A NNALES DE L ’I. H. P., SECTION A
R OLCI C IPOLATTI
On the instability of ground states for a Davey- Stewartson system
Annales de l’I. H. P., section A, tome 58, n
o1 (1993), p. 85-104
<http://www.numdam.org/item?id=AIHPA_1993__58_1_85_0>
© Gauthier-Villars, 1993, tous droits réservés.
L’accès aux archives de la revue « Annales de l’I. H. P., section A » implique l’accord avec les conditions générales d’utilisation (http://www.numdam.
org/conditions). Toute utilisation commerciale ou impression systématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fichier doit contenir la présente mention de copyright.
Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques
http://www.numdam.org/
85
On the instability of ground states
for
aDavey-Stewartson system
Rolci
CIPOLATTI (*)
Departamento de Matemática Aplicada,Institudo de Matematica, Universidade Federal do Rio de Janeiro,
C.P. 68530, Rio de Janeiro, Brasil
Vol. 58, n° 1,1993, Physique théorique
ABSTRACT. - In this work we
study
theinstability properties
ofground-
states for the
equation
in
1R2
and1R3.
We prove that the set(where
p is aground-state)
is unstableby
the flow of theequation ( * ) provided a ((x 2014 2) ~
0.RESUME. 2014 Dans cet article nous étudions l’instabilité des
ground-states
pour
l’équation
.dans
[R2
and[R3.
Nous prouvons que l’ensemble(ou
p estground-state)
est instable pour Ie flux del’équation (* )
sia(fx-2)~0.
(*) This work was partially supported by Conselho Nacional de Desenvolvimento Cienti- fico e Tecnológico - CNPq (Brasil), Proc.: 202176/90-8.
86 R. CIPOLATTI
1. INTRODUCTION
In this paper we
study
theinstability
of Ground-States for theequation
where A is the usual
Laplacian operator
in[R",
a and b arepositive constants,
andE 1
is thepseudo-differential operator
withsymbol
.
The
equation ( 1.1 )
has itsorigin
in fluid mechanicswhere,
foru = N = 2,
it describes the evolution of
weakly
nonlinear water waveshaving
apredominant
direction of travel. Moreprecisely, ( 1. 1 )
is the N-dimensional extension of theDavey-Stewartson systems
in theelliptic-elliptic
case,namely
(A,
andv > 0)
which describes the time evolution of two-dimensional surface of water waveshaving
apropagation preponderantly
in the x-direction
(see [6]).
The
Cauchy problem
for the D. S.systems
in allphysical
relevant caseshas been studied in
[7] by
functionalanalytical
methods. Thestanding
waves for
equation ( 1.1 )
has been treatedby
the author in[5],
with theexistence of Ground-States obtained
by
means of P.-L. Lions’sconcentration-compactness
method.By standing
waves we meanspecial periodic
solutions of the formwhere 03C9 ~ R and cp E
H1
The so-called Ground-States arestanding
waves which minimize the action among all nontrivial solutions of the form
( 1. 3).
The
problem
ofstability
andinstability
ofstanding
waves for nonlinearSchrodinger equations
has been studiedby
several authors. Let us mention the papers ofBerestycki-Cazenave [1] ]
on theinstability
ofground
states,Cazenave-Lions
[4]
for the existence of stablestanding
waves and thepapers of Cazenave
[2],
Shatah-Stauss[10]
and Grillakis-Shatah- Strauss[9].
With the
meaning
ofstability
to beprecised later,
we may summarizeour main result as follows
(see
theorem3 . 15):
THEOREM. - Let
Ne{2, 3}
and assume a(oc - 2) _
o. Then all Ground- Statesfor equation ( 1. 1 )
are unstable.We
organize
this paper as follows: in section 2 we introduce the notation and webriefly
review some results on the existence of solutions for theCauchy problem
forequation ( 1. 1 )
and existence andregularity
for itsAnnales de l’Institut Henri Poincaré - Physique théorique
standing
waves. In section 3 we state and prove our resultsconcerning
the
instability
of Ground-States. Let us mention the theorem 3 . 12 whichgives
sufficient conditions for theinstability;
it isessentially
due toGonçal-
ves Ribeiro
(see [8]).
We finish this paper with an
appendix
where theconcavity
criterion of Grillakis-Shatah-Strauss is stablished for ourproblem
in the case N = 2.2. EXISTENCE OF SOLUTIONS
In this section we shall
briefly
review some results about the existence of solutions for the initial valueproblem
and the existence ofstanding
waves for the
equation (1.1).
Throughout
this paper we consider the LP spaces, 1_p __
+ oo, of com-plex
functions onjRN
with their usual norms denotedby 1.lp.
We considerL2
as a real Hilbert space endowed with the scalarproduct
and we denote
by
HS and the usual Sobolev spaces on Inparticular
we denoteby 11.11 I
the usual norm ofH~
1 related to(2 .1 ) .
Wealso denote
by sob(N)
the Sobolevexponent,
i. e.,Then we can state the
following
result about the local existence of weak solutions of theCauchy problem
for( 1.1 )
in the energy spaceH 1 (see [3],
thm.
4.3.1,
p.65).
THEOREM 2 . 1. -
3}
and exe]0,
sob(N) - 2[.
Then thefollow-
ing
holds:(i )
For any uo EH 1,
there existT *, T*
> 0 and aunique
maximal solutionof (1.1)
such that u
(0)
= uo. Themaximality
is in the sense that limII
u(t) II
= + 00t i T*
if T*
+ 00(resp.
limII I u (t) II =
+ 00 +(0).
t i - T*
(ii )
We have conservationof charge
and energy, that is88 R. CIPOLATTI
for
all tE ] - T*, T *[,
whereFor more
specific
resultsconcerning
theCauchy problem
for theDavey-
Stewartson
systems,
we refer the reader to[7].
In order to
simply
thenotation,
from now on we will denoteby
U(t)
uo the maximal solutionu (t)
of( 1.1 )
which satisfies the initial condition uo.Looking
forstanding
waves forequation ( 1.1 )
leads us to consider thestationary equation
whose solutions are critical
points
of theLagrangian
S definedby (see [5])
Let us introduce the
following
sets(~ being
the set ofGround-States).
For 03C9 and b
given positive
constants, letwhere
If we consider the functionals
then we can state the
following
result about the existence of Ground-States for
( 1.1 )
.Annales de l’Institut Henri Poincaré - Physique théorique
THEOREM 2 . 2. - Let N
E {2, 3}
and(ex, a)
E Then thefollowing
holds:
(i) If
N = 2 then tp ~ Gif
andonly if
tp solves thefollowing
minimizationproblem:
where
~o = { ’"
E HI 0,
V(~)
=0 }.
(ii) If
N = 3 then there exists a constant > 0 such that pif
andonly if
cp solvesthe following
minimizationproblem:
where
= ~ ~
EH I 0,
V(~) _
(iii)
Problems(2.4)
and(2. 5)
have solutions.(iv)
For all cp E~,
there exists a real valuedpositive function
cpo such that cp =ez~
Proof -
For theproofs
of(i ), (ii )
and(iii )
we refer the reader to[5],
theorems 2 1 and 2. 2. To prove
(iv),
we mayproceed
as in Cazenave[3], p. 171.
0We have the
following
resultconcerning
theregularity
of solutions of(2 . 3):
THEOREM 2 . 3. -
If N E {2, 3 },
ocE]o,
sob(N) - 2[
and cp is a solutionof
(2 . 3),
then thefollowing
holds:(1 )
p EW3, P,
Bif p E[2, 00[.
(ii)
limI x --~ 00
(iii )
There existpositive
constants C and v such that(iv) Moreover, i, f ’ oc > 1,
then(i)-(iii) hold for b’ j =1,
..., N.Proof. -
For theproof
of(i)-(iii)
we refer reader to[5],
thm. 2.4.In order to prove
(iv),
we notethat 03C8
=ai
cp satisfies theequation
-039403C8+03C903C8=bE1(|03C6|2)03C8+2bE1(R(03C603C8)03C6 03B1a|03C6|03B1-203C6R(03C603C8)-a|03C6|03B103C8,
and the
arguments
in theproof
of the theorem 2.4 in[5]
areapplicable, provided
a > 1. DWe end this section with the
following
LEMMA 2.4. - Let
Ne{2, 3},
co, bpositive
constants,(a, a)
ELet ji=O
Then the sets90 R. CIPOLATTI
are invariant
regions
under theflow of U. Moreover, if
uo then U(t)
uo isglobal.
Proof. -
Let uo Then S(U (t) uo)
= S(uo)
S(cp). By
the definitionof
and Theorem 2 . 3 we have Since the functiont H V
(U (t) uo)
is continuous on]
-T~, T*[,
we have V(U (t) uo) Q
ifor
T * [.
Assuming
now that uo we haveand we conclude the
proof.
D3. INSTABILITY
For any
X c H
wedefine V(X, õ)
the5-neighborhood
of Xin
H 1 by
where
Bs (v)
is the ball inBs I u - v II 8}.
DEFINITION 3 . 1. -
We say
thatXc H 1
isstable by the flow of U if
and
only, if, for
any8>0,
there exists ~ > 0 suchthat, for
all uo E(X, 8), T*(Mo)=+00
and U(t)
u0 ~ r(X, E)
V t ~ 0.If u is a
periodic
solution of(1.1),
then orbitalstability
of u meansstability
of the closedorbit {M (t) I t
Eby
the flow of U.Remark 3.2. - Due to the invariances under translation and multi-
plication by
we never have orbitalstability
for thestanding
waves forequation (1.1). Indeed,
let and consider the closed orbit Given s>0 and aunitary
letM,
(t, x)
(p(x -
2 Ety).
Then weeasily verify
thatwhere
(x)
= ~’’ cP(x). Furthermore,
- tp inHI
asbut for
8>0,
where
(p(x)=(p(2014~).
Sincecp (x) --~ 0 as x ~
- aJ(see
theorem2 . 3),
weinfer from
Lebesgue’s
theorem that(I P I * (sy)
- 0 ass T
oo and weconclude that lim dist
(uE (t); C~~) >
cp~2.
For each y E let
iy
the translationoperator
definedby iy
v = v(.
+y).
If
YeLP,
we denoteQy = { e‘8 iy v ~
leER,
y E v EY ~ .
With this notationwe introduce the
following
set which is a natural extension of the closedAnnales de l’Institut Henri Poincaré - Physique théorique
orbit
(!) p
in thestudy
ofstability
in our context:In
fact,
it is well known that(see [3]),
in the pure power case(which corresponds
in our context to the case b = 0 anda 0), Q
is stableby
theflow
ifOa4/N. Moreover, S~~
is unstable ifprovided
cp E ~.LEMMA 3. 3. - For each
Y E HI
have Yõ) =
(y. õ).Proof -
Let u E Y(Qy, õ).
Then there exist e E[RN
and w e Y such thatõ)
andconsequently
Conversely,
if u E (Y. s~, then u =e~e iy
v for some e EIR,
y E andv E f’
(Y, 8). Hence,
there exists w e Y suchthat ~w - 03BD~
8 and the conclu-sion follows since D
For each
vEHI,
we introduce thesymmetric
matrixwhere
LEMMA 3 . 4. - Let cp E
H2
such that ~ isstrictly positive definite.
Then there exists Eo> 0 such
that, for
all v E ’~Eo),
there exists aunique
~
(v)
EQcp satisfying
Moreover, the function
~ :’Y~ (SZ~, Eo) ~!1cp
isC2
with derivative~’ :
’Y~’ ~52~, Eo) ~ ~(HB- given by
where ’~’
(~2~, Eo)
--~ (1~ areC2 functionals satisfying for j = 1,
...,N,
e E
E~,
y E V(Qq>’ Eo):
In
particular, for
v E0,
..., N andk =1,
..., N we haveProof -
Let E > 0 and consider the function92 R. CIPOLATTI
Since cp E
H2,
weeasily verify
that F EC3
in all its variables andDenoting by
Grad F =(~ F, ayi F,
...,ayN F)
andby
Hess F the hessian matrix of F in the variables(6, y),
weget
It follows from the
implicit
function theorem that there exist80 >0, 80 >0
and R > 0 and N + 1 functions of classC2
such
that,
for all v EBEo (p)
we have:Hess F (v, Ao (v), A (v))
is astrictly positive
definitematrix,
where
A (v) _ (A (~),
...,AN (v)). Therefore,
if we denoteby
then,
from(3. 9)
and(3 .10),
we inferthat,
for v EBEo (cp),
/(v)
is theunique
element ofBto ((p) satisfying
for all
n BEo (cp). Moreover,
sincefor all
6B y’,
it follows from Lemma 3.3 thatAo
and A may be extended to
1/ (Qcp, 80)
in such a way that(mod 2 ~),
V S E
R, V y
E[RN
and(3 . 1)
holds.We obtain
(3.2) easily by taking
derivatives in(3 . 11).
Let
Eo), ~ Then,
we infer from(3 . 6)
and(3 . 9)
that(v + s ~; ~’J~(~+~(j)))2==0,
for allsufficiently
small.By differentiating
this last
identity
in s, we obtainAnnales de l’lnstitut Henri Poincaré - Physique théorique
which, by (3.2), gives
usfor all v E 1/
(SZ~, Eo).
The same
arguments
with(3. 7), (3 . 9)
and(3. 2) gives
usIn
particular,
for v ES2~
we obtainSince
(iv; v)~ = 0
is invertiblefor
allwe deduce
(3 . 3)-(3 . 5) easily
from(3 . 14).
DIf then it follows from Theorem
2. 2-(iv)
that(~(p;~(p)2==0,V/=l,...,N
and weeasily verify
from(3 .14)
thatMoreover, by solving
thesystem (3 . 14)
we obtain0 ( )
Ai (v)
as a linear combination ofa
...,aN v
for all Inparticular,
for N = 2 we obtain
explicitly
from(3.14) that,
for every v E5~~:
where
LEMMA 3 . 5. - Let N
E {2, 3 ~ . If
cpE ~,
then ~ isstrictly positive
definite.
~’roo, f : - ~
isalways positive
definitebecause,
V 3 E we maywrite
Assume that
~((p)9.9=0
forsome 3 = (30, 9’) 7~0
and considerf (t)
= (p(t 9’). Since at f (t)
=(i 9o
(p(t 9’)
+ 9’. V p(t 3’)),
we havewhich is in contradiction with theorem 2. 3. 0
Remark 3 . 6. - It follows
directly
from(3 . 6), (3. 7)
and(3 . 9) that,
forall v E 1/
Eo):
94 R. CIPOLATTI
In
particular,
from(3 . 2)
we obtain for all v E(5~~, Eo)
andall ~
EH 1:
Consider the
operator P : L2
x(L2B{
0})
-L2
definedby
Note that ~ is the
orthogonal projection (in
theL2 setting)
of v on theclosed
hyperplane {M
EL2 (u; w)2
=0}.
LEMMA 3 . 7. - For all v E ~
(D~, Eo),
thefollowing
holds:Proof -
From the definition of P we obtainFrom Lemma 3.4 and
Pitagoras’ Theorem,
we haveand the conclusion follows
since (v) I2 =
cpI2.
DRemark 3 . 8. - It follows
directly
from Remark 3 . 6 that(EP (v,
~(v));
JV’
(v)
= 0 for every v E YEo) and ())
EH 1. Moreover,
from Theorem2.2-(iv),
the functions in thefollowing
sets aremutually orthogonal
inL2:
For
each 03C8 ~ H1
and v E Y(Qp, Eo),
we defineNote that
(v) = -
i~’w). Therefore,
we infer from(3 . 2)
thatLEMMA 3 . 9. - Let cp E
H2
be such that ~ isstrictly positive definite.
Then there exists Eo> 0 such that Bif uo E 1/
(52~, Eo),
we canfind TEo
> 0 suchthat,
Vx/
EHl, 3
C > 0for
which the solution u(t)
= U(t)
uoof (1 . 1) satisfies
Annales de l’lnstitut Henri Poincaré - Physique théorique
Proof -
Let Eogiven
in Lemma 3 . 4 and defineFrom the
continuity
ofU (t) uo,
we haveTEo
> 0 and the result followseasily
from(2 . 2) and M 2 = ~ 2-
DFor
each 03C8 ~ H1
we consider the functionalThen the
following
holds:LEMMA 3 . 10. - For
each ~
E~,~
is aC2-functional satisfying:
Proof : - (3 . 19)
follows from theproperties
ofAo
and A(see
Lemma3 . 4). By differentiating ~,~
we obtain from(3 . 18)
thatIn
particular,
for v = p we haveand we infer from
(3 . 4)
and(3 . 5)
thatWe obtain
(3 . 20)
and(3 . 21 )
from(3 . 24)
and the well known formula 0Remark 3.11. - Note that
(3 . 23)
may be written asLet us
point
out that iv andbeing tangent
toS2~ at, v,
the identities(3 . 20)
and(3.21)
saythat,
for allBf1 E HI,
the functional~,~ : ’~’ Eo)
- [R defines afield
that is transversal to This indicates that thetrajectories created by ~~
andstarting
nearQ~
may scape1/
(03A903C6. s)
in a finite time if ~ is smallenough.
This is thekey
idea for thefollowing
result which isessentially
due toGonçalves
Ribeiro[8].
THEOREM
then there exists ~ > 0 and a
sequence ~
in YE) satisfying
96 R. CIPOLATTI
(i ) u~
~ p in oo .(ii )
Forall j EN, U (t) U j is global
but scapes ~(Qp, E)
in afinite
time.Proof -
First ofall,
let Eogiven
in Lemma 3 . 4 andconsider ~
EH~
1such that
For each vo E ~
(Oq>, Eo),
we consider the initial valueproblem
A
straightfoward
calculation with the identities(3.12)
and(3.13)
leadsus to
get
the estimates:where
Co depends on ~ and ~ (p
~
Since
A~ : B£o
- R isC2, 0,
..., N(see
Lemma3 . 4), by taking
Eo if necessary, we may assume that there exists
C
> 0 suchthat,
forall v E 1/
(Qq>, Eo)
Therefore, by differentiating
theidentity (3.23)
weget
where C
depends only
onCi
It follows from
(3.29)
that the initial valueproblem (3.27)
admits aunique
maximal solution(Jo];
1/’Eo)),
where (Jo = (Jo(vo)
is some
positive
constant.Moreover,
for each E1 Eo, there exists03C31 > 0
such that 60
(vo) >__
61,V vo
E1/’(Qcp, E1).
Taking
E 1 Eo and(J 1 > 0
asabove,
we may define the nonlinear semi- groupwhere ~
(s) vo
is theunique
maximal solution of(3 . 27).
Note that U is a
C1-function
in both variablesand,
for vo E j’"(03A903C6, GI)’
the function
s ~--~ ~ (s) vo
isC2. Moreover,
it follows from the invariances ofj, j = 0,
...,N (see
Lemma3 . 4),
comutes withei03B8 03C4y
forLet
and
Annales de l’lnstitut Henri Poincaré - Physique théorique
Since is
C2,
we mayapply Taylors’
theorem toget
for some
v E ]0, 1[.
Since R is continuous and R0,
there exists82 ~
~1and such that
for all
~e] 2014
cr 2’cr 2[,
V vo EBE2 (cp).
Since(s)
comutes with we obtainfrom Lemma 3. 3 :
In
particular,
for vo = ~(’t)
q>, with ’t =1= 0sufficiently small,
weget
So, by taking
s = - i 0 we obtainFurthermore,
from(3 . 30)
we haveand we infer from
(3 . 32)
and(3.33) that,
for some03 ~
a2:On the other hand we have
[because
otherwise we would havetangent
toLJ.10
at cp(po
= 0 ifN = 2)
and since p minimizes S onLJ.10’
we would have(
S"~+ (cp); ~* (c~) ~ >_
0 in contradiction with(3. 26)]. Moreover,
since
(c~) _ - ~+ [see (3 . 23)],
we may assume without loss ofgenerality
that(
V’(cp);
i~~, (cp) ~
o.So,
for T > 0 smallenough
we have(where ~
= 0 if N = 2and ~ ==
if N =3).
Define[see (2.6)].
Then D is invariantby
the flow of U and we have from(3 . 33), (3 . 34)
and(3.36)
thatLet
TjE]O, N,
such that 0as j ~
00 and consider cp.Then p in
H1 as j ~
oo which proves(i ). Moreover,
from(3 . 37)
and98 R. CIPOLATTI
Lemma 2 . 4 we infer that U
(t) u~
islocal,
for allj.
In order to conclude theproof, we
needonly
toverify
that escapes ’t"(.Qq>, E3)
for some83
>0,
forConsider the function
A : ] - a i,
definedby
A
(s, vo)
= V(~ (s) vo).
Since[see (3 . 35)]
A(0, v) = Jl
andit follows from the
implicit
function theorem that there existsE2 __
82 suchthat,
for every vo E r(QI>’ E2),
there exists aunique
s(vo) satisfying
From
(3 . 30)
and(3 . 38)
we obtain(with
83 =min {8~ ~}):
If we introduce
t 1 [ ~,
it fol-lows from
(3 . 39)
that VtE] ] 0, TJ, 3 sjE] -
03,oj satisfying
From the conservation laws
(2.2)
and(3 . 37)
we infer thatOn the other
hand,
we know thatwhich
gives
From Lemma 3. 9 and
(3. 40)
we infer thatT j 00,
whichcompletes
the
proof.
DIn order to
give
conditions to assure(3.25),
we need thefollowing ellementary
Lemma:LEMMA 3.13. - For we have the
following
identities :Annales de l’lnstitut Henri Poincaré - Physique théorique
Proof -
The identities(i )
and(ii )
are trivial consequences of Gauss’Theorem. In order to prove
(iii),
we remember thefollowing
well knownidentity:
V/} = -
div(ç ~ ~ ~ ~),
where ff denotes the Fourier transform.By denoting W = 1 ~ 2,
we have from Parsevalidentity
and the definition ofEi:
and the conclusion follows because div
1(03BE))
= N 03C31(03BE).
DPROPOSITION 3.14. - Let
Ne{2, 3}, (o, ~>0, (a,
withand Then we have
Proof. -
From Theorem 2.2 we may assume without loss ofgenerality
that p is real. real valued. It follows from Remark 3.11 that i
/I ~~p) _
~(0/, p),
wherewith weget
where, Since
for
and remembering
that pand ware real,
weget:
100 R. CIPOLATTI
Taking
into account that p solves(2. 3),
we haveFrom Theorem 2 . 3 we have
that WEH1
and since= (p it follows from Lemma 3 . 13 that
Since
(q>; = - N I q> 12,
we may rewrite(1. 41)
as:2
In order to calculate
Ji,
let~n ([R2;
such that cpn - x. V cp inH~
1 and consider(Remark
that n -oo.) Taking
into account that p solves(2. 3)
andapplying
Gausstheorem,
we obtainAnnales de l’Institut Henri Poincaré - Physique théorique
Integration by parts gives
and we obtain
Letting n -
oo andusing again
Lemma3 .13,
weget:
Now, applying Proposition
2 .5-(iii)
of[5]
andremembering
that03C6x.~03C6=1 2x,~(|03C6|
2)
weget
03C6 x . V p
2
x . V(I
P12)
weget
Denoting by f (§) = 3’ { (ç)
andremembering
thatwe obtain from Parseval
identity:
Since
the conclusion follows
by merging (3 . 42)
and(3 .43).
0With Theorem 3 . 12 and
Proposition
3 . 14 we areready
to stabilish ourmain
result, namely:
THEOREM 3.15. - Let
Ne{2, 3},
co,(a,
andIf
a> 1 and a
(rJ. - 2) ~ 0
then unstableby
theflow of
U.102 R. CIPOLATTI
The
proof
followsdirectly
from Theorem 3.12 andProposition
3. 14excepted
the case N = 2 and a(a - 2)
=-0 for which the condition(3 . 25)
isnot assured. This is a
special
case for which a verysimple
idea ofWeinstein[11]
isapplicable
and that we treat as the nextproposition.
PROPOSITION 3 . 16. - Let
(x=2, N = 2,
a b and(pe~.
Thenu (t, global
solutionof ( 1 .1 )
which is unstable in the sensethat there such that
H1 as n T
oo and U(t)
cpn blows up in afinite
time.Proof. -
Since we have Hence $(~.cp)
0 for all~>1. Let =
~,,~
where1,
1 and consider un(t)
= U(t)
Since . cp E L2 (see
Theorem2 . 2),
it follows that the functionx ~ ~ ~
is well defined and we have(see [7],
thm.3 . 2,
p.
495):
dtt In (t)
= 16 E(Un (t))
.But
by
conservation of energy we whichconcludes the
proof.
D4. APPENDIX
In a well known paper
(see [9]), stability
andinstability
ofstationary
states for abstract Hamiltonian
systems
are obtainedby
means of theconvexity
of a real valued function d(m)
which is related to the action of thesystem:
In our context, thisapproach
leads us to consider the functionwhere
(in
the case N = 2 forinstance)
This
point
of view does not seem to work in our context because wedo not have a
precise
characterisation for the kernel of S"(cp). Indeed, equation (2. 3)
does not have radial solutions(cf. [5])
andfor j=i,
..., N ifNevertheless,
it is inter-esting
to remark that the same condition on theparameters a
and a whichimplies instability
ofGround-States,
assures theconcavity
of the functionMore
precisely,
_THEOREM 4 . 1. - Let
N = 2,
co, ~>0.7/~((x-2)0,
then thefunction
is concave on
]0,
+00[.
Annales de l’Institut Henri Poincaré - Physique théorique
Proof. -
First ofall,
note that if~(a-2)0,
thenIro
is well definedon
]0,
+oo [. Indeed,
if a 2 then(a, a)
E V aE,
V co > 0. On the otherhand,
if a > 2 and a 0 then(a, a)
E b, > 0. Assume a(oc -- 2)
0 andlet D>0. From Theorem 2. 3 the minimisation
problem (2.4)
has asolution which we choose from now on. The
following ellementary
facts hold:
(i )
For each we have(ii)
inf{T
>0
=0}
= 1( because
otherwiseL - ~
.By continuity,
we haveLet us consider the function
Since
F ( 1, 0) = 0
andat F (1, 0)=/~(1)=4IJ~ ~>0~
it follows fromthe
implicity
function theorem that there exists 8>0 and a C°°-functiona :
]
- s,E[
- [R such thatFrom
(4 . 1 ), (4 . 2)
and the definition ofF,
we haveIn
particular,
from the definition of1~
and(4. 4)
we obtainfrom which we deduce
Letting h 1
0 andtaking
into account theregularity
of 03C3(h),
we obtainBut from
(4 . 3)
we see that(h)
and astraightforward
calcula-tion
gives
us104 R. CIPOLATTI
which
completes
theproof.
DRemark 4. 2. - The result of Theorem 4 .1 seems to be true for N = 3
although
thearguments
used in itsproof
do not work.Indeed, according
to Theorem
2. 2,
is aground
state if andonly
if is a solution ofproblem (2.5),
where yo = ~ro =(I~~3)3~2 (cf [5]).
Thisstrong dependence
on co does not lead us to
proceed
as before.ACKNOWLEDGEMENTS
I would like to thank T. Cazenave for his
helpful
remarks. I amgrateful
to the Laboratoire
d’Analyse Numerique
de l’Université de Paris-XI for itshospitality.
REFERENCES
[1] H. BERESTYCKI and T. CAZENAVE, Instabilité des états stationnaires dans les équations
de Schrödinger et de Klein-Gordon non linéaires, C. R. Acad. Sci., Paris, T. 293, Series I, 1981, pp. 489-492.
[2] T. CAZENAVE, Stability and Instability of Stationary States in Nonlinear Schrödinger Equations, Contributions to nonlinear partial differential equations I, Research Notes in Math., Vol. 89, Pitman, London, 1983.
[3] T. CAZENAVE. An Introduction to Nonlinear Schrödinger Equations, Textos de Métodos Matemáticos, Vol. 22, I.M.U.F.R.J., Rio de Janeiro, 1989.
[4] T. CAZENAVE and P.-L. LIONS, Orbital Stability of Standing Waves for Nonlinear Schrödinger Equations, Commun. Math. Phys., Vol. 85, 1982, pp. 549-561.
[5] R. CIPOLATTI, On the Existence of Standing Waves for a Davey-Stewartson System,
Commun. in P.D.E. (to appear).
[6] A. DAVEY and K. STEWARTSON, On Three-Dimensional Packets of Surface Waves, Proc. R. Soc. A., Vol. 338, 1974, pp. 101-110.
[7] J.-M. GHIDAGLIA and J.-C. SAUT, On the Initial Value Problem for the Davey-Stewart-
son Systems, Nonlinearity, Vol. 3, 1990, pp. 475-506.
[8] J. M. GONÇALVES RIBEIRO, Instability of Symmetric Stationary States for Some Nonlin-
ear Schrödinger Equations with an External Magnetic Field, Ann. Inst. H. Poincaré;
Phys. Théor., Vol. 54, No. 4, 1992, pp. 403-433.
[9] M. GRILLAKIS, J. SHATAH and W. A. STRAUSS, Stability Theory of Solitary Waves in
Presence of Symmetry I, J. Functional Analysis, Vol. 74, 1987, pp. 160-197.
[10] J. SHATAH and W. A. STRAUSS, Instability of Nonlinear Bound States, Commun Math.
Phys., Vol. 100, 1985, pp. 173-190.
[11] M. I. WEINSTEIN, Nonlinear Schrödinger Equations and Sharp Interpolation Estimates, Comm. Math. Physics, Vol. 87, 1983, pp. 567-576.
( Manuscript received September 27, 1991;
revised version received May 11, 1992.)
Annales de l’Institut Henri Poincaré - Physique théorique