• Aucun résultat trouvé

On the extinction of the solution for an elliptic equation in a cylindrical domain

N/A
N/A
Protected

Academic year: 2022

Partager "On the extinction of the solution for an elliptic equation in a cylindrical domain"

Copied!
9
0
0

Texte intégral

(1)

A NNALES MATHÉMATIQUES B LAISE P ASCAL

T HEODORE K. B ONI

On the extinction of the solution for an elliptic equation in a cylindrical domain

Annales mathématiques Blaise Pascal, tome 6, n

o

1 (1999), p. 13-20

<http://www.numdam.org/item?id=AMBP_1999__6_1_13_0>

© Annales mathématiques Blaise Pascal, 1999, tous droits réservés.

L’accès aux archives de la revue « Annales mathématiques Blaise Pascal » (http://

math.univ-bpclermont.fr/ambp/) 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/

(2)

ON THE EXTINCTION OF THE SOLUTION FOR AN ELLIPTIC

EQUATION

IN A CYLINDRICAL DOMAIN

THEODORE K.BONI

ABSTRACT. - We obtain some conditions under which there is extinction in finite time of the solution of an

elliptic equation

in a

cylindrical

domain. We also show how the solution of our

elliptic equation approaches

its

stationary

solution.

Keywords. Asymptotic behavior, extinction,

dead core.

AMS classifications. 35 B

40,

35 J 60.

1. Introduction

Let 03A9 be a bounded domain in IRn with smooth

boundary

~03A9. We consider the

following boundary

value

problem

for

t):

~2u ~t2 + 0394u - 03BBf(u) = 0 in

03A9

(4, oo), ( I.l )

on aS~ x

(0, oo), (1.2)

=

uo(x)

in

~,

with 0

uo(x)

1,

(1.3)

where

h(x)

is a

nonnegative

continuous function on

~03A9, uo( x)

is a continuous

function in S~ which satisfies the

compatibility

condition

uo(x)

=

h(x)

on ~03A9.

(1.4)

For

positive

values of s,

f(s)

is a

positive

and

increasing

function with

f(O)

= 0.

Proposition

1.1. Under the above

assumptions,

there exists a solution u

of (1.1) -.(1.3).

The

proof

of the

proposition

is based on the

following lemma

Comparison

lemma 1.2

(maximum principle).

Let u, v E

C(03A9

x

[0, oo))

satis-

fying

the

following inequalities:

~2u ~t2 + Lu - f(u) ~ ~2v ~t2 + Lv - f(v) in 03A9 (0, ~),

u > v on an x

(0, ~), u(x, 0) > v(x, 0)

in St.

Then we have

u(x, t)

>

v(x, t)

in S2 x

(0, oo).

Proof. Assume that w = u-v takes

negative

values and let m = inf

w(x, t)

0.

The function

_

m(t)

= x E

IT}

(3)

14

is

obviously

continuous and

m(0) > 0, inf m(t)

= m. Introduce the function

L(t,C) = m 2 -

Ct.

We have

L(t, C) m(t)

in

~0, oo)

for C

large

and the infimum of all constants C with this

property,

call it

Co,

has the property that

Co

>

0, L(t, Co) m(t),

there is to > 0 with

L(to, Co)

=

m(to).

Let xo be such that

m(to)

=

w(xo, to).

Since

w(xo, to) 0,

i.e

u(xo, to) v(xo, to),

it follows from the second

inequality

of the lemma that xo is in H. Now we have

f (u(xa, ta))

,

w(x, to)

>

w(xo, to)

in 03A9 ~

(Lw)(xo, to)

>

0,

L(t, Co)

with

equality

for t = to ==~ 0

in contradiction to the assumed first

inequality

of the lemma. 0

Proof of

Proposition

1.1. Since Problem

(1.1)-(1.3)

has a

comparison lemma,

to prove our

proposition,

it suffices to show that

( 1.1 ) - (1.3)

has both a super-

solution and a subsolution

(see [10]).

Let us notice that 0 is a subsolution and 1 is a

supersolution,

which leads us to the result. D

By

a

stationary

solution of

(1.1) - (1.3),

we mean a function

s(x) satisfying

a f (s(x))

= 0 in

fI, (1.5)

s(x)

=

h(x)

on ~03A9.

(1.6)

Definition 1.3. We say that there is extinction in

finite

time

for

the solution

u(x, t) if

there exists a

finite

time

To

such that

u(x, t)

= 0

for

x E

03A9,

t >

To.

In this paper, we are interested in the extinction of the solution u.

Elliptic equations

in

cylindrical

domains have been the

subject

of research of many authors

(see [1], [2], [5], [6], [7], [8], [11], [12]

and the references cited

therein). Concerning

the

extinction,

Kondratiev and Veron in

[8]

have shown

that under some

assumptions,

there is extinction in finite time for the solution u in the case where

f(s)

= sP with p 1 and the Dirichlet condition

replaced by

that of Neumann. Our aim in this paper is to

generalize

this result

considering

the

problem

described in

(1.1) - (1.3).

We also obtain the

asymptotic

behavior

as t -~ oo of the solution u

showing

how it

approaches

its

stationary

solution.

We shall see in this paper that the

stationary

solution of u

plays

an

important

(4)

role in its

asymptotic

behavior as t ~ oo. This

phenomenon

is well known in

the

parabolic

case

(see,

for instance

[3], [4]).

Our paper is

organized

as follows.

In the next

section,

we

give

some conditions of extinction and in the last

section,

we

study

the

asymptotic

behavior of the solution u as t ~ oo.

2. Conditions of extinction

In this

section,

we show that under some

conditions,

there is extinction in finite time for the solution M of

(1.1)

2014

(1.3).

Introduce the function

= ~ /(cr)~7.

Let

J(s)

=

d03C3 F(03C3) (2.1)

and let be the inverse function of

J(~).

In this

notation,

the IVP

a()

=

-~/2AF(Q-()),

= 1,

(2.2)

has the solution

03B1(t) = K(t203BB) for

t > 0 if

/ ds F(s) = ~, (2.3)

03B1(t)

=

K(t203BB)

for t

20142014 / ds F(s)

if

/ ~10ds F(s)

oo,

(2.4)

= 0 for t ~

1 203BB 10ds F(s)

if

10ds F(s)

~.

(2.5)

Remark 2.1. If

/(.s)

= then

a()

is denned as follows:

if

p>~

a~)=e-~

if

p=l,

~)=[1-~~’~’~

if

pl,

where

[a:j+

==

max{x,0}.

Theorem 2.2.

(a) Suppose

that h(x) = 0.

7~

(5)

16

then there is extinction in

finite

time

for u(x, t).

.

(b)

Assume that

infx~03A9uo(x)

= p > 0.

If

i

ds _

~

-~’

then

u(x, t)

> 0 in 03A9 x

(0, ~).

Proof. (a)

Let

a(t)

be a solution of

(2.2). Then,

we

easily

show that

a(t)

is a

supersolution

of

(l.l) - (1.3)

and the maximum

principle implies

that

0

u(x,t) 03B1(t)

in 03A9 x

(0,T).

Therefore,

the result follows from

(2.5).

(b)

Let

v(t)

be a solution of the

following

differential

equation.

v’(t) = - 2aF(v(t~i, v(0)

= p.

(2.6)

Then

v(t)

is a subsolution of

(l.l) - (1.3).

The maximum

principle implies

that

u(x, t)

>

v(t).

It follows from = oo that

v(t)

> 0 for t >

0,

which

o

~(S)

leads us to the result. 0

In the above

theorem,

we have obtained the extinction in finite time of the solution

u of

( 1.1 ) - (1.3)

in the case where the

boundary

values are zero. The

question

which appears is what

happens

if the

boundary

values are

positive?

To answer

this

question,

we shall need the

following

definition:

Definition 2.3. Let

cp( x)

be the

stationary

solution

of

the solution u

of (1.1) - (1.3).

We say that there exists a dead core

for 03C6(x) if

there exists a set

no

c 03A9

such that

03C6(x)

= 0

for

x E

03A9o.

The

following

lemma which may be found in

[4]

will be used later. It

gives

us

information on the

following steady-state problem:

039403C6(x)

= i in

SZ, (2.7)

h(x)

on ~03A9.

(2.8)

Lemma 2.4. Let

h(x)

> 0.

Suppose

that

f E C2(o,1), f (o)

= 0,

f (1)

=1, > 0 on

(0,1),

either

f ~~(s)

> 0 on

(d,1)

or

f’~(s)

0 on

(o, I).

If

(6)

then there is no dead core

for

any ~.

If

-

a

F(s)

oa,

then a dead core exists

for sufficiently large a,

and

for

any xo E

03A9,

xo

belongs

to

the dead core.

If --- oo, for xa E

SI,

define

03BBo

as follows:

ao =

inf{03C6(xo, 03BB)

=

0}, (2.9)

a

where

03C6(x, a)

is the solution of

(2.7) - (2.8).

Now,

introduce our result in the case where the

boundary

values are

positive.

Theorem 2.5. Assume that

assumptions of

Lemma

2.4

hoLd. For

fixed

xo E

03A9,

choose 03BB >

03BBo,

where

03BBo

is

defined

in

(2.9).

(a~ If

-

a

F(S) --

oo,

then

u(xo, t)

= 0

for

1 1 ds

t >

1

---

.

_

2(a _ 03BBo)

a

F(s).

(b~ ~f

01 ds F(s) = o0

and

minx~03A9uo(x)

>

0,

then

u(xo, t)

> 0

for

all t.

Proof. (a)

Put

t)

=

~(t)

+

4~(x)~

where

j~(t)

is a solution of

;~~(t)

=

2(~ - ao)F~~(t))~ ~(~)

=1

and the solution of

(2.7) - (2.8)

with a =

~o.

We compute that

wtt + 0394w -

03BBf(w)

=

03B2"(t)

+

039403C6(x)

-

03BBf(w)

_

03BB - 03BBo)f(03B2) + 03BBof(03C6(x))

--

03BBf(w)

-

(a --

~o +

ao - a)f tw)

= o~

It is easy to see that

(7)

18

w(x, 0) > uo(x)

in Q.

The maximum

principle implies

that

w(x, t)

>

u(x,t)

in 03A9 x

(0, oo),

and the result follows from

(2.5)

and

(2.9).

The

proof

of Theorem 2.5

(b)

is as the

proof

of Theorem 2.2

(b).

0

3.

Asymptotic

behavior.

In this

section,

we show how the solution u of

( 1.1 ) - (1.3) approaches

its sta-

tionary

solution p. We assume that

uo( x)

>

cp(x)

in S~. The maximum

principle implies

that

u(x, t)

> in 03A9 x

(0, oo).

The

following

lemma will be used in the

proof

of the theorem in below.

Lemma 3.1. Let

f(8)

be a convex

function

with

f(O)

= 0 and let a and b be two

nonnegative

numbers. Then we have

f(a

+

b)

>

f(a)

+

f (b).

Proof.

Put a =

inf ~a, b~

and

p

=

sup~a, b~.

Since

f(8)

is convex with

f (o)

=

0,

then

using Taylor’s formula,

we

get

f(a+b)

=

f (~+a)

> ~

f (~)+af ~(a)

~

f (~)+f ~a)

=

I(a)+ f(b).

Hence the result. D

Theorem 3.2.

(a) If f

is convex, then

o

t)

-

r(xl 03B1(t),

where

a(t)

is a solution

of (2.2).

(b) If f

is concave, then

0 ~ u(x,t) -

03C6(x) ~ 03B6(t),

where

~(t)

is a solution

of

~ (t)

=

~fr(2)~~t)~ ~(0)

= l.

Proof. (a)

Put

t)

=

a(t)

+

~(x)~

We compute that

wtt + 0394w - 03BBf(w) =

03B1"(t)

+ 039403C6(x) -

03BBf(w)

=

03BB[f(03B1)

+

f(03C6x))

-

f(03C9)].

Since

f(s)

is convex, it follows from Lemma 3.1 that

wtt + 0394w -

03BBf(w)

0 in 03A9 x

(9, oo).

(8)

Obviously,

we have

t~ >_ t)

on aSZ x

(0, oo), w(x, 0) > ~uo(~)

in SZ.

It follows from the maximum

principle

that

t)

>

t)

in SZ x

(0, oo),

which

yields

the first result.

(b)

Let

z(~, t) _ ~(t) ~- ~(~)~

.

We

compute

that

zee + _

03B6"(t) + 039403C6(x) - 03BBf(z) - ‘a.f~(2)C(t) "~" a(f(~(~)) -.f(z)1~

Since

a f

is concave, then

using Taylor’s formula,

we

get

a~.f~z) - f{~P~~))1

?

’~f~~z)~= - 4~(~)) =’~f~~z)~(t)-

It follows from z 2 that

~(f(z) -

>

~f~(2)~(t),

which

implies

that

zit + 0394z - 03BBf(z) 0 in SZ x

(0, oo).

We

easily

show that

z(x,t) > u(~,t)

on aS2 x

(0, cxJ), z(x,0) >_ uo(x)

in SZ.

The maximum

principle implies

that

z(x,t) > u(~,t)

in

S2 x (O,oc),

which

gives

the second result. 0

References

[1]

T. K.

Boni,

On the

asymptotic

behavior

of

solutions

for

some semilinear

parabolic

and

elliptic equations of

second order with nonlinear

boundary

con-

ditions,

Nonl. Anal.

TMA,

to appear.

(9)

20

[2]

T. K.

Boni,

Sur le comportement

asymptotique

des solutions pour certaines

équations paraboliques

et

elliptiques

semi-linéaires du second

ordre,

C. R.

Acad. Sci.

Paris,

t.

327,

Série

I,

5

(1998),

p. 455-460.

[3]

C. Bandle - I.

Stakgold,

The

formation of

the dead core in

parabolic

reaction

diffusin problems

Trans. Amer. Math. Soc. Vol.

286,

1

(1984)

275-

293.

[4]

C. Bandle - T. Nanbu - I.

Stakgold,

Porous medium

equation

with

absorption

SIAM J. Math. Anal. Vol.

20,

5

(1998)

1268-1278.

[5]

G. V.

Grishina, Asymptotic

behavior

of

solutions to

weakly

nonlinear equa- tions in tube

domains,

Diff.

Equat.,

Vol.

31,

3

(1995),

429-434.

[6]

V. A. Kondratiev - O. A.

Oleinik,

Some results

for

nonlinear

elliptic equations

in

cylindrical

domains. in:

Operational

calculus

spectral theory, Birkhauser, 1992,

185-195.

[7]

V. A. Kondratiev - O. A.

Oleinik,

On

asymptotic of

solutions

of

non-

linear second order

elliptic equations

in

cylindrical

domains. in: "Nonlinear.

Diff.

Equat. Appl.", 22, Birkhauser, 1996,

160-173.

[8]

V. A. Kondratiev - L.

Véron, Asymptotic

behavior

of

solutions

of

some

nonlinear

parabolic

or

elliptic equations, Asymptotic Analysis,

14

(1997)

117-156.

[9]

M. H. Protter - H. F.

Weinberger,

"Maximum

principles

in differential

equations",

Prentice

Hall, Englewood Cliffs, NJ, 1967.

[10]

D. H.

Sattinger,

Monotone methods in nonlinear

elliptic

and

parabolic boundary

value

problems,

Indiana Univ. Math. J. Vol.

21,

11

(1972),

979-

1000.

[11]

L.

Veron, Comportement asymptotique

des solutions

d’équations elliptiques

semi-linéaires dans Rn. Annali Math. Pura

Appl., 127 (1981),

25-50.

[12]

L.

Veron, Equations

d’évolution semi-linéaires du second ordre dans L1 Rev. Roumaine Math. Pures

Appl.,

XXVII

(1982),

95-123.

[13]

W.

Walter,

"Differential- und

Integral-Ungleichungen", Springer, Berlin,

1964.

Theodore K. BONI Universite Paul Sabatier

UFR-MIG,

MIP

118 route de Narbonne 31062

Toulouse,

France.

E. Mail:

boni@mip.ups-tlse.fr

Références

Documents relatifs

In this paper, we propose a deterministic control interpretation, via “two persons repeated games”, for a broad class of fully nonlinear equations of elliptic or parabolic type with

This paper describes a fairly simple method for proving the classical solvability of certain fully nonlinear second order elliptic equations, provided the coefficient of the

MARCUS, Large solutions of semilinear elliptic equations: existence, uniqueness and asymptotic behavior, Jl. MARCUS, Large solutions of semilinear elliptic equations

MARCUS M., Large solutions of semilinear elliptic equations: existence, uniqueness and asymptotic behaviour, J. DYNKIN, A probabilistic approach to one class

V´eron, Uniqueness and asymptotic behavior of solutions with boundary blow-up for a class of nonlinear elliptic equations, Annal.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

This paper indeed contains new Liouville type results of independent interest for the solutions of some elliptic equations in half-spaces R N−1 × (0, +∞) with homogeneous

Veron, Uniqueness and asymptotic behaviour of solutions with boundary blow-up for a class of nonlinear elliptic equations, Ann. Veron, Existence and uniqueness results for