity of the longitudinal éiu ¡ ot + au i ax + iJlu ! ari = oq¡ ! fh¡.
1.1 (5)
waves <lim ini shes with the wave numbe r whereas
in the latter case the s ituation is reversed. The The form o f the p e rturbation is easily e stab-
di ffere nce is t hat in t he firs t case the coo rdinate lished by cons idering the limiting c ase of a two -
x is measured io t he di r ectlon of wave propaga- d imensional wave of small ampl itude u = exp (-iwt +
t ion and in the latter c:is e it is measured in the ikr) and sm all wavelength (s ic) in the x direction,
revers e d irection [2!. kx « 1. In this c ase, th e se cond and thir-d terms in
The Ko rteweg and de \"ries equation is well Eq. (5) can be neglected. For such a wave, moving
i\nown [l-4]. Jn particular, i t has been shown that along t he x axis of the coordinates chosen wíth the
::-: thc dc ve !opm ent of a ny initial pertur bation u(x, O} v elocicy c, the oscillatíon irequc?ncy for neptl ve
an important role is played by spcci:il s oluti0ns to dispersion is
Eq . (1) óf the s olitary -wave type :

u= "'° (.r, t) = ai(p(:: - .ro)) , (2) and for a wave with positive dispe r sion , in which
wliere a is th c w:we 3mplitude , and x = at is the cas e the wave move s in the :leg:it ive dtrection along
w:; ._.c ph:ls~. The iunct ion i (0 , s:nisfying the the x axis ,
equations k:
(J) = - kc -k.,c:::::- ~ ¡..:. .
- :r
- ( + ii' -r /"'=O, -! 7 '¡:f + í" = O (3)
Hence one c:in see that
tliat fo llvw fro rn Eq. (ll (whe re the p rime indícates
Ciffc r cnti:l ti on with respect to ~ ) , equals • couespon<lin~ Mer::be1 ...\c a.i.:r::y .::f Sc1ences of th~ ;.:ssR.
539
540 B. B. KADOMTSEV AND V . l. PETVIASHVILI

(6) part of the perturbation u1 "" cons t · f' that constl-


tutes a solution of the homogeneous equat ion (8)
whc r e the s igns - and + correspond to the nega- be zero, which in this c ase implies a correctly
ti ve and pos i ti ve di spers ion, respecti vel y. cho sen phase "<!· Hence, to second o rde r in e, we
We shall furthe r limit ourselves to the linear have
approximation (8Jeo /ay) « l (the oscillation ampli- oro l. fJt = a. (9)
tud e of the phase Xo can nevertheless be much
larger than the width of the solitary wave - a- 1/2). [n the next approximation with respect to t,
Even with the línear approximation. the sys tem o f ¡t wíll be ade quate to conside r in <Pi only terms of
equations (5) and (6) is rather complicated so we the first order of magnt.ide witb respect to <Pt so
s hall concentrate on tbe limLting case o f very large that wi.thin a linear approximation we have
wavelengths along y when (8xo/cy) « (a /c) t/ 2 < l.
In this cas e. <P is small and Eqs. (5) and (6) may be
solved by the Krylov-Bogolyubov method, i.e., by
Consequently the second-order correction u 2 is
introducing slowly varying para.meters. Tberefore
we shall introduce in Eq. (5) the variable ~ = b~1 f il':zo f (lO)
.fa(X - Xo) instead Of X and rewrite the equation
, , iJ {
'"il .,- a 1, 8( \ - u, + u, + ll"u: \
a~ · J =
011 .J. ta.
- -¡¡ : .._ T i!:¡• •

. .L. iJ'u\ v a - (0%G


.- u+- -J = -f-iJ;,
· o~ -01 - a.
" d ( "'
We shall now multiply Eq. (10) by f and ín-
J.'
4"-
a~ ~ª ' ª~'
tegrate it with respect to ~. It can be shcrwn by ua-
ing Eq. (3) that the integral of the second term
(7)
on the left-hand side of Eq. (10), integrated by parta,
becomes zero and cons idering the arder of magni•
where the right-hand side can be consídered small. tude OÍ 8~ f 8t ,_ E 3. it follOWS that up to third-or-
In the zeroth approximation the right-hand side is der terms we llave
made equal to zero and it follows that u = ~ =
aj(~). In the s ubsequent approximation we put u= ~ _ ±'.?taiJ•:zo
a1 - 3 ar . (11}
u0 + u 1 , linear ize the left-hand side of Eq. (7), and
substitute u 0 in the right-hand side:
where the factor 2/3 is dueto averaging with re-
a,.n Q;
a 1I - ,
U 1 .. _
I u 1 ~. º"""
Ü~·: j = a•;'.f' (i>i
0
7Jl - ) lfo
a - ¡¡ Z -
. ocp
by , {S) spect to ~ (NB < zf > = 3/ 4 <.f 2 >). F r om Eq. {11)
one can s ee that the amplitude oscillations are in
where fact s~bstant ial ly s maller tha.n the phase oscilla-
tions x0 •
r = df í d;. == i + 'iz~r. The fun ction u 2 can be found approximately if
and l/l is to be· found from Eq. (6) by substituting u= uz = A{l-z) is assumed and if A is a funct lon weak-
u 0 in íts right~hand side . ly dependent on ~, i.e., its higher-order derivativei
We shall c onsider the quantities a/ éH and with res pect to ~ are neglected, With such approxi•
a/ ay to be small and of thc o rder e. As we shall mations and with the use of Eq. {11), Eq . (10} be-
see l:lter, the oscillations of the ph.'.lse x 0(y) are comes
much larger than the os c illations of the amplitude
a, so that the va riablt: part of the amplitude ;; ( \-·).JA .:.. a''•~= -~~(/ -· :O/') (l:?)
- 01 · a~ ~ 01 · , ·•
s hould be con s idered small and of the ord er E:(;,.
To m:linta in accur acy up to the order e , one should
Apart from the solitary wave thi s rcl ation de-
retain th e first term only on the right-h and s 1d c of
scribe s also a w:ive prop:ig:iting opposite to ·.;:
Eq. (8) . Fu rt hermo re, as can be proved by means
o f Eq. (3); the solution u, = (uio f ot~ a} :, where a..i ¡ ai + a•..a..i / a~ = o. (13)
: = / + 'h'U', can be found from Eq. (8 ). An addl-
tion to thc basic solution lli) is pr oportional to z.1nd, as so that A = f(t - ~ a- 31 2 ) , where F is an arbitrary
can be seen, corresponds to small variations oi the func tion. [n the case of 3 soli t:i. ry wnve. one mlY
so!itlrywaveampl itudea. Wecanrequire, without ignore in Eq . (l~l thc deri v:i.tive of A witb respect
:rny loss of gener::ility, that the addition be zero. T his to time, and if in the s tream advancing aga inst the
meaos in fact that the amplitude h:is been chosen solitary wave the pcrturbation is 3bsent, then A is
C•Hrectly. Sim ilarly, one cou lJ r cquirc that the fou nu from Eq , (12) by int~gra ting it with rcspcct

....
STABILITY OF SOLITARY WAVES
541

eo ~, wíth the boundary condition A ::: O at ~ = .... Assuming tlut there is no perturbation in the
in particular to the left away frorn the sol itary
strcam advancing toward the solitary wave we have
wave we have
rp.:: O at ~ ::: + oo, and ~ can be easily found from Eq.
A (L) = F (t):::;:: - t º" e·.
-¡-'ji-- • (15). Multíplicatwn of Eq. (10) by j and intcgr:'l-
tion with respect to ~, with the small terms au~ /Ot
Accordir.gl·:, from an oscillating solit!l.ry wave and arpd&f present, serve to demonstrate that the
there propagat~s bach'Ward a long-wave perturba- small additional terms - c,Z rcsult in d:unping the
tion of the trave!ling type whose amplitude is pro- solitary wave oscillations.
ponionai to aa Í at. We ha ve thus demonstrated that with negati ve
We find from Eqs. (9) and (10) a relation for dispersion (e.g., waves in shallow water), "bend-
phase oscillations of the solitary wave Jeo: ing" of a solitary wave lead.s to elastic oscilla-
tions with a weak damping. With positive disper-
(14) sioo, the solitary wave is unstable with regard to
two-dímensional perturbations of the bending type
where the amplitude a, according to our approxí- and as such is not líkely to last a long time.
m:ition, can be taken as constant. If the nonlinear term in Eq. {10) is negative,
Equation (14) indicates that in a medium with then the situation is tbe reverse: the soHtary
a positive dispersion (minus sign) the solitary wave wave with positive dispersion is stable and the
is unstable wíth regard to a bending and its small solitary wave with negative dispersion is unstable.
perturbations are growing with time. The growth
increment of such perturbations is proportional to
the square root of the :unplitude, i.e., it is rather LITERATURE CITED
large •
. 1n the case of negative dispersion (plus sign). 1. N. J. Z abusk.-y and M. O. Kruskal, Phys. Rev.
Eq. (14) leads to harmonic oscillations, and in or- Lett., l l1 240 (1965).
der to sol ve the stability (or instability) problem 2. Yu. A. Berezin aod v. l. Karpman, Zh. Eksp.
of the solitary wave one needs to examine the terms Teor. Fiz., g, 1557 (1966) [Sov. Phys. - JETP,
of order e 3. Therefore it is essential to pre- 24, 1049 (1967)}.
.. -. 3. V. I. Karpman, Zh_ Eksp. Teor. Fiz., g, 1657
re- ·· serve in Eq. (10) the additional second-order term
:q. (11) o~ satisirin~ the rclation (1967) fSov. Phys. - JETP, 25. 1102 Cl967)l •
~ in -.· 4. V. I. Karpman and V. P. Sokoiov, Zh. Eksp.
.la- • c'l<p,= - ~~z. (15) Teor. Fi2., i1_, 1568 (1968) [Sov. Phys.-JETP,
a~ 2 ~·a 11y
....;- ,ll. 83 9 (1968) l .
y i(. -
·.ve:ik-
·ative•
Jproxi-
JC-

(12)

1 de-

(13)

ary
t:lV

.-\ ,:,
.:et

You might also like