The initial value problem for the derivative nonlinear Schrödinger equation in the energy space

The initial value problem for the derivative nonlinear Schrödinger equation in the energy space

Nonl,near Analysrs, Theory, Methods Printed in Great Britain. & Applrcarrons, Vol. 20, No. 7, pp. 823-833, 1993. 0 0362-546X/93 $6.00+ .OO 1993 Pe...

653KB Sizes 0 Downloads 36 Views

Nonl,near Analysrs, Theory, Methods Printed in Great Britain.

& Applrcarrons,

Vol. 20, No. 7, pp. 823-833,

1993. 0

0362-546X/93 $6.00+ .OO 1993 Pergamon Press Ltd

THE INITIAL VALUE PROBLEM FOR THE DERIVATIVE NONLINEAR SCHRODINGER EQUATION IN THE ENERGY SPACE NAKAO HAYASHI Department of Mathematics, Faculty of Engineering, Gunma University, Kiryu 376, Japan (Received 1 October 1991; received for publication 2 June 1992) Key words and phrases: The derivative nonlinear SchrGdinger equation, the energy space.

1. INTRODUCTION THIS PAPER

is concerned with the Cauchy problem for the derivative nonlinear Schrodinger

equation iy/, + wXX+ 2iS(ly/lzyl), = 0, XEL? i Y(O, x) = %9,

(t, x) E IRx m,

where 6 = +i. This equation was first derived by [l, 21 to study the propagation polarized nonlinear Alfvtn waves in plasma. Our purpose in this paper is to show the following theorem.

(1.1) of circular

1.1. We assume that 4(x) E H’( E?)and ]I4 )ILzis sufficiently small. Then there exists a unique global solution I+Y(~, x) such that

THEOREM

l#Y(l,x) E C(lR; H’(R)) f-l L&$(E?;P3(lR)).

(1.2)

In [3,4] Tsutsumi and Fukuda proved that the local existence of solutions to (1.1) under the condition such that 4 E H”(R) (s > 3/2) and the global existence of solutions to (1.1) under the conditions such that 4 E H2(Q and 1)411 Hl are sufficiently small. Hence our result is an improvement of [3,4]. Remark 1.1. In [5], Kenig et al. established interesting results for the Korteweg-de equation + UU, = 0, (t, x) E IRx IR, u, + UXXX XE IR. u(O,x) = 6(x), i

Vries

(1.3)

They proved that if 4 E H”(E) (s > 3/4), then (1.3) is locally well-posed in time and if 4 E H’(R), then (1.3) is globally well-posed in time. Their method depends on the smoothing property of solutions to the Airy equation and their results are improvements of the previous papers (refer to [5] and references therein). Our result is analogous to theorem 1.2 in [5]. However, our method is different from that in [5]. It seems that the “L* - Lq estimate”, which works well for the usual nonlinear Schrddinger equation, is not directly applicable to (l.l), since (1.1) has a space derivative in the nonlinear term which causes the so-called derivative loss. To overcome this difficulty, we make use of the gauge transformations stated in Section 2 823

824

N. HAYASHI

which enable us to treat the problem in a way similar to the usual equation. We give notation and function spaces used in this paper.

nonlinear

Schrodinger

Notation and function spaces We let Lp(lR) = (f(x); f(x) is measurable

on R, 11 f llLp< ~01, where 1)f ll$ = (SIR)f(x)Ip dx) if E R) ifp = 05, and,;e let ZPP(R) = (f(x) E P(R); we put H ’ (R) = P(R). We denote by H,,!.P= cj”=, IIa:f lip < a). For simplicity (a, *)Lz the inner product in Z’(R). For any interval Z of R and a Banach space B with the norm functions from Z to B, and we let P(Z; B) be II* llLIl we let C(Z; B) be the space of continuous the space consisting of strongly measurable B valued functions u(t) defined on Z such that \Iu(t)lIB E G’(Z). Different positive constants will be denoted by the same letter C. If necessary, by C(*, . . . . *) we denote constants depending on the quantities appearing in parentheses.

;If7,p < COand Ilf lILw= ess*sup(l f(x)l;x

2. PROOF

In order to prove theorem

i

OF THEOREM

1.1 we consider

i(wj), +

(WjLx+ 2i41Wj12Vj)x = 0,

Wj(O9

=

x,

1.1

x

4j(x),

E

(t,

x)

E

m

x

R, (2.1)

I?,

where ~j(X) E H”(R) and rPj + 4 strongly in H’(R) as j + co. In [4, theorem 31 it was shown that for each j E ~PJthere exists a unique global solution wj of small. We remark (2.1) satisfying wj(t, X) E C”(R; H”(R)) provided that Il~jIlN I is sufficiently here that if we take the Sobolev inequality (lemma 2.1) into account, then the result for wj mentioned above holds valid under the condition such that ll~jIlL2 is sufficiently small. For the nonlinear Schrodinger equation of the form ily, + wXX+

m, 4 = @(x),

Id4w = 0,

(NW

Weinstein [6, theorem A] proved that a sufficient condition for global existence space is \(@llrz < 6 by using the “best constant” for the Sobolev inequality. We first recall the Gagliardo-Nirenberg inequality and the space-time estimate of free Schrodinger equations. LEMMA

integers

2.1. Let q, r be any real numbers such that j I m. Then

satisfying

in the energy of a solution

1 5 q, r s co and let j and m be nonnegative

ll~l4lLP5 ~llm41aL’I14L~“‘, where 1 -=j+a

i-m

P

(

+(1-a): >

for all a in the interval j/m 5 a 5 1, and M is a positive constant r and a. For the proof

see, e.g. Friedman

[7].

depending

only on m, j, q,

825

Initial value problem

LEMMA 2.2. For any $ E L2(lR) we have

where 2 5 r 5 00, 2/q = (l/2)

- (l/r)

u(t)4

and

= &

IKe

-(X-v)*/4ri 4Q) dye

.i For the proof

see, e.g. Ginibre

Remark 2.1. Lemma

and Velo [S].

2.2 was first shown by [9] in the case r = p = 6 and improved

We next recall lemma 1 in [4] which is derived by the infinite covered by [lo] (see also the Appendix). LEMMA2.3. Let wj be the solution

family of conservation

by PI. laws dis-

of (2.1). Then

lIVj(t)lIL2= ll~jllLz~ ll~xVj(f>llZZ + 2~211Vj(f)lli6 + 3~Im(~xVj(f),lVj(t)12Vj(f))L2 = Ila,4jllL2+ 2~2114jllP6 + 36Im(~,djj 14j124j)Lz. By lemma

2.1 and lemma

LEMMA2.4. We assume

2.3 we easily have lemma

that I,Y~is the solution

2.4.

of (2.1) and Il~jIlrz is sufficiently

small.

Then

Ilu/j(OllH1 s c(l14jllH1). We are now in a position

to prove the theorem.

Proof of theorem 1 .l. We only consider analogously.

the case t > 0, since the case t < 0 can be treated

We shall prove that ~wj) is a Cauchy

sequence

in C([O, T]; H’(R))

II L”([O,

T]; H’,‘(R))

for any T > 0. For that purpose we generate some new nonlinear equations from (2.1) through the gauge transformations. Following the paper by Kundu [l 11, we first multiply both sides of (2.1) by

Ej = E,(t, x) = exp ia (

X i-,

IVj(t, W)I’dw * >

Then we have i(Vj),

+

(Vj),

+

2iSlVj12(Vj),

= I.7

Vj(0,

where

X)

=

@j(X)

id

exp (!

i Vj(t,X)

=

Wj(t,X)exp

-m

0,

(t, X) E IR x I?,

I+j(W)I’dw 3 1

XE R,

/wj(t, w)12 dw >

(see also the Appendix

in [12]).

(2.2)

N. HAYASHI

826

both sides of (2.2) by Ej and Ejd, to obtain

We again multiply i(U!“) J i(d2’)

J

f f

+ (ZP) J +

(ZP’) J

zz F!”

xx

=

xx

J



Fc2’ J



I+j(W)I’ dw

(2.3)

3

i

where F!”

= 2~4~!1))2~!2)

Ff2’ J

=

J

u,“)(t,

(see the Appendix).

x)

=

vj(t,

J

J

x)

-2i&~!~‘)~u!”



J

J

IVj(t, w)12 dw 3 )

exp

We note here that (Uj”),

= j&lUj(t)(2+)

+ Uj’z).

(2.4)

The proof of theorem 1.1 will be preceded by a series of lemmas. For the sake of brevity in lemma 2.5 and lemma 2.6 we suppress the subscriptj of uj, t//j and Fje We first prove lemma 2.5. LEMMA

2.5. Let u(t) and u (2) be solutions SUP

Odf57 Proof.

By the definition

by lemmas

h4~~ll,~

5 c.

suP

suP

this inequality

LEMMA 2 . 6 . Let uc2)

Il”‘l’(t)llff~c c(ll+jllH’)IlU(2)(t)llL2 4 c(ll~jllH1)~

and (2.5) the lemma be the solution ” (1.O

where C(T)

II Iv/IW~II~~+ /l~,mll,4.

2.1 and (2.5) to (2.4) to obtain

ostr7 From

+

2.4 and 2.1 we have ost57

We apply lemma

Il”‘2’(t)llL2 5 c(l14jlIH’)*

suP

O~fzzT

of u(l),

Il~“)(&l Hence,

II~(l)mP +

of (2.3). Then we have for any T > 0

is a positive

constant

follows.

n

of (2.3). Then we have for any T > 0

IIu’2’Ct)IIsdt5 C(T) * C(II+jIIHI), >

independent

of j.

(2.5)

827

Initial value problem

Proof. By the integral

equations

with (2.3) we have

associated

f

d2’(t) = u(t)d2’(o) +

U(t - s)P’(.s)

ds.

.i 0 The Lp-Lq estimate

yields

llLP(&3 We use lemma

ds. = llU(t)u’2’(0)IlL~+ c f (t - s)-“~IIF(~)(s)II~~/~ i .O

2.1 to obtain l(P(.s)llp

From

(2.6)

(2.6), (2.7) and lemma

5 CJlU(1)(s)l(~~llU’2’(s)ll~3.

2.5 it follows

(2.7)

that

IId2)(t)llL3 I IIu(t>d2)(o)ll~3 + c(II~~II~~) (t - S)-“611U’2’(S)llt3dS. (2.8) In what follows, inequality “IIu(‘)(s)ll;: (1.0 The second

we let C, = C((J4jjllH~). By (2.8) and the singular

for simplicity

” )IU(s)~‘~‘(O)112 ds ) “12 ds “I2 5 ) (i 0

term of the right-hand

&(I;

+

side of (2.9) is estimated

(Iu(2)(s)11~‘11 dsji1’i2.

integral

(2.9)

by (2.10)

“6t3”. )

We let

(iIf

lld2’(s)l/$ ds “12.

Y(f)=

0

Then we have by lemma

)

2.2, (2.9) and (2.10)

y(t) I clp(o)~~L* Since by lemma

(2.11)

+ C,t3’4yZ(t).

2.1

Il”‘2’(o)IlL2 5 c
from (2.11)

y(t) I C&l Now we choose

+ Yy2(t)).

(2.12)

r, > 0 so small that Ti <

(2C&4’3,

sup

y(t) 5 2CH.

then OSlST,

By using lemma

2.5, we repeat

Remark 2.2. The same argument

the above argument as in lemma

to obtain

the lemma.

n

2.6 has been used in [13, 141.

828

N.

HAYASHI

We next prove that (~7’) is a Cauchy sequence in C([O, T]; L’(R)) fl L12([0, T]; L3(lR)) for any T > 0 and I = 1,2. LEMMA2.7. Let uj” and uj’2’be solutions of (2.3). Then we have for any T > 0 and I = 1,2 I I(uj”‘(t) - ~ei(i)lI::d1)l/12

5 C(T) . C,ll~j

- ~kllH1,

(i sup

IIu,“‘(t) -

uf’(t)IILz 5 C(T) *CHII$j-

where C(T) is a positive constant independent

of j and k,

cff = c(l14jllH’ Proof.

$kllHl,

9 l14kllH1).

We let W”‘(t) = u!‘)(t) - z#@). J

Then we see that i(W(‘)) I + (WC’))xx = F!‘) _ F(‘) k .

(2.13)

J

From the integral equation associated with (2.13) it follows that II~(WllL3

5 Ilw)(~vm3

+ c

’ (t - S)-qFqS)

- Ff)(S)(lL3/Z

J

ds.

(2.14)

0

By Holder’s inequality we obtain

IIFj”’ - FpyL3/* I IIUj1)llLrnllUj1)llL311Uj(2) - U~2qr3 + (llUjl)llp

+ Il~6’~Il~~)ll~~~II~3II~j(~~ -

4’)11~3,

(2.15)

and IIFi’z’- F$2$‘/Z I (]IUj’z’II L3 + IIu~2)IIL3)lluj(1)llL~IIuj(2) - u~2)llL3 +

Ilup’y53~~4)

-

(2.16)

zd/y’II,-.

Lemma 2.1 and (2.4) imply I]Uj”’- U:“]lL~ 5 C(]&(u,cl) - u/$“)l]#]#) I C(IJUi(2)I

From (2.15)-(2.17),

IIF”’ J

c(I2.p

-

Uf)llL3

f4pI(L3

+

+

- u/$‘)Il;P

(Iluy’ll:cc

+

~~up&#.4j’1~

-

up)~L3)1’311z4j1)

(1 + IIuj(“)];G + IIu~1)](~~3))Iz4j1) - upll~3.

-

z&‘ll~~’

(2.17)

lemmas 2.1 and 2.5 it follows that

- Ff’(s)llL3/2 4

c,

.

1 +

c n=j,k

11~~2’(~)112,3 ,i, >

lI~“‘(4llL3.

(2.18)

Initial value problem

By (2.14), (2.18), lemmas

2.2, 2.6 and the singular

829

integral

inequality

we have

‘j IIW”‘(s)JI$ d&s “12 5 cI( W(‘)(O)IILZ (1

0

> +

C(T)&

* (t’l”2

+

P)

i

I=1

t IlW(‘)(s)(I;: (.r

0

ds “12. >

(2.19)

We next prove that

Ilw”‘(o)llL25 cffI14j - 4kllH1. By the mean value theorem U>“(O) - ZAj”(O) =

(~j(X)

we see that there exists a constant B (0 < 0 < 1) such that ‘1. - 4k(X)) exp 2id _m 14j(w)12dw ) ( i

+ &(x) =

(+j(X)

-

i

exp

23

(

@k(X))

+ 2iS&(x)

(I--

w

x

x exp 23 e ((1 . We apply lemma

(2.20)

2.1 and Holder’s

X --

‘X

ui --m (

2s

I@j(W)12d w) - ew(Zid x i

_-m

IQj(w)I’dw

jlw bkCw)12 dw)]

)

I+j(W)I’ - Id/c(W)12dW ) I$j(W)I’dw + (1 - 0)

inequality

b,h912 dw

>>

.

(2.21)

to (2.21) to obtain

II”j’l’(o)- u!‘)(“)llLz5 ll4j - 4kllL’ + cl14kIILzlll4jl’ - 14k1211L1 5 ll4j - 4klILZ + cffl14j - 4kllL2. By a direct calculation

(2.22)

we have

zp(O) - u12’(0)

x 14j(W)12 dw --m ) dw)] . x 14ji(W)12 dw) - ew(2id I:_ 14Awl12 -m

830

N. HAYASHI

In the same way as in the proof

of (2.22)

IIu,‘2’(0) - U/$2’(o)ll,z 5

llax(4j

4k)llL2+ cII l4jl’+j - 14k124kllLZ

-

l+jl’ - l~k1211L~ + (IIMIIL~ + cl1l+k124kllL2)lI 5 cHll+j - GkllH’.

(2.23)

By (2.22) and (2.23) we have (2.20). From (2.20) we see that the first part of the lemma holds valid for small T > 0. Hence we have the first part of the lemma for any T > 0 by the standard argument. We next prove the second part of the lemma. Multiplying both sides of (2.13) by IV(‘), integrating in x, taking the imaginary part, we obtain $ (lIV”‘(t)112,z = 2 Im(Fj”‘(t) 5 CjIFj”(t) We apply (2.18) to the above inequality

IlW”‘(t)ll;Zs c,

$

- F$“(t),

W”)(t))rz

- F~l)(t)llL2/~((W(l)(f)llL1.

to see that

* (1 +

ll~%l1:-) f: lw”‘(ollt~,

c

I=

n=j,k

from which it follows

I

that

lIw’“(t)ll:~ I (1 W”)(O)l($ +

c,

” I(

c

1+

+

5 IIw”‘(o)(I:*

cH

Ibf2’(dlt3 ,i, 1lW”‘Wll:3 ds

n=j,k

,O

t5/6

l t2/3

where we have used Holder’s inequality. Lemma W imply the second part of the lemma. The following LEMMA 2.8.

lemma

is important

Let wj be the solution

(I

,::

dsy’6]

IIu;2)(s)((;:.

in the proof

d)ll’,

2.6, (2.20) and the first part of the lemma

of the theorem.

of (2.1). Then we have for any T > 0

II’Yj(t)- WkCt)IIG~~’ di>“125 C(T) ’ cHI16j SUP

(\‘: 11 W”‘(s)((s

j,

IIY’j(t)-

vk(f)llH1

5

C’(T)

’ cHll$j

-

(DkllH’r

-

6kllH1,

05157

where C(T) Proof.

is a positive

constant

independent

of j and k.

We have vj(t, x) = uj”‘(t, x) exp

luj”(t,

.

W)12dw >

(2.24)

831

Initial value problem

By a simple calculation a,wj(t, X) = lJ,uj”(t,

_~ lu,!“(t, w)12dw

x) - 2i8(~4j(~)(‘~,!‘)(t, x)] exp

.

Since we have .

(2.25)

In the same way as in the proof of (2.20) we have by (2.24), (2.25) and lemma 2.5 IIVj(t3-x) - ‘Yk(f*x)II~l 5 Cm 1 + i

C

l=l

n=j,k

(

Ilu!i”(t)llLz ‘Ii, lIuj”(t)- ~i’)(t)llL~ (2-W >

and

II’J’j(fv X) -

Wk(f,

4 x)IIHI.35 c’ 1 + ; c (Il@(t& + I\@(t)\\& I=1 n=j,k > ( x

,il

(IIq(0

-

uf’(t)JJ,3

From (2.26), (2.27) and lemma 2.7 the lemma follows.

+

))u’[‘(t) J

-

z@(f)JIp).

(2.27)

n

Proof of theorem 1.1 completed. From lemma 2.8 we conclude that there exists a t,~= t+v(t,x) satisfying (1.2) such that for any T > 0 SUP

IIV’j(t)

-

W(f)l/H1

+

‘IIV’j(t)- W(t)IIS,jdt+ 0

1 0

OStST

as j -+ 00. It is easily seen that y is a unique solution of (1 .l>. This completes the proof of theorem 1.1. Remark 2.3. We note here that global existence in time of Schwartz class solutions of (1.1) has been shown by Lee [15] and smoothing property of global solutions of (1.1) is investigated by [12] under some conditions on data. Acknow/edgement-The for remarks.

author

would like to thank

Professor

Y. Tsutsumi

for valuable

Note added in proof-T. Ozawa and the author proved that a sufficient condition space for (1.1) is ]l$llr~ < &% by making use of the function E, defined in Section number X& is the same as that of (NLS) obtained in [6].

comments,

and to the referee

for global existence in the energy 2 and the method used in [6]. The

REFERENCES 1. Mro W., OGINO T., MINAMI K. & TAKEDA S., Modified nonlinear Schrodinger equation for Alfven waves propagating along the magnetic field in cold plasmas, J. P&s. Sot. Japan 41, 265-271 (1976). 2. MJOLHUS E., On the modulational instability of hydromagnetic waves parallel to the magnetic field, J. Plasma Phys. 16, 321-334 (1976). 3. TSUTSUMIM. & FUKUDA I., On solutions of the derivative nonlinear Schrodinger equation. Existence and uniqueness theorem, Funkcialaj Ekvacioj 23, 259-277 (1980).

832

N. HAYASHI

4. TSUTSUMIM. & FUKUDA I., On solutions

of the derivative

nonlinear

SchrGdinger

equation

II, Funkciuluj Ekvacioj

24, 85-94 (1981). 5. KENIG C. E., PONCE G. & VEGA L., Well-posedness of the initial value problem for the Korteweg-de Vries equation, J. Am. math. Sot. 4, 323-347 (1991). 6. WEINSTEIN M. I., Nonlinear SchrGdinger equations and sharp interpolation estimates, Communs Math. Phys 87,

567-576 (1983). I. FRIEDMAN A., Partial Differential Equations. Holt (Rinehart & Winston), New York (1969). 8. GINIBRE J. & VELO G., Scattering theory in the energy space for a class of nonlinear Schradinger

equations,

J. Math. pure appl. 64, 363-401 (1985). 9. STRICHARTZ R. S., Restriction of Fourier transforms to quadratic surfaces and decay of solutions of wave equations, Duke math. J. 44, 705-714 (1977). 10. KAUP D. J. & NEWELL A. C., An exact solution for a derivative nonlinear SchrBdinger equation, J. Math. Phys. 19,789-801(1978).

11. KUNDU A., Landau-Lifshitz and higher-order nonlinear systems gauge generated from nonlinear Schbdinger type equations, J. Math. Phys. 25, 3433-3438 (1984). 12. HAYASHI N., Analyticity and smoothing effect of solutions to the derivative nonlinear Schradinger equation, unpublished manuscript (1991). for nonlinear Schradinger equations and nonlinear groups, Funkcialaj Ekvacioj 30, 13. TSUTSUMI Y., L2-solutions

115-125 (1987). 14. HAYASHI N.,

Schrbdinger 15. LEE J. H.,

NAKAMITSU K. & TSUUUMI M., On solutions of the initial value problem for the nonlinear equations in one space dimension, Math. Z. 192, 637-650 (1986). Global solvability of the derivative nonlinear SchrGdinger equation, Trans. Am. math. Sot. 314,

107-118 (1989).

APPENDIX PROPOSITION A. 1. We assume

that u = u(t, x) is a solution

of

iv, + u,, + 2iS(o(zux = 0 and LJE C”(lR; H”(m)).

Then

(A.1)

V, = Eu and V, = Eu, satisfy

i(V,), + (VI), = 3idV*T 1 2, i(V2), + (V,), = -2idV*F * 1, where *x

Iv(t, w)l* dw

E = E(t, x) = exp id (

Proof. Multiplying

!-_

>

both sides of (A.l) by E, we obtain

i(V,), + (V,), + 2iSlV,12Vz = iE,v + 2E,u, + E,,u. We again

apply

(A.3

both sides of (A.1) by E?J, to have

I(V,), + (V,),, + 2iSlV, 12uxxE= -2iB(V,K By a simple calculation

+ V,y)V,

+ iE, v, + 2E,u,, + E,,u,

(A.3)

we see that

iE, = (i&V, Vz - V,q) + d21V,14)E, E,, = (i&V,% + V,V,) - SzIV1“)E, and 2E, = 2iS/ L’l*E. Therefore,

it follows

that

iEt + 2E xxu + E xxu = 2iSlVI’V 1 21 I 2 + 2i&‘*v i&v, From

(A.2)-(A.4)

the result follows.

+ 2E,u,, n

+ E,,u,

= 2i~lV,j’u,,E

+ 2iS[V212V,.

(A.4) (A.5)

833

Initial value problem For the convenience

of the reader

we give the proof

Proof of lemma 2.3. For simplicity we suppress the real part and integrating in x, we obtain

of lemma 2.3 which is different the subscript

j. Multiplying

from that in [4].

both

sides of (2.2) by q,

taking

By a direct calculation

(A.7) From (A.6) and (A.7) it follows

that (A.8)

/WV,w)l’dw m

>

we have

(A.9) and -tic5 By (A.8)-(A.lO)

IulZ(u,V - lIG)dx

I .m

we have the lemma.

n

= -*is

I ,R

IwIZ(w,P

- ViK)dx

+

~211wlli~.

(A.lO)