• 検索結果がありません。

We show that if the initial data of the following problemsut+uux+uxxx+η(Hux+Huxxx

N/A
N/A
Protected

Academic year: 2022

シェア "We show that if the initial data of the following problemsut+uux+uxxx+η(Hux+Huxxx"

Copied!
21
0
0

読み込み中.... (全文を見る)

全文

(1)

Electronic Journal of Qualitative Theory of Differential Equations 2005, No. 20, 1-21;http://www.math.u-szeged.hu/ejqtde/

SPATIAL ANALYTICITY OF SOLUTIONS OF A NONLOCAL PERTURBATION OF THE KDV EQUATION

BORYS ALVAREZ SAMANIEGO

IMECC-UNICAMP, C.P. 6065, 13083-970, Campinas, SP, Brazil.

Abstract. LetHdenote the Hilbert transform andη0. We show that if the initial data of the following problemsut+uux+uxxx+η(Hux+Huxxx) = 0, u(·,0) =φ(·) andvt+12(vx)2+vxxx+η(Hvx+Hvxxx) = 0, v(·,0) =ψ(·) has an analytic continuation to a strip containing the real axis, then the solution has the same property, although the width of the strip might diminish with time. Whenη >0 and the initial data is complex-valued we prove local well- posedness of the two problems above in spaces of analytic functions, which implies the constancy over time of the radius of the strip of analyticity in the complex plane around the real axis.

1. Introduction

We are interested in studying spatial analyticity of solutions of the following problems:

ut+uux+uxxx+η(Hux+Huxxx) = 0, u(·,0) =φ(·), (1) vt+1

2(vx)2+vxxx+η(Hvx+Hvxxx) = 0, v(·,0) =ψ(·), (2) where H denotes the Hilbert transform given byHf(x) = π1PR

−∞

f(y)

yxdy forf ∈ S(R) the Schwartz space of rapidly decreasingC(R) functions, P represents the principal value of the integral and the parameter η is an arbitrary nonnegative number. It is known that([Hf)(ξ) =isgn(ξ) ˆf(ξ), for allf ∈Hs(R), where

sgn (ξ) =

−1, ξ <0, 1, ξ >0.

Equation (1) was derived by Ostrovskyet.al. (see [9] for more details) to describe the radiational instability of long non-linear waves in a stratified fluid caused by internal wave radiation from a shear layer; the fourth term corresponds to thewave amplification and the fifth term represents damping. It models the motion of a homogeneous finite-thickness fluid layer with densityδ1, which moves at a constant speed U, slipping over an immobile infinitely deep stratified fluid with a density δ2 > δ1. The upper boundary of the layer is supposed to be rigid and the lower one is contiguous to the infinitely deep fluid. Here u(x, t) is the deviation of the

2000Mathematics Subject Classification. 35Q53, 35B30, 35A07.

Key words and phrases. Spatial analyticity, Hilbert transform, KdV equation.

Research supported by FAPESP/Brazil under grant No. 2002/02522-0.

EJQTDE, 2005 No. 20, p. 1

(2)

interface from its equilibrium position. Let us remark that some numerical results for periodic and solitary-wave solutions of equation (1) were obtained by Bao-Feng Feng and T. Kawahara [4].

The Cauchy problems associated to (1) and (2) were studied in [1], where it was proved that problems (1) and (2) are globally well-posed in Hs(R) for s≥1, considering real-valued solutions.

In this paper we are interested in proving that if the initial condition of the problem (1) (resp. (2)) is analytic and has an analytic continuation to a strip containing the real axis, then the solution of (1) (resp. (2)) has the same property.

Section 2 is devoted to studying the case when the solutions are real-valued on the real axis at any time, and η ≥ 0. Hence, the results obtained in [1] about the initial value problems associated to (1) and (2) will be helpful. We use the method developed by Kato&Masuda [8] which estimates certain families of Liapunov functions for the solutions, to prove global spatial analyticity of the solutions, but the width of the strip might decrease with time.

Section 3 shows that problems (1) and (2) admit a Gevrey-class analysis. For η >0, we prove local well-posedness of problem (1) (resp. (2)) in Xσ,s forσ >0 ands >1/2 (resp. s >3/2); here, the initial data can be complex-valued. So, if the initial data of problem (1) (resp. (2)) is analytic and has an analytic continuation to a strip containing the real axis, then the solution of (1) (resp. (2)) has the same property, maintaining the width of the strip in time. It should be mentioned that it was recently proved by Gruji´c&Kalisch [5] a result on local well-posedness of the generalized KdV equation (KdV is an abbreviation for Korteweg-de Vries) in spaces of analytic functions on a strip containing the real axis without shrinking the width of the strip in time; their proof uses space-time estimates and Bourgain-type spaces.

Here we do not make use of Bourgain spaces, we mainly use some properties of the Semigroup associated to the linear part of problem (1), namely Lemmas 3.1 and 3.2, to prove local well-posedness of problem (1) in Xσ,s. Moreover, proceeding as in [2], where Bona&Gruji´c studied some KdV-type equations, we prove for real-valued solutions and η ≥ 0 that if the initial state belongs to a Gevrey class, then the solution of (1) (resp. (2)) remains in this class for all time but the width of the strip of analyticity may diminish as a function of time.

Finally, in Section 4 we consider η ≥ 0 in (1) and complex-valued initial data in Xr-spaces forr >0. Similar as in [6], where analyticity of solutions of the KdV equation was studied, we use Banach’s fixed point theorem in a suitable function space in order to find a local solution of problem (1) that is analytic and has an EJQTDE, 2005 No. 20, p. 2

(3)

analytic extension to a strip around the real axis although the radius of the strip of analyticity in the complex plane around the real axis may decrease with time.

Notation:

• fˆ=Ff : the Fourier transform off (F1: the inverse of the Fourier transform), where fˆ(ξ) = 1R

eiξxf(x)dx forf ∈L1(R).

• k · ks, (·,·)s: the norm and the inner product respectively in Hs(R) (Sobolev space of ordersofL2 type), s∈R. kfk2s≡R

(1 +|ξ|2)s|fˆ(ξ)|2dξ.

k · k=k · k0: the L2(R) norm. (·,·) denotes the inner product onL2(R).

H(R)≡ ∩Hs(R).

• H: the Hilbert transform.

• B(X, Y): set of bounded linear operators onX toY. IfX =Y we writeB(X).

k · kB(X,Y): the operator norm inB(X, Y).

• S(r) ={x+iy∈C; x∈R, |y|< r}, forr >0.

A(r): the set of all analytic functionsf onS(r) such thatf ∈L2(S(r0)) for each 0< r0< r and thatf(x)∈Rforx∈R.

• A= (I−∂x2)1/2,Xσ,s=D(AseσA) the domain of the operatorAseσA.

• Lp={f;f is measurable onR,kfkLp<∞}, where kfkLp= R

|f(x)|pdx1/p

if 1≤p <+∞, and kfkL = ess supx∈R|f(x)|,f is an equivalence class.

• Lr={f ∈L2;kfk2Lr = (f, f)Lr = (cosh(2rξ) ˆf ,fˆ)<+∞}.

• Xr=

f∈L2;kfk2Xr = (f, f)Xr =P1

j=02j(cosh(2rξ)+ξsinh(2rξ)) ˆf ,f)ˆ <∞ .

• Yr=

f ∈L2;∂xf ∈L2,kfk2Yr = (f, f)Yr =P1

j=02j+2cosh(2rξ) ˆf ,fˆ)<+∞ .

• kfkpm,p=Pm

j=0k∂xjfkpLp, 1≤p <+∞. kfkm,=Pm

j=0k∂xjfk.

• Hp(r): the analytic Hardy space on the strip S(r).

Hp(r) ={F;F is analytic onS(r),kFkHp(r)= sup|y|<rkF(·+iy)kLp<∞}.

• Hm,p(r) ={F ∈Hp(r);kFkpHm,p(r)=Pm

j=0k∂zjFkpHp(r)<∞}.

• C(I;X) : set of continuous functions on the intervalI into the Banach spaceX.

• Cω(I;X) : the set of weakly continuous functions from I to X.

• <(z): the real part of the complex number z.

2. Real-valued initial data.

We deduce in this Section global analyticity (in space variables) of solutions of problems (1) and (2) when the initial data and the corresponding solution take real values on the real axis and supposing moreover that the initial data has an analytic continuation that is analytic in a strip containing the real axis. We will use the fact that problems (1) and (2) are globally well possed in Hs(R) fors ≥ 1, when the solution of the two previously mentioned problems take real values on the real axis EJQTDE, 2005 No. 20, p. 3

(4)

at any time. More precisely we have the following two theorems (for real-valued solutions) which can be found in [1].

Theorem 2.1. Let s ≥ 1. If φ ∈ Hs(R), then for each η > 0 there exists a unique u = uη ∈ C([0,∞);Hs(R)) solution to the problem (1) such that ∂tu ∈ C([0,∞);Hs3(R)).

Proof. See Theorem 4.2 in [1].

Theorem 2.2. Let s ≥ 1. If ψ ∈ Hs(R), then for each η > 0 there exists a unique v = vη ∈ C([0,∞);Hs(R)) solution to the problem (2) such that ∂tv ∈ C([0,∞);Hs3(R)).

Proof. See Theorem 4.1 in [1].

Theorem 2.3 (resp. 2.4) states that if the initial state has an analytic continu- ation that belongs to A(r0) for somer0 >0 then the solutionu(t) (resp. v(t)) of problem (1) (resp. (2)), with η≥0, also has an analytic continuation belonging to A(r1) for allt ∈ [0, T], wherer1 might decrease with time. Theorem 2.3, below, is an application of the method developed by Kato&Masuda in [8] to study global analyticity (in space variables) of some partial differential equations. Similar as in the proof of Theorem 2 in [8] we consider Hm+5(R) ≡ Z ⊂X ≡ Hm+2(R) and Φσ;m(v)≡ 12kvk2σ,2;mdefined on an appropriate open setO⊂Z, where

kfk2σ,s;m≡ Xm j=0

e2jσ

(j!)2k∂xjfk2s and kfk2σ,s≡ X j=0

e2jσ

(j!)2k∂xjfk2s, s∈R. Lemma 2.1. LetF(v)be defined by F(v)≡ −vvx−vxxx−η(Hvx+Hvxxx). Then there exist constants c, γ >0such that for every v∈Hm+5(R),

hF(v), DΦσ;m(v)i ≤c(η+kvk2σ;m(v) +γq

Φσ;m(v)∂σΦσ;m(v). (3) Proof. It is not difficult to see that DΦσ;m(v) =Pm

j=0 e2jσ

(j!)2(−∂x)jA4xjv, where A= (1−∂x2)1/2. Then

hF(v), DΦσ;m(v)i= Xm j=0

e2jσ (j!)2

− ∂xj(v∂xv), ∂xjv

2−η ∂xj(H∂xv+H∂x3v), ∂jxv

2

= Xm j=0

e2jσ (j!)2

− v∂xj+1v, ∂xjv

2−Qj(v)−η ∂xj(H∂xv+H∂x3v), ∂xjv

2

, where Qj(v) = Pj

k=1 j k

xkv∂xjk+1v, ∂xjv

2, and Q0(v) ≡ 0. By using Kato’s inequality (K) in the Appendix we have that

|(v∂xj+1v, ∂xjv)2| ≤ck∂xvk1k∂xjvk22≤ckvk2k∂xjvk22.

EJQTDE, 2005 No. 20, p. 4

(5)

Moreover

− ∂xj(H∂xv+H∂x3v), ∂xjv

2 =

Z

(1 +ξ2)2ξ2j(|ξ| − |ξ|3)|v(ξ)ˆ |2

≤ Z

(1 +ξ2)2ξ2j|ˆv(ξ)|2dξ=k∂jxvk22. (4) Then

hF(v), DΦσ;m(v)i ≤c(kvk2+η)Φσ;m(v)− Xm j=0

e2jσ

(j!)2Qj(v). (5) Now, using the Schwarz inequality and the formulakf gk2≤γ kfk2kgk1+kfk1kgk2 we get

|Qj(v)| ≤γ Xj k=1

j k

k∂xjvk2 k∂xkvk1k∂xjk+1vk2+k∂kxvk2k∂xjk+1vk1 . (6) We denote as in [8], bjej!k∂jxvk2, B2 ≡ Pm

j=0b2j = 2Φσ;m(v) and ˜B2 ≡ Pm

j=1jb2j = ∂σΦσ;m(v). By using (6) it follows, as a particular case of Lemma 3.1 in [8], that

Xm j=0

e2jσ

(j!)2|Qj(v)| ≤ γ Xm j=1

Xj k=1

bj

bk1

k (j−k+ 1)bjk+1+bjbkbjk

= γ

Xm k=1

bk1

k Xm j=k

bj(j−k+ 1)bjk+1

Xm k=1

bk

Xm j=k

bjbjk

≤ γB˜2 Xm k=1

1 k2

1/2 Xm k=1

b2k11/2

+γBB˜ Xm k=1

bk

√k

≤ 2γBB˜2+γBB˜ Xm k=1

1 k2

1/2 Xm k=1

kb2k1/2

≤4γBB˜2. (7) By replacing the inequality (7) into (5), the Lemma follows.

Next, we enunciate Lemma 2.4 in [8] and we give, for expository completeness, a proof of this trivial result.

Lemma 2.2. Letφn ∈A(r), n= 1,2, ...be a sequence withkφnkσ,2bounded, where eσ< r. Ifφn→0in H asn→ ∞, thenkφnkσ0,2→0for each σ0< σ.

Proof. Letσ0< σ. ChooseMsuch thatkφnkσ,2≤Mfor alln. Sincee2jσ

0

(j!)2k∂xjφnk22

M2

e2j(σ−σ0), it follows, by using the dominated convergence theorem, that

nlim→∞nk2σ0,2= X j=0

e2jσ0 (j!)2( lim

n→∞k∂xjφnk22) = 0.

EJQTDE, 2005 No. 20, p. 5

(6)

Theorem 2.3. Let η ≥ 0 and T > 0. Let u ∈ C([0, T];H(R)) be a solution of (1). If u(0) = φ ∈ A(r0) for some r0 > 0, there exists r1 > 0 such that u∈C([0, T];A(r1)).

Proof. Let us remark that φ ∈ A(r0) implies, by Lemma 2.2 in [8], that φ ∈ H(R). So, it follows from Theorem 2.1 and from Corollary 4.7 in [7] that u ∈ C([0, T];H(R)). We proceed as in the proof of Theorem 2 in [8], to prove that Φσ;mis a Liapunov family for (1) onO=Z =Hm+5(R), considering the functions:

α(r) = γ√

r, β(r) =c(η+M)r ≡Kr, for r ≥ 0, where M = maxt[0,T]ku(t)k2, and γ, c >0 are constants given by Lemma 2.1,ρ(t) = 12kφk2b,2eKt where b < σ0, eσ0 < r0, and

σ(t) =b−

√2γ

K kφkb,2(eK2t−1), t∈[0, T]. (8) We have thatu(t)∈A(eσ(t)), for allt∈[0, T]. Thenu(t)∈A(r1) for allt∈[0, T], where r1 = eσ(T). The continuity of u, as in the proof of Theorem 2 in [8], is a

consequence of Lemma 2.2 above.

Theorem 2.4. Let η ≥ 0 and T > 0. Let v ∈ C([0, T];H(R)) be a solution of (2). If v(0) = ψ ∈ A(r0) for some r0 > 0, there exists r1 > 0 such that v∈C([0, T];A(r1)).

Proof. We remark thatu≡vx∈C([0, T];H(R)) is a solution of (1) withu(0) = ψ0. Since ψ0 ∈ H(R) and kψ0kσ,0 ≤ kψkσ,1 for each σ such that eσ < r0 it follows, as a consequence of Lemma 2.2 in [8], that ψ0 ∈ A(r0). So, by Theorem 2.3, there existsr1 >0 such that vx∈C([0, T];A(r1)). Now, since v(t)∈H(R) andkv(t)k2σ,0≤supt[0,T]kv(t)k2+esupt[0,T]ku(t)k2σ,0<∞for eachσsuch that eσ < r1 and for all t ∈ [0, T], it follows that v(t)∈ A(r1) for all t ∈ [0, T]. The

continuity ofv follows as in the previous Theorem.

3. Gevrey Class Regularity.

Now, we make a Gevrey-class analysis of problems (1) and (2). Let us remark that, since each function f ∈ Xσ,s = D(AseσA) with σ > 0 and s ≥ 0, satisfies kfk2Lσ ≤ R

e|ξ||fˆ(ξ)|2dξ ≤ kfk2Xσ,s < ∞, it follows from Theorem 1 in [6] that f has an analytic extensionF ∈H2(σ), where σ > 0 is the radius of the strip of analyticity in the complex plane around the real axis. Theorem 3.1 (resp. Theorem 3.2) states that problem (1) (resp. (2)), forη >0 and complex-valued initial data, is locally well-posed in Xσ,s where σ > 0 is a fixed number and s is a suitable nonnegative number. Theorem 3.1 (resp. 3.2)) implies that if the initial condition of problem (1) (resp. (2)) is analytic and has an analytic continuation to a strip EJQTDE, 2005 No. 20, p. 6

(7)

containing the real axis, then the solution of (1) (resp. (2)) has the same property, without reducing the width of the strip in time.

Fort≥0 andξ∈R, let

Fη(t, ξ) = e(iξ3+η(|ξ|−|ξ|3))t

Eη(t)f = F1(Fη(t,·) ˆf), f∈L2(R). (9) It is not difficult to see that|Fη(t, ξ)| ≤eηt, for allt≥0 andξ∈R.

Lemma 3.1. Letη > 0. Then (Eη(t))t0 is a C0-semigroup on Xσ,s for σ >0 ands∈R. Moreover,

kEη(t)kB(Xσ,s)≤eηt. (10) Whenη= 0, kEη(t)kB(Xσ,s)= 1 for allt≥0.

Proof. Similar to the proof of Lemma 2.1 in [1].

Lemma 3.2. Let t >0, λ≥0, η >0, σ > 0and s∈R be given. Then Eη(t)∈ B(Xσ,s, Xσ,s+λ). Moreover,

kEη(t)φkXσ,s+λ ≤cλ

eηt+ 1 (ηt)λ/3

kφkXσ,s, (11) whereφ∈Xσ,s andcλ is a constant depending only onλ.

Proof.

kEη(t)φk2Xσ,s+λ ≤ cλ Z

(1 +ξ2)se2σ(1+ξ2)1/2|Fη(t, ξ)|2|φ(ξ)ˆ |2dξ +

Z

ξ(1 +ξ2)se2σ(1+ξ2)1/2e2ηt(|ξ|−|ξ|3)|φ(ξ)ˆ |2

≤ cλ

e2ηt+ sup

ξ∈R

ξe2ηt(|ξ|3−|ξ|)

kφk2Xσ,s. (12) On the other hand,

sup

ξ∈R|ξ|λeηt(|ξ|3−|ξ|)= sup

ξ0

ξλeηt(ξ3ξ)≤√

2λeηt+cλ 1

(ηt)λ/3. (13)

The lemma follows immediately from (12) and (13).

Next theorem proves, without using Bourgain-type spaces, local well-posedness to problem (1) withη >0 inXσ,sforσ >0,s >1/2, and complex-valued initial data.

Theorem 3.1. Letη > 0, σ > 0and s > 1/2 be given. If φ ∈ Xσ,s, then there exist T =T(s,σ,η,kφkXσ,s) >0 and a unique function u∈C([0, T];Xσ,s) satisfying the integral equation

u(t) =Eη(t)φ−1 2

Z t 0

Eη(t−t0)∂x(u2(t0))dt0. (14) EJQTDE, 2005 No. 20, p. 7

(8)

Proof. LetM,T >0 be fixed but arbitrary. Let us consider the map Af(t) =Eη(t)φ−1

2 Z t

0

Eη(t−t0)∂x(f2(t0))dt0, defined on the complete metric space

Θs,σ,η(T) ={f ∈C([0, T];Xσ,s); sup

t[0,T]kf(t)−Eη(t)φkXσ,s ≤M},

whereT >0 will be suitably chosen later. First, we prove that iff ∈Θs,σ,η(T) then Af ∈C([0, T];Xσ,s). Without loss of generality, we may assume thatτ > t > 0.

Then

kAf(t)−Af(τ)kXσ,s ≤ k(Eη(t)−Eη(τ))φkXσ,s +F(t, τ) +G(t, τ), whereF(t, τ) andG(t, τ) will be estimated below.

G(t, τ) ≡ Z τ

t

kEη(τ−t0)∂x(f2(t0))kXσ,sdt0

≤ cs

η(M+eηTkφkXσ,s)2(eη(τt)−1 + 3

2(η(τ −t))23)→0 as τ ↓t, where in the last inequality we have used Lemmas 3.1 and 3.2 (withλ= 1) and the fact thatXσ,s is a Banach algebra fors >1/2 andσ ≥0 (Lemma 6 in [2]). Since τ −t0 ≥t−t0, for all t0 ∈[0, t], it follows from Lemma 3.2 and from the triangle inequality that

k(Eη(t−t0)−Eη(τ−t0))∂x(f2(t0))kXσ,s≤cs(M+eηTkφkXσ,s)2 eη(Tt0)+(η(t−t0))13 ,

and the expression on the right hand side of the last inequality belongs toL1([0, t], dt0).

The fact thatk(Eη(t−t0)−Eη(τ−t0))∂x(f2(t0))kXσ,s →0 asτ ↓tis a consequence of the dominated convergence theorem. So, by using again the dominated convergence theorem, we have that

F(t, τ)≡ Z t

0 k(Eη(t−t0)−Eη(τ−t0))∂x(f2(t0))kXσ,sdt0 →0 as τ ↓t.

Now, we prove thatA(Θs,σ,η(T))⊂Θs,σ,η(T), forT = ˜T >0 sufficiently small. Let u∈Θs,σ,η(T). Then

kAu(t)−Eη(t)φkXσ,s ≤ Z t

0 kEη(t−t0)∂x(u2(t0))kXσ,sdt0

≤ cs

η (M+eηTkφkXσ,s)2h(T), (15) where h(T)≡eηT −1 +32(ηT)2/3. By choosingT = ˜T >0 sufficiently small, the right hand side of (15) is less thanM. Finally , we claim that there exists ˆT ∈(0,T˜] EJQTDE, 2005 No. 20, p. 8

(9)

such thatAis a contraction on Θs,σ,η( ˆT). Lett∈[0,T˜],u, v∈Θs,σ,η( ˜T). Then kAu(t)−Av(t)kXσ,s ≤ c1

Z t 0

eη(tt0)+ 1 (η(t−t0))1/3

k∂x(u2(t0)−v2(t0))kXσ,s−1dt0

≤ cs(M+eηtkφkXσ,s) sup

t0[0,T]˜

ku(t0)−v(t0)kXσ,s

· Z t

0

eη(tt0)+ 1 (η(t−t0))1/3

dt0

≤ cs

η(M+eηT˜kφkXσ,s)h( ˜T) sup

t0[0,T]˜

ku(t0)−v(t0)kXσ,s. So, by choosing ˆT ∈(0,T] such that˜ cηs(M+eηTˆkφkXσ,s)h( ˆT)<1, the claim follows.

Hence A has a unique fixed point u∈Θs,σ,η( ˆT), which satisfies (14). Uniqueness of the solutionu∈C([0,Tˆ];Xσ,s) follows from Proposition 3.2 in [1], which implies uniqueness of the solution in the classC([0,T];ˆ Hs(R)) fors >1/2.

Proposition 3.1. Problem (1) is equivalent to the integral equation (14). More precisely, if u ∈ C([0, T];Xσ,s), s > 1/2 is a solution of (1) then u satisfies (14). Reciprocally, if u ∈ C([0, T];Xσ,s), s > 1/2 is a solution of (14) then u∈C1([0, T];Xσ,s3)and satisfies (1).

Proof. Similar to the proof of Proposition 3.1 in [1]. However, here we use Lemmas

3.1 and 3.2.

Theorem 3.2. Let η >0, σ >0and s >3/2be given. Letψ∈Xσ,s. Then there exist T =T(s,σ,η,kψkXσ,s)>0and a uniquev∈C([0, T];Xσ,s)solution of (2).

Proof. Sinceψ∈Xσ,s, it follows thatφ≡ψ0∈Xσ,s1. By Theorem 3.1 and Propo- sition 3.1, there exist T = T(s,σ,η,kψkXσ,s) >0 and a unique u∈C([0, T];Xσ,s1) satisfying (14) and (1). Let us define

v(t)≡Eη(t)ψ−1 2

Z t 0

Eη(t−t0)u2(t0)dt0, t∈[0, T]. (16) It follows easily from (16) and from the uniqueness of the solution of (14) that

xv(t) =u(t). Sinceu∈C([0, T];Xσ,s1), it follows from Lemmas 3.1 and 3.2 that v∈C([0, T];Xσ,s). Now, by similar calculations as in the proof of Proposition 3.1 in [1], we have thatv∈C1([0, T];Xσ,s3) and satisfies (2).

Theorems 3.3 and 3.4, below, consider the case of real-valued solutions (on the real axis) to problems (1) and (2) respectively, forη≥0. So, Theorems 2.1 and 2.2 will be required again. The following theorem is proved similarly to Theorem 11 in [2], where the rate of decrease of the uniform radius of analyticity for KdV-type equations of the formut+G(u)ux−Lux = 0 was also studied, whereu=u(x, t), EJQTDE, 2005 No. 20, p. 9

(10)

forx, t∈R,Gis a function that is analytic at least in a neighborhood of zero inC, but real-valued on the real axis, andLis a homogeneus Fourier multiplier operator defined by Lu(ξ) =c |ξ|µu(ξˆ ), for someµ > 0. In the case of problem (1) we have thatLu(ξ) = [ξc 2−ηi(sgn(ξ)−ξ|ξ|)]ˆu(ξ).

Theorem 3.3. Letη ≥0 andT > 0. Suppose thatφ ∈Xσ0,s, for some σ0 > 0 and s > 5/2. Suppose moreover that φ(x) ∈ R for x ∈ R. Then the solution u of problem (1) satisfies u ∈ C([0, T];Xσ(T),s), where σ(t) is a positive monotone decreasing function given by (24).

Proof. Letη, T, σ0, andsbe as in the hypothesis of the theorem. Letr≡s−1>3/2.

By the Remark at the end of this Section we have that φ ∈A(σ0). So, by using Lemma 2.2 in [8], we have that φ ∈ H(R). Then u ∈ C([0, T];Hs(R)), which follows from Theorem 2.1 and from Corollary 4.7 in [7]. Letv≡ux, then

vt+v2+vxxx+η(Hvx+Hvxxx) +uvx= 0, v(0) =φ0. (17) Letσ∈C1([0, T];R) be a positive function such thatσ0<0 andσ(0) =σ0. Then

1 2

d

dtkv(t)k2Xσ(t),r−σ0(t)kv(t)k2Xσ(t),r+1/2 =<

Z

(1+ξ2)re2σ(t)(1+ξ2)1/2tˆv(t, ξ)ˆv(t, ξ)dξ.

It follows from the last expression and from (17) that 1

2 d

dtkv(t)k2Xσ(t),r −σ0(t)kv(t)k2Xσ(t),r+1/2 ≤I1+I2+ηkv(t)k2Xσ(t),r, (18)

where

I1≡ |(Areσ(t)Av2(t), Areσ(t)Av(t))|, I2≡ |(Areσ(t)A(uvx)(t), Areσ(t)Av(t))|.

I1andI2are particular cases of the corresponding ones in [2], takingG(u) =u. For the sake of completeness we estimate them here. Sincer >1/2, by using Lemmas 6 and 9 in [2], we have that

I1 ≤ crkAreσ(t)Av(t)k3≤crkArv(t)k3+ ˜crσ(t)kAr+1/3eσ(t)Av(t)k3

≤ crkArv(t)k3+ ˜crσ(t)kAreσ(t)Av(t)kkAr+1/2eσ(t)Av(t)k2. (19) Sincer >3/2, by using Lemma 10 in [2], we obtain

I2≤crkAr+1u(t)kkArv(t)k2+ ˜crσ(t)kAr+1eσ(t)Au(t)kkAr+1/2eσ(t)Av(t)k2. (20) Sinceu∈C([0, T];Hs(R)), it follows thatkAr+1u(t)k ≤c(r,T). Moreover

kAr+1eσ(t)Au(t)k2= Z

(1 +ξ2)re2σ(t)(1+ξ2)1/2|u(ξ, t)ˆ |2dξ+kAreσ(t)Av(t)k2

≤e20ku(t)k2r+ 2kAreσ(t)Av(t)k2≤c(r,T,σ0)+ 2kAreσ(t)Av(t)k2, (21) EJQTDE, 2005 No. 20, p. 10

(11)

wherec(r,T,σ0) is a positive constant depending only onr, T andσ0. Replacing the last inequalities into (20) and sincev∈C([0, T];Hr(R)), we get

I2≤c(r,T)+ ˜crσ(t) c(r,T,σ0)+√

2kAreσ(t)Av(t)k

kAr+1/2eσ(t)Av(t)k2. (22) Let us remark thatc(r,T)andc(r,T,σ0)are positive, continuous, non-decreasing func- tions of the variableT∈[0,+∞). Now, by using (19) and (22) into (18), we obtain

1 2

d

dtkv(t)k2Xσ(t),r −σ0(t)kv(t)k2Xσ(t),r+1/2

≤c(r,T)+c(r,T,σ0)σ(t) 1 +kv(t)kXσ(t),r

kv(t)k2Xσ(t),r+1/2+ηkv(t)k2Xσ(t),r. Then

d

dtkv(t)k2Xσ(t),r + 2 −σ0(t)−c(r,T,σ0)σ(t)(1 +kv(t)kXσ(t),r)

kv(t)k2Xσ(t),r+1/2

≤c(r,T)+ 2ηkv(t)k2Xσ(t),r. (23)

Inequality (23) implies thatv(t)∈Xσ(t),r for allt∈[0, T], where

σ(t) =σ0eKt, (24)

K=c(r,T,σ0)(1 +c(r,T,σ0,φ)eηT), andc(r,T,σ0,φ)= kAreσ0Aφ0k2+c(r,T)T1/2

. More precisely we have that

kv(t)kXσ(t),r ≤c(r,T,σ0,φ)eηT, t∈[0, T]. (25) Now, we will prove assertion (25). In fact let

T≡sup{T >0;∃!u∈C([0, T];Xσ(T),s) solution of (1), sup

t[0,T]ku(t)kXσ(t),s <∞}. We claim that T= +∞. Suppose by contradiction thatT<∞. It follows from Theorem 3.1, Proposition 3.1 and Theorem 1 in [5] (for η = 0) that T >0. Let T < T˜ . We have that

sup

t[0,T]˜

kv(t)kXσ(t),r ≤ sup

t[0,T˜]

ku(t)kXσ(t),s ≡M( ˜T)≡M.

By choosingσ(t) =σ0eKt˜ fort∈[0,T˜], where ˜K≡c(r,T ,σ˜ 0)(1 +M), we have that

−σ0(t)−c(r,T ,σ˜ 0)σ(t)(1 +kv(t)kXσ(t),r) =σ(t)c(r,T ,σ˜ 0)(M− kv(t)kXσ(t),r)≥0, for allt∈[0,T]. So, it follows from (23) that˜

d

dtkv(t)k2Xσ(t),r ≤c(r,T˜)+ 2ηkv(t)k2Xσ(t),r,

for allt∈[0,T˜]. Now, Gronwall´s inequality implies thatkv(t)kXσ(t),r ≤c(r,T ,σ˜ 0,φ)eηt, for allt∈[0,T˜], wherec(r,T ,σ˜ 0,φ)≡(kφ0k2Xσ0,r+c(r,T˜)T˜)1/2is a continuous, positive, non-decreasing function of ˜T ∈[0,∞). So, we can replace above the upper bound M byc(r,T ,σ˜ 0,φ)eηT˜ and take σ(t) = σ0eKt, K ≡c(r,T ,σ˜ 0)(1 +c(r,T ,σ˜ 0,φ)eηT˜), for t∈[0,T˜] . Sincec(r,T ,σ˜ 0)andc(r,T ,σ˜ 0,φ)are continuous, non-decreasing functions of EJQTDE, 2005 No. 20, p. 11

(12)

T˜∈[0,∞), we can chooseσ(t) =σ0eKtˆ , where ˆK≡c(r,T0)(1 +c(r,T0,φ)eηT) for allt∈[0, T], and applying again the local theory we obtain a contradiction. Fi- nally, it follows from (21) and (25) thatu(t)∈Xσ(t),s⊂Xσ(T),sfor allt∈[0, T].

Theorem 3.4. Letη ≥ 0and T >0. Suppose that ψ ∈ Xσ0,s, for some σ0 > 0 and s > 7/2. Suppose moreover that ψ(x) ∈ R for x ∈ R. Then the solution v of problem (2) satisfies v ∈ C([0, T];Xσ(T),s), where σ(t) is a positive monotone decreasing function given by (24).

Proof. Since ψ0 ∈ Xσ0,s1, and s−1 > 5/2, it follows from Theorem 3.3 that u≡vx∈C([0, T];Xσ(T),s1), whereσis given by (24). Making similar calculations to (21), we see thatv(t)∈Xσ(t),s for allt∈[0, T]. Finally, since

kv(t)−v(τ)k2Xσ(T),s ≤e22σ(T)kv(t)−v(τ)k2s1+ku(t)−u(τ)k2Xσ(T),s1,

it follows thatv∈C([0, T];Xσ(T),s).

Remark: Forr >0, we have







Xr,s⊂Lr, if s≥0.

Xr,s⊂Xr, if s≥3/2.

Xr⊂Lr.

If f ∈Lr, and f(x)∈R for x∈R, then f ∈A(r).

(26)

The first statement above was already proved at the beginning of this Section. The second part in (26) follows from the inequalitykfk2Xr ≤4kfk2Xr,s fors≥3/2. The fact thatXr⊂Lris obvious. Finally, suppose thatf ∈Lrandf(x)∈Rforx∈R; then we already know (see the first paragraph of this Section) thatfhas an analytic extensionF ∈H2(r); moreover, for every 0< r0< r we have

Z r0

r0

Z +

−∞ |F(x+iy)|2dxdy ≤ Z r0

r0

sup

|y|<r0

Z +

−∞ |F(x+iy)|2dx

dy≤2rkFk2H2(r). 4. Analyticity of Local Solutions of (1) in Xr-Spaces.

In this section we show that if the initial dataφof (1), with η ≥0, is analytic and has an analytic continuation to a strip containing the real axis, then there exists aT >0 such that the solutionu(t) of (1) has the same property for all t∈[0, T], but the width of the strip may decrease as a function of time. Similar to the proof of Theorem 2 in [6], we establish in Theorem 4.1 existence and analyticity of the solution of problem (1) simultaneously including the case when the initial condition is complex-valued on the real axis, more precisely whenφ∈Xσ0 for someσ0 >0.

The following lemma, which states a close relation between spacesXr andYr, will be mainly used to prove thatu∈Cω([0, T];Xσ(T)) in Theorem 4.1.

EJQTDE, 2005 No. 20, p. 12

(13)

Lemma 4.1. Yr is a dense subset of Xr, for r >0.

Proof. Letf ∈Yr. We have that Z

cosh(2rξ)|fˆ(ξ)|2dξ ≤ cosh(2r)kfk2+ Z

|ξ|>1

cosh(2rξ)|fˆ(ξ)|2

≤ cosh(2r)(kfk2+kfk2Yr), Z

ξsinh(2rξ)|fˆ(ξ)|2dξ ≤ sinh(2r)kfk2+ Z

|ξ|>1

ξ2cosh(2rξ)|fˆ(ξ)|2

≤ cosh(2r)(kfk2+kfk2Yr), and similarlyR

ξ3sinh(2rξ)|fˆ(ξ)|2dξ ≤cosh(2r)(kfk2+kfk2Yr). So, using the last three inequalities, it is not difficult to prove that

kfk2Xr ≤4 cosh(2r)(kfk2+kfk2Yr). (27) It follows from the last inequality thatYr⊂Xr. Now, let us consider the set

L2,0(R)=

g∈L2(R); ∃K >0, |ˆg(ξ)| ≤Ka.e. inR,g(ξ) = 0 a.e. inˆ {ξ;|ξ|> K} . Letg∈L2,0(R), thenP1

j=0

R ξ2j+2cosh(2rξ)|ˆg(ξ)|2dξ≤(K2+K4) cosh(2rK)kgk2, sog∈Yr. Now we will prove thatL2,0(R) is a dense set inXr. Letf be an element ofXr. For eachn∈N, take fn∈L2,0(R) given by

fcn(ξ) =

fˆ(ξ), if|ξ| ≤nand|fˆ(ξ)| ≤n 0, otherwise.

It follows easily from the definition of fn that |cfn(ξ)| ≤ |f(ξ)ˆ |, for all n ∈N and ξ ∈ R. Moreover, cfn(ξ) → fˆ(ξ) as n→ ∞, for all ξ ∈R. So, by the dominated convergence theorem, we have thatkfn−fkXr →0 asn→ ∞. This concludes the

proof.

Lemma 4.2. (i.) Suppose thatF ∈H2,2(r). Letf be the trace of F on the real line. Thenf ∈Yr and

kfkYr≤√

2kFkH2,2(r). (28)

(ii.) Conversely, suppose that f ∈Yr. Then f has an analytic extension F ∈ H2,2(r)and

kFkH2,2(r)≤p

10 cosh(2r)(kfk+kfkYr). (29) Proof. (i.) Since F ∈ H2,2(r), we see that ∂zF, ∂z2F ∈ H2(r). Then, by using Theorem 1 in [6], we have that f0, f00∈Lrand moreover

kfk2Yr =kf0k2Lr+kf00k2Lr ≤2(k∂zFk2H2(r)+k∂z2Fk2H2(r))≤2kFk2H2,2(r). EJQTDE, 2005 No. 20, p. 13

(14)

(ii.) By Lemma 4.1, Yr ⊂ Xr. So, it follows from Lemma 2.1 in [6] that f has an analytic extension F on S(r) such that k∂zFkH1,2(r) ≤ √

2kfkYr and kFkH1,2(r)≤√

2kfkXr. Now, using the last inequalities, we have that kFk2H2,2(r) = kFk2H1,2(r)+k∂z2Fk2H2(r)≤ kFk2H1,2(r)+k∂zFk2H1,2(r)

≤ 2(kfk2Xr+kfk2Yr)≤10 cosh(2r)(kfk2+kfk2Yr),

where in the last inequality we have used (27).

Lemma 4.3. Letr >0. Suppose that F, G∈H2,2(r). Then F G∈H2,2(r) and kF GkH2,2(r)≤ckFkH2,2(r)kGkH2,2(r). (30) Proof. Using Leibniz’s rule and Sobolev’s inequality (kGkH(r)≤ckGkH1,2(r)) we have that

k∂z2(F G)kH2(r) ≤ k∂z2FkH2(r)kGkH(r)+ 2k∂zFkH(r)k∂zGkH2(r)

+kFkH(r)k∂z2GkH2(r).

≤ ckFkH2,2(r)kGkH2,2(r).

Moreover, by Lemma 2.3 in [6], we have thatkF GkH1,2(r)≤ckFkH1,2(r)kGkH1,2(r). SincekF Gk2H2,2(r)=kF Gk2H1,2(r)+k∂z2(F G)k2H2(r), the proof of the Lemma follows

from the last two inequalities.

Lemma 4.4 and Corollary 4.1, below, are particular cases of Lemma 2.4 and its Corollary in Hayashi [6] respectively.

Lemma 4.4. There exists a polynomial ˜aof which coefficients are all nonnegative with the following property: If r >0and f, g, v∈Xr∩Yr, then

|(f ∂xf−g∂xg, v)Xr| ≤˜a(kfkXr,kgkXr) kfkYrkf−gkXr+kf−gkYr

kvkXr+kvkYr

. Corollary 4.1. There exists a polynomiala˜1with nonnegative coefficients such that iff, v∈Xr∩Yr, then

|(f ∂xf, v)Xr| ≤a˜1(kfkXr)kfkYr kvkXr+kvkYr

. (31)

Now, we state the main theorem of this Section.

Theorem 4.1. Let η ≥0. If φ ∈ Xσ0 for some σ0 > 0, then there exist aT = T(kφkXσ0, η, σ0) > 0 and a positive monotone decreasing function σ(t) satisfying σ(0) = σ0 and such that (1) has a unique solution u ∈ Cω([0, T];Xσ(T)). When η= 0, u∈C([0, T];Xσ(T)).

EJQTDE, 2005 No. 20, p. 14

(15)

Proof. Letη ≥0, σ0 >0, and φ∈Xσ0. Letσ(t) =σ0eAt/σ0, where the positive constantAwill be conveniently chosen later. ForT >0, consider the space B(T) =

f : [0, T]×R7→C;kfk2B(T)= sup

0tTkf(t)k2Xσ(t)+A Z T

0 kf(t)k2Yσ(t)dt <+∞ and

Bρ(T) ={f ∈B(T);kfkB(T)≤ρ}.

Letρ= 4kφkXσ0. Forv ∈Bρ(T), we define the mapping M byu=M v, whereu is the solution of the linearized problem

tu+∂x3u+η(H∂xu+H∂x3u) =−v∂xv

u(0) =φ. (32)

More precisely ucan be obtained as follows. Take the Fourier transform to (32), then

tu(t, ξ)ˆ −iξ3u(t, ξ) +ˆ η(−|ξ|+|ξ|3)ˆu(t, ξ) =−dvvx(t, ξ).

Now, integrating the last expression between 0 andt, it follows that ˆ

u(t, ξ) =e(iξ3+η(|ξ|−|ξ|3))tφ(ξ)ˆ − Z t

0

e(iξ3+η(|ξ|−|ξ|3))(tτ)vvdx(τ, ξ)dτ, (33) where the last integral is well defined, since

Z t

0 |dvvx(τ, ξ)|dτ = Z t

0

dv(τ)∗v[x(τ)

(ξ)dτ = Z t

0

Z ˆ

v(τ, ξ−ω)ωˆv(τ, ω)dωdτ

≤ Z t

0 kv(τ)kkv(τ)k1dτ ≤ρ2t, and in the last inequality we have used the fact that

kv(τ)k1≤ kv(τ)kXσ(τ) ≤ kvkB(T)≤ρ.

Let us chooseA, T >0 such that

eAT /σ0 > 1

2. (34)

Let v ∈ Bρ(T). It will be proved that u = M v ∈ Bρ(T), for suitably chosen T, A >0. Now, it is not difficult to see that

d

dtku(t)k2Xσ(t) = 2σ0(t) X1 j=0

Z

ξ2j ξsinh(2σ(t)ξ) +ξ2cosh(2σ(t)ξ)

|u(t, ξ)ˆ |2

+2<

X1 j=0

Z

ξ2j cosh(2σ(t)ξ) +ξsinh(2σ(t)ξ)

tu(t, ξ)ˆˆ u(t, ξ)dξ.

EJQTDE, 2005 No. 20, p. 15

(16)

Taking the Fourier transform to (32), multiplying both sides of the obtained equa- tion byξ2j cosh(2σ(t)ξ) +ξsinh(2σ(t)ξ)

ˆ

u(t, ξ), integrating with respect toξ, sum- ming the terms for j= 0,1 and finally taking the real part we get

1 2

d

dtku(t)k2Xσ(t)−σ0(t) X1 j=0

Z

ξ2j ξsinh(2σ(t)ξ) +ξ2cosh(2σ(t)ξ)

|u(t, ξ)ˆ |2

+η X1 j=0

Z

ξ2j cosh(2σ(t)ξ) +ξsinh(2σ(t)ξ)

(−|ξ|+|ξ|3)|u(t, ξ)ˆ |2

=−<(vvx, u)Xσ(t).

Sinceσ0(t) =−AeAtσ0 and|ξ| − |ξ|3≤1, for allξ∈R, it follows that 1

2 d

dtku(t)k2Xσ(t)+AeAtσ0ku(t)k2Yσ(t) ≤ −<(vvx, u)Xσ(t)+ηku(t)k2Xσ(t). Using (34) and Corollary 4.1, it follows from the last inequality that

1 2

d

dtku(t)k2Xσ(t)+A

2ku(t)k2Yσ(t) ≤a(ρ)kv(t)kYσ(t)(ku(t)kXσ(t)+ku(t)kYσ(t))+ηku(t)k2Xσ(t), where a(·) is a polynomial with nonnegative coefficients. Now integrating the last inequality from 0 tot we get

sup

0tTku(t)k2Xσ(t)+A Z T

0 ku(t)k2Yσ(t)dt≤2

kφk2Xσ0+ 2a(ρ) Z T

0 kv(t)k2Yσ(t)dt1/2

· Z T

0 ku(t)k2Xσ(t)dt1/2

+ Z T

0 ku(t)k2Yσ(t)dt1/2 + 2η

Z T

0 ku(t)k2Xσ(t)dt .

Then,

kuk2B(T) ≤ ρ2

8 +4ρa(ρ)

√A

√T+ 1

√A

kukB(T)+ 4ηTkuk2B(T)

≤ ρ2 8 +1

2

4ρa(ρ)

√A

√T+ 1

√A 2

+ (1

2+ 4ηT)kuk2B(T). By choosingT >0 small enough such that

ηT < 1

12, (35)

we have that

kuk2B(T)≤ 3

4+48a2(ρ) A

√T+ 1

√A 2

ρ2.

Now we takeA, T >0 such that a(ρ)√

A

√T+ 1

√A < 1

8√

3. (36)

So, choosingA, T >0 such that (34)-(36) are satisfied, it follows that u=M v ∈ Bρ(T).

Now let us prove that the mapping M is a contraction, defined from Bρ(T) into EJQTDE, 2005 No. 20, p. 16

参照

関連したドキュメント

(6) It is well known that the dyadic decomposition is useful to define the product of two distributions.. Proof of Existence Results 4.1. Global existence for small initial data..

The objective of this paper is to apply the two-variable G /G, 1/G-expansion method to find the exact traveling wave solutions of the following nonlinear 11-dimensional KdV-

It is not a bad idea but it means that since a differential field automorphism of L|[x 0 ] is given by a birational transformation c 7→ ϕ(c) of the space of initial conditions, we

In this section we state our main theorems concerning the existence of a unique local solution to (SDP) and the continuous dependence on the initial data... τ is the initial time of

boundary condition guarantees the existence of global solutions without smallness conditions for the initial data, whereas posing a general linear boundary condition we did not

In this article we consider the problem of unique continuation for high-order equations of Korteweg-de Vries type which include the kdV hierarchy.. It is proved that if the difference

The Dubrovin–Novikov procedure is well justified in the averaging of the Gardner–Zakharov–Faddeev bracket on the m-phase solutions of KdV for any m and provides a local

However, few efforts are devoted to the existence of global attractor for generalized dissipative four-order KDV equation with nonlinearity in unbounded domains.. When the equation