• No results found

On the solvability of the boundary value problem without initial condition for Schrödinger systems in infinite cylinders

N/A
N/A
Protected

Academic year: 2020

Share "On the solvability of the boundary value problem without initial condition for Schrödinger systems in infinite cylinders"

Copied!
9
0
0

Loading.... (view fulltext now)

Full text

(1)

R E S E A R C H

Open Access

On the solvability of the boundary value

problem without initial condition for

Schrödinger systems in infinite cylinders

Nguyen Manh Hung

1

and Nguyen Thi Lien

2*

*Correspondence: [email protected] 2Department of Mathematics, Hanoi National University of Education, Hanoi, Vietnam Full list of author information is available at the end of the article

Abstract

In this paper, we deal with the boundary value problems without initial condition for Schrödinger systems in cylinders. We establish several results on the existence and uniqueness of solutions.

Keywords: generalized solution; problems without initial condition; cylinders

Introduction

The initial boundary value for the Schrödinger equation in cylinders with base containing conical points was established in []. Such a problem for parabolic systems was studied in Sobolev spaces with weights []. The boundary value problem without initial condition for parabolic equation was investigated in [].

In the present paper, we consider the boundary value problem without initial condition for Schrödinger systems in cylinders. Firstly, following the method in [], we prove the existence of solutionsuh of problems with initial conditionst=h. Then, by lettingh→ –∞, the solvability of a problem without initial condition is obtained.

This paper is organized as follows. In the first section, we state the problem. In Sec-tion , we present the results on the unique solvability of problems with initial condiSec-tion for Schrödinger systems in cylinders. The well-posedness of the problem without initial condition is dealt with in Sections  and .

1 Setting the problem

Letbe a bounded domain inRn(n) with boundaryS=. Fora<b, setb a=

×(a,b),Sab=S×(a,b). If (a,b) =R, we useRto refer toandSRto refer toS. For each multi-indexα= (α, . . . ,αn)∈Nn, set|α|=α+· · ·+αnand∂α=∂xα· · ·

αn

xn.

Denoteu(x,t) = (u(x,t), . . . ,us(x,t)),Dαu= (Dαu, . . . ,Dαus),|Dαu|=

s

i=|Dαui|and

utj= ( ju

∂tj , . . . ,

∂jus

∂tj ) ,|utj|=si=| ju

i ∂tj |.

Let us introduce some functional spaces (see []) used in this paper.

We useHk() to be the space ofs-dimensional vector functions defined inwith the

norm

uHk()=

k

|α|=

udx

 

.

(2)

Denote byHk,l(b

a) the space consisting of all vector functionsu:ba−→Cssatisfying

uHk,l(b a)=

ba

k

|α|=

Dαu+

l

j=

|utj|dx dt

 

,

andHk,l(–γ,b

a) is the space of vector functions with the norm

uHk,l(–γ,ba)=

ba

k

|α|=

Dα

u+

l

j=

|utj|

e–γtdx dt

 

.

In particular,

uHk,(–γ,b a)=

k

|α|=

ba

Dαue–γtdx dt

 

.

Especially, we setL(–γ,ba) =H,(–γ,ba).

Denote byHk,l(–γ,ba) the completion of infinitely differentiable vector functions van-ishing nearSb

awith respect toHl,k(–γ,ba) norm.

Now we introduce a differential operator of order m

L(x,t,D) =

m

|p|,|q|=

(–)|p|Dpapq(x,t)Dq ,

whereapqares×smatrices with the bounded complex-valued components inR,apq=

aqp(a∗qpis the transposed conjugate matrix toapq). Set

B(t,u,v) =

m

|p|,|q|=

apqDquDpv dx, t∈R.

We assume further that the formB(t,·,·) isHm-elliptic uniformly with respect tot∈R, which means there exists a constantμ>  independent oftandusuch that

B(t,u,u)μu(·,t)Hm() (.)

for alluHm() and a.e.tR, whereHm() is a subspace ofHm(), the dense subset

of infinitely differentiable complexs-dimensional vector functions with compact support in.

Now we consider the following problem in the cylinderR:

(–)m–iL(x,t,D)uut=f(x,t) inR, (.)

∂ju ∂νj

SR= , j= , . . . ,m– , (.)

(3)

LetfL(–γ,R), a complex vector-valued functionu∈ ◦

Hm,(–γ,

R) is called a gen-eralized solution of problem (.)-(.) if and only if, for anyT> , the equality

(–)m–i

T

–∞

B(t,u,η)dt+

T–∞

uηtdx dt=

T–∞

fηdx dt (.)

holds for allηHm,(γ,R),η(x,t) =  withtT.

2 The unique solvability of problems with initial condition

Firstly, for anyh∈R, we study the following problem in the cylinderh :

(–)m–iL(x,t,D)vvt=f(x,t) inh , (.)

v|t=h= , x, (.)

∂jv

∂νj

Sh = , j= , . . . ,m– , (.)

whereνis the unit vector of outer normal to the surrounding surfaceSh .

The solutionv(x,t) is surveyed in the generalized sense. That meansvHm,(–γ,h ) is a generalized solution if and only if, for anyT> , we have

(–)m–i

T h

B(t,v,η)dt+

Th

vηtdx dt=

Th

fηdx dt (.)

for allηHm,(γ,

h ),η(x,t) =  for alltT.

In [], the unique solvability of problem (.)-(.) is studied in the caseh=  andf,ft

L∞(,∞;L()). Now, by the same method, we consider that problem in the caseh∈R

andf,ftL(–γ,h ).

Theorem . Assume that

(i) sup{|∂apqt |: (x,t)h , ≤ |p|,|q| ≤m}=μ<∞; (ii) ft,fL(–γ,h ).

Then,for allγ >γ=m μμ,m

=

|α|≤m,there exists a uniquely generalized solution v

Hm,(–γ,h )of problem(.)-(.)satisfying vHm,(–γ,

h)≤C

fL

(–γ,h)+ft

L(–γ,h)

, (.)

where C is a nonnegative constant independent of h,v,and f.

Proof The uniqueness is proved in a similar way as in []. We omit the details here. Now we prove the existence by the Galerkin approximating method. Suppose that{ϕk}∞k=is an

orthogonal basis ofHm() which is orthonormal inL(). For anyN∈N, we consider

the functionvN(x,t) =N

k=CNk(t)ϕk(x), where (CNk(t))Nk= is the solution of the ordinary

differential system

(–)m–iBt,vN,ϕk

vNt ϕkdx=

fϕkdx, (.)

(4)

So, multiplying both sides of (.) bydtd(CkN(t)) and taking the sum with respect tokfrom  toN, we arrive at

(–)m

m

|p|,|q|=

apqDqvNDpvNt dxi

vNt vN t dx=i

f vN t dx.

Adding this equation to its complex conjugate, integrating with respect totfromhtoT, and then integrating by parts, we get

(–)m

m

|p|,|q|=

apq(x,T)DqvN(x,T)DpvN(x,T)dx

= (–)m

m

|p|,|q|=

Th

∂apq

∂t D

qvNDpvNdx dt

– Im

f(x,T)vN(x,T)dx

Th

ftvNdx dt

. (.)

Using (.) and the Cauchy inequality, we receive from (.) that

vN(·,T)Hm()

mμ+

μ–

T h

vN(·,t)Hm()dt

+ 

(μ–)

f(·,T)L()+

T h

ft(·,t)L()dt

. (.)

Using the Gronwall-Bellman inequality, put α=mμ–μ+, from (.) we obtain

vN(·,T)Hm()≤α

T h

eα(Tt)f(·,t)L()+

T h

fs(·,s)L

()ds

dt

+ 

(μ–)

f(·,T)L()+

T h

ft(·,t)

L()dt

.

Multiplying both sides of this equation bye–γTand integrating with respect toTfromh

to∞, we get

vNHm,(–γ,h)≤

(μ–)

h

e–γTf(·,T)L()dT

+ 

(μ–)

h

e–γT

T h

ft(·,t)

L()dt dT

+ α

h

e–γT

T h

eα(Tt)f(·,t)L()dt dT

+ α

h

e–γT

T h

eα(Tt)

t h

fs(·,s)

L()ds dt dT. (.)

We denote byI,II,III,IV the terms from the first, second, third, fourth, respectively, of the right-hand sides of (.). We will give estimations for these terms. Firstly

I= 

(μ–)

f

(5)

and

II= 

(μ–)

h

ft(·,t)L()

t

e–γTdT dt=  γ (μ–)

ftL(–γ,h).

Because ofinf<<μ

+ μ– =

μ = γ, for anyγ >γ, we can choose>  satisfying α=

+

μ– < γ. Next, the termIIIcan be estimated by

III= α

h

e–αtf(·,t)

L()

t

e(αγ)TdT dt= α

γαf

L(–γ,h).

The last term,IV, is equal to

IV= α

h

fs(·,s)

L()

s

e–αt

t

e(αγ)TdT dt ds= α

γ(γα)ft

L(–γ,h).

Combining the above estimations, we get from (.) that

vNHm,(–γ,h)≤C

f

L(–γ,h)+ftL(–γ,h)

, (.)

where the constantCis independent ofh,N.

From this inequality, by standard weakly convergent arguments (see []), we can con-clude that the sequence{vN}

N=possesses a subsequence convergent to a vector function

vHm,(–γ,

h ), which is a generalized solution of problem (.)-(.). Moreover, it

fol-lows from (.) that (.) holds.

3 The uniqueness of generalized solution of problem (1.2)-(1.3)

Theorem . Ifγ > and|∂apqt |<μeγt,∀t∈R,∀|p|,|q| ≤m,problem(.)-(.)has no

more than one solution.

Proof Assume thatu(x,t) andu(x,t) are two generalized solutions of problem (.)-(.),

setu(x,t) =u(x,t) –u(x,t). For anyT> ,bT, denote

η(x,t) =

⎧ ⎨ ⎩ t

bu(x,τ), –∞ ≤tb,

, btT.

Thenη(x,T) = ,ηHm,(γ,T

–∞) andηt(x,t) =u(x,t), –∞ ≤tb.

From the definition of generalized solution, we obtain

(–)m

m

|p|,|q|=

b–∞

apqDpηtDqηdx dt+i

b–∞

|ηt|dx dt= . (.)

Adding (.) to its complex conjugate, we discover

(–)m

m

|p|,|q|=

b–∞ apq

(DpηDpη)

(6)

which leads to

m

|p|,|q|=

b–∞ ∂apq

∂t D

pηDqηdx dt

= lim

τ→–∞

m

|p|,|q|=

τ

b

apq(x,τ)Dpu(x,τ)Dqu(x,τ)dx dτ. (.)

Using the assumption of Theorem . and the Cauchy inequality, the left-hand sides of (.) can be estimated by

m

|p|,|q|=

b–∞ ∂apq

∂t D

pηDqηdx dtμ

m

|p|,|q|=

b–∞

DqηDpeγtdx dt

μ

m

|p|,|q|=

b–∞

Dqηeγt+Dpηeγt dx dt

μHm,(γ,b –∞).

From (.) we have

B(τ,η,η) =

m

|p|,|q|=

apq(x,τ)D(x,τ)D(x,τ)dxμη(·,τ) 

Hm(),

which implies

lim

τ→–∞

m

|p|,|q|=

apq(x,τ)D(x,τ)D(x,τ)dxμ lim

τ→–∞η(·,τ) 

Hm().

Thus

lim

τ→–∞η(·,τ)

Hm()Hm,(γ,b

–∞). (.)

Set

vp(x,t) =

–∞

t

Dpu(x,τ), –∞ ≤tb,

then

Dpη(x,t) =

t b

Dpu(x,τ)=vp(x,b) –vp(x,t).

Replacing them into (.), noting thatlimτ→–∞vp(x,t) = , yields m

|p|=

vp(x,b)dx

C

m

|p|=

b–∞

(7)

≤C

m

|p|=

b–∞

eγtvp(x,b)

+vp(x,t)

dx dt

Ceγb

m

|p|=

vp(x,b)dx+C

m

|p|=

b–∞

eγtvp(x,t)

dx dt.

Setting

J(t) =

m

|p|=

vp(x,t)dx,

we have

 –Ceγb J(b)C

b

–∞e

γtJ(t)dt.

So

J(b)≤C

b

–∞e

γtJ(t)dt, b

–∞,  γ ln

 C

,

where the positive constantCdepends only onμandμ.

Using the Gronwall-Bellman inequality, we get

J(t)≡ on

–∞,  γ ln

 C

.

Sou(x,t) =  almost everywheret∈(–∞,γ lnC]. Because of the uniqueness of the so-lution of the problem with initial condition (.)-(.), we implyu(x,t) =u(x,t) almost

everywheret∈R.

4 The existence of generalized solution

The generalized solution of problem (.)-(.) can be approximated by a sequence of so-lutions of problems with initial condition (.)-(.).

It is known that there is a smooth functionχ(t) which is equal to  on [,∞), is equal to  on (–∞, ] and assumes value in [, ] on [; ] (see [, Th. .] for more details). Moreover, we can suppose that all derivatives ofχ(t) are bounded. Leth∈(–∞, ] be an integer. Settingfh(x,t) =χ(t–h)f(x,t), we then get

fh(x,t) =

⎧ ⎨ ⎩

f(x,t) ifth+ ,

 ifth.

Moreover, iff,ftL(–γ,R),fh,fthL(–γ,R) and

fhL

(–γ,R)≤ f 

L(–γ,R), (.)

fthL

(–γ,R)≤C

ftL(–γ,R)+fL(–γ,R) , (.)

(8)

Let us consider generalized solutionuhandukof problems (.)-(.) in cylindersh

andk withf(x,t) is replaced byfh(x,t) andfk(x,t) respectively. Withh>k,uh can be

understood inHm,(–γ,

k ) withuh(x,t) = ,kth.

Defineukh(x,t) =uk(x,t) –uh(x,t), thenukh(x,t) is the generalized solution of problem

(.)-(.) in cylinderk withf(x,t) is replaced byfkh(x,t) =fk(x,t) –fh(x,t). According to (.),

ukh◦

Hm,(–γ,R)=ukh

Hm,(–γ,k)

CfhfkL(–γ,

k)+

fthftkL(–γ,

k )

.

Because

fhfkL(–γ, R)=f

hfk

L(–γ,k)=

h+

k

e–γtfhfkL

()dt

=

h+

k

e–γtχ(t–h) –χ(t–k)· fL()dt

≤

h+

k

e–γtfL()dt.

Because of the fact thatfL(–γ,R),lim

h+

k e–

γtf

L()dt=  whenh,k→–∞. So

limfhfkL(–γ,R)=  whenh,k→–∞. Repeating this argument, we discoverlimfthfk

tL(–γ,R)=  whenh,k→–∞. It follows that{uh}–h∞=is a Cauchy sequence anduhis

convergent touinHm,(–γ,

R).

In conclusion, we haveuhHm,(–γ,

R) satisfying

(–)m–i

T h

Bt,uh,η dt+

Th

uhηtdx dt=

Th

fhηdx dt (.)

for allT> ,ηHm,(γ,

h ),η(x,t) =  withtT.

Because of the fact thatuh(x,t) = ,fh(x,t) = ,th, (.) leads to

(–)m–i

T

–∞

Bt,uh,η dt+

T–∞

uhηtdx dt=

T–∞

fηdx dt (.)

for allT> ,ηHm,(γ,R),η(x,t) =  for alltT.

Forf(x,t)L(–γ,R), sendingh→–∞, (.) is written as

(–)m–i

T

–∞

B(t,u,η)dt+

T–∞

uηtdx dt=

T–∞

fηdx dt (.)

for allT> ,ηHm,(γ,

R),η(x,t) =  for alltT.

That meansu(x,t) is a generalized solution of problem (.)-(.). We obtain our main result.

Theorem . Assume that:

(9)

(ii) |∂apqt | ≤μ·eγt,for all(x,t)R,≤ |p|,|q| ≤m;

(iii) f,ftL(–γ,R). Then, for allγ >γ= m

μμ, m

=

|α|≤m, there exists a uniquely generalized solution

u(x,t)Hm,(–γ,

R)of problem(.)-(.)satisfying

u◦

Hm,(–γ,R)C

fL(–γ,R)+ftL(–γ,R)

.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors studied, read and approved the final manuscript.

Author details

1National Institute of Education Management, Hanoi, Vietnam.2Department of Mathematics, Hanoi National University of Education, Hanoi, Vietnam.

Acknowledgements

This research is funded by Vietnam National Foundation for Science and Technology Development (NAFOSTED) under grant number 101.01-2011.30.

Received: 29 January 2013 Accepted: 16 June 2013 Published: 1 July 2013 References

1. Hung, NM, Anh, CT: Asymptotic expansions of solutions of the first initial boundary value problem for the Schrodinger system near conical points of the boundary. Differ. Uravn.46(2), 285-289 (2010)

2. Hung, NM, Anh, NT: Regularity of solutions of initial-boundary value problems for parabolic equations in domains with conical points. J. Differ. Equ.245(7), 1801-1818 (2008)

3. Bokalo, NM: Problem without initial conditions for some classes of nonlinear parabolic equations. J. Sov. Math.51, 2291-2322 (1990)

4. Adams, RA: Sobolev Spaces. Academic Press, San Diego (1975)

5. Renardy, M, Rogers, C: An Introduction to Partial Differential Equations. Springer, Berlin (2004)

doi:10.1186/1687-2770-2013-156

References

Related documents