• Nenhum resultado encontrado

Regularity of Weak Solutions to Some Anisotropic Elliptic Equations (II)

N/A
N/A
Protected

Academic year: 2017

Share "Regularity of Weak Solutions to Some Anisotropic Elliptic Equations (II)"

Copied!
7
0
0

Texto

(1)

Regularity of Weak Solutions to Some Anisotropic

Elliptic Equations

GAO Yanmin

College of Mathematics and Information Science, Hebei University, Baoding, 071002, China.email:1245712005@qq.com

GAO Hongya

College of Mathematics and Information Science, Hebei University, Baoding, 071002, China.email:ghy@hbu.cn

Abstract

We consider boundary value problem of the form This paper we deals with the system of N partial differential equations

n

X

i=1

Di(aαi(x, Du(x))) = n

X

i=1

DiFiα(x), α= 1,· · ·N.

We show that regularity of boundary datumu forcesuto have regular-ity as well, provided we assume suitable ellipticregular-ity and growth conditions on aαi.

Mathematics Subject Classification: 49N60, 35J60.

Keywords: Regularity, weak solution, anisotropic, elliptic systems.

1

Introduction

In a recent paper [1], Leonetti and Petricca considered the system ofN partial differential equations

n

X

i=1

Di(aαi(x, Du(x))) = 0, x∈Ω, α= 1,2,· · ·, N. (1.1)

Under the boundary condition

(2)

the authors showed that higher integrability of the boundary datum u∗ forces solutions u to have higher integrability as well, provided aα

i satisfy suitable

ellipticity and growth conditions. Among all the results, they obtained a the-orem see [1, Thethe-orem 1.3].

In this paper, we consider a more general problem. We refer the reader to [1] for the notations and symbols used in this paper. Let us consider the following system of N partial differential equations

n

X

i=1

Di(aαi(x, Du(x))) = n

X

i=1

DiFiα(x), α= 1,· · ·, N, (1.3)

and suppose that the Carath´eodory functions aα

i(x, z) : Ω×R n

→ R satisfy the monotonicity condition

ν

n

X

i=1

|zi −zei|pi ≤ n

X

i=1

(aαi(x, z)−aαi(x,ze))(ziα−zeiα), (1.4)

for some positive constant ν,and

|aαi(x, z)| ≤c(1 + n

X

j=1

|zj|pj) 1−pi1

, i= 1,2,· · ·, n, α= 1,2,· · ·, N, (1.5)

We work in Marcinkiewicz spaces: if q >1, then the spaceMm(Ω) consists of measurable functions g on Ω such that

sup

t>0

t|{x∈Ω :|g(x)|> t}|m1 <∞.

This condition is equivalently stated as

|||g(x)|||m = sup E⊂Ω,|E|>0

1

|E|m1′

Z

E|g(x)|dx <∞,

where m′ is the conjugate exponent of q, 1 m +

1

m′ = 1. We recall that L q(Ω)

is a proper subspace of Lqweak(Ω), and if g ∈ L q

weak(Ω) for some q > 1, then

g ∈Lq−ε(Ω) for every 0< εq1.

The following Sobolev type inequality is also proved: there exists a positive constant c, depending only on Ω, such that

kvkLr(Ω) ≤c

n

Y

i=1

k∂ivk 1

n

Lpi(Ω), ∀r ∈[1, p∗], (1.6)

for any v∈C1

0(Ω) where pi >1 for i= 1,2,· · ·, n. In the following the letter c

(3)

of (1.1), withm > p′, we consider the casep∞= ¯p∗, wherep∞ = max{pn,p¯∗},

pn = max{pi}, and we also consider the elliptic systems problem withf ∈Mm

too, not only this but also our method is different from the previous one.For some recent developments on anisotropic functionals and anisotropic elliptic equations and systems, see[2-3].

2

Preliminary Notes

We mow introduce some symbols used in this paper. Let Tk(u) is the usual

truncation of u at level k >0, that is,

Tk(u) = max{−k,min{k, u}}.

Moreover, let

Gk(u) =u−Tk(u).

Definition 2.1 A function u ∈ u∗+W01,(pi)(Ω;RN) is called a solution to the boundary value problem

  

n

P

i=1Di(a α

i(x, Du(x))) =fα(x), x∈Ω,

u(x) =u∗(x), x∈∂Ω,

(2.7)

if

Z

Ω n

X

i=1 N

X

α=1

i(x, Du(x))·Diϕα(x)dx =

Z

Ω N

X

α=1

fα(x)·ϕα(x)dx, (2.8)

holds true for any ϕ ∈W1,(pi)

0 (Ω;RN), α= 1,2,· · ·, N.

3

Main Results

These are the main results of the paper.

Theorem 3.1 Let f ∈ Mm(Ω;RN), u

∗ ∈W1,1(Ω;RN) with Diu∗ ∈Mpim,

i= 1,2,· · ·, n, and under previous assumptions (1.4)−(1.5),

i) If m> np, then there exists a weak solution u for the problem (1.3), andu−u∗

is bounded;

ii) If m= n

p, then there exists a weak solution u for the problem (1.3) and a

constant β >0 such that Z

Ωe

β|u−u∗|<; iii) If (p∗)′

< m < np, then there exists a weak solution u for the problem(1.3)

and u−u∗ belongs to Ms with s satisfies

s = mp

(p1)

mp+p∗mp∗ =

mn(p−1)

(4)

Proof of Theorem 3.1. Let us fix the componentβ ∈ {1,2,· · ·, N}and let us consider the test function

ϕ = (0,· · ·,0, ϕβ,0,· · ·,0),

where, for k ∈(0,+∞) we takeϕβ =G

k(uβ −uβ∗) in (2.8). We have

n X i=1 Z Ωa β

i(x, Du)DiGk(uβ−uβ∗) = Z

Ωf β

Gk(uβ−uβ∗),

because n X i=1 Z Ωa β

i(x, Du)DiGk(uβ−uβ∗) =

n

X

i=1

Z

Ak

aβi(x, Du)Di(uβ−uβ∗) = Z

Ak

fβ(uβ−uβ),

where Ak ={|uβ−uβ∗| > k}. Hence by (1.4), (1.5) and Young inequality we obtain that ˜ ν n X i=1 Z Ak

|Diuβ −Diuβ∗|pi

≤ n X i=1 Z Ak

(aβi(x, Du)−aβi(x, Du∗))(Diuβ−Diuβ)

= Z

Ak

fβ(uβ −uβ)−

n

X

i=1

Z

Ak

aβi(x, Duβ∗)(Diuβ −Diuβ∗)

Z

Ak

|fβ(uβ −uβ)|+

n

X

i=1

Z

Ak

|aβi(x, Du∗)||Diuβ−Diuβ∗|

Z

Ak

|fβ(uβ −uβ)|+c

n X i=1 Z Ak (1 + N X j=1

|Diu∗|pj)1−

1

pi(Diuβ Diuβ

∗)

Z

Ak

|fβ(uβ −uβ)|+c(ε)

n X i=1 Z Ak (1 + n X i=1

|Diu∗|pi) +ε

n

X

i=1

Z

Ak

|Diuβ−Diuβ|pi.

Then Z

Ak

|Diuβ−Diuβ|pi ≤c(

Z

Ak

|fβ(uβ−uβ)|+|Ak|+ n

X

i=1

Z

Ak

|Diu∗|pi). (3.10)

Therefore, by (1.6), with r=p∗, H¨older inequality and (3.10), we get

c( Z

Ak

|uβ−uβ|p∗

)p1∗

≤ n Y i=1 ( Z Ak

|Diuβ−Diuβ∗|pi)

1

pin

≤ (

Z

Ak

|fβ(uβ −uβ)|+|Ak|+ n

X

i=1

Z

Ak

|Diu∗|pi)

1

p

≤ (

Z

Ak

|fβ|(p∗)′)

1 (p∗)

( Z

Ak

|uβ−uβ|p∗

)p1∗ +|A k|+

n

X

i=1

Z

Ak

|Diu∗|pi]

1

p.

(5)

Since f ∈Mm(Ω) and D

iu∗ ∈Mpim, and m≥(p∗)′, we have Z

Ak

|f|(p∗)

≤c|Ak|1− (p∗)

m ,

Z

Ak

|Diu∗|pi ≤c|Ak|1− 1

m.

Then by applying Young inequality, |Ak| 1

m ≤ |Ω|

1

m and (3.11) becomes

c( Z

Ak

|uβ−uβ|p∗

)pp∗

≤ (|Ak|1− (p∗)′

m )

1 (p∗)

( Z

Ak

|uβ−uβ|p∗

)p1∗ +|A

k|+|Ak|1− 1

m

≤ |Ak| 1 (p∗)′−

1

m

( Z

Ak

|uβ −uβ|p∗)p1∗ + 1

k|Ak|

1 (p∗)′−

1

m · |Ak|

1

m(

Z

Ak

|uβ−uβ|p∗)p1∗

+1

k|Ak|

m1 Z

Ak

|uβ −uβ| ≤ |Ak|

1 (p∗)′−m1

( Z

Ak

|uβ −uβ|p∗

)p1∗ +c|A k|

1 (p∗)′−m1

( Z

Ak

|uβ−uβ|p∗

)p1∗

+c|Ak|− 1

m(

Z

Ak

|uβ −uβ|p∗

)p1∗|A k|

1 (p∗)

≤ c(ε)|Ak| ( 1

(p∗)′− 1

m)(p)

+ε( Z

Ak

|uβ−uβ|p∗

)pp∗.

Hence by applying H¨older inequality with exponentsp∗ and (p∗)′toR

Ω|Gk(uβ−

∗)|= R

Ak|u

βuβ

∗| and by simplifying, we obtain

Z

Ω|Gk(u β

−uβ)| ≤c|Ak| ( 1

(p∗)′−m1)p−11 +1−p1∗

. (3.12)

We define g(k) =R|Gk(uβ−uβ∗)| and we recall that g

(k)=−|Ak|, for almost

every k (see[4], [5]). We obtain, from(3.12), that

g(k)γ1 ≤ −cg′(k),

with γ = ( 1

(p∗)′ − m1)p11 + 1− p1∗. Therefore

1≤ −cg′(k)g(k)−1γ =− c

1− 1γ(g(k)

1−1

γ)′. (3.13)

If we are in case i) of Theorem 3.1, we note that

1− 1

γ >0.

Therefore, by integrating (3.13) from 0 to k, we get

k ≤ −c[g(k)1−1γ −g(0)1−

1

(6)

i.e.

cg(k)1−γ1 ≤ −k+ckuβ−uβ

∗k

1−1γ

L1(Ω).

Sinceg(k) is a non-negative and decreasing function, from the latter inequality we deduce that there exists k0, such that g(k0)=0, and so uβ−uβ∗ ∈ L∞(Ω). In case ii) of Theorem 3.1, since m = n

p,γ = 1, we have

1≤ −cg

(x)

g(x). By integrating from 0 to k, we have

k

c ≤log[

kuβuβ

∗kL1(Ω)

g(k) ], and since the function t→et increases, we obtain

ekc ≤ ku

β uβ

∗kL1(Ω)

g(k) ⇒g(k)e

k

c ≤ kuβ −uβ

∗kL1(Ω).

So, recalling that

g(k) = Z

Ω|Gk(u β

−uβ)| ≥

Z

A2k

|Gk(uβ−uβ∗)| ≥k|A2k|, (3.14)

Hence, if k≥1, we have

g(k)≥ |A2k| ⇒ |A2k|e

k

c ≤ kuβ−uβ

∗kL1(Ω).

Hence, if k≥2, we get

|Ak|e

k

2c ≤ kuβ−uβ

∗kL1(Ω). (3.15)

We prove now that the previous inequality implies that

+X

k=0

ekτ|Ak|<∞,

with 0< τ < 21c. Indeed, by (3.15),

+∞ X

k=0

ekτ|Ak| ≤(1 +e)|Ω|+ +∞ X

k=2

kuβ uβ

∗kL1(Ω)

ek(21c−τ)

<∞.

Since

+∞ X

k=0

ekτ|Ak|<+∞ ⇒

Z

Ωe β|uβuβ

(7)

ii) is proved. To conclude, we consider case iii). In this case we have

1− 1

γ <0.

Therefore,

1≤c(g(k)1−γ1)′.

By integration from 0 to k, we obtain

k≤ c[g(k)1−γ1 −g(0)1−

1

γ]≤cg(k)1−

1

γ,

and so

g(k)−1+1γ ≤ c

k ⇒g(k)≤ c k1−γγ

.

Therefore, by (3.14), it holds true that

|A2k| ≤

g(k)

k ≤ c k1−γγk

= c

k1−1γ

.

By recalling the definition of γ, we obtain 1

γ =

mn(p−1)

n−mp =s,

so that uβ uβ

∗ ∈Ms(Ω). This ends the proof of Theorem 1.1.

References

[1] F.Leonetti, P.V.Petricca, Global summability for solutions to some anisotropic elliptic systems, Adv Diff.Eqns.,11-12(2015), 1165-1186.

[2] F.Leonetti,F.Siepe,Integrability for solutions to some anisotropic elliptic equations,NonlinearAnal.,75(2012),2867-2873.

[3] I.Alessandra,F.Leonetti,Global integrability for weak soiutions to some anisotropic equations,Nonlinear Anallysis,113(2015),430-434.

[4] P.Hartman,G.Stampacchia,On some non-linear elliptic differential func-tional equations, Acta Math.115(1966),271-310.

[5] O.Ladyzhenskaya,N.Ural’tseva Linear and quasilinear elliptic equa-tion,Math.in Science and Engineering,46,Academic Press(1968).

Referências

Documentos relacionados

Também pode ser usado para criar planos de orientação para os dentes pilares de próteses parciais removíveis e determinar o caminho de inserção de uma ponte Maryland prótese

Cotam-se nesta categoria todas as unidades de registo que remetam para a ilegitimação da violência letal quando o sujeito- alvo perpetrador de um crime é. equipado a uma potencial

Embora discordem em alguns pontos, verifica-se que nos pontos-chave de análise desta dissertação existe convergência de opiniões, nomeadamente na importância dos

Este projecto apresenta uma beneficiação entre Vimioso e Carção e Vinhos, portanto, tendo em conta as características do projecto optou-se por nível 1 com a classificação

Gráfico 1 – Relação das categorias estabelecidas das fronteiras lingüísticas com sua duração, a partir da leitura do texto oral “A história do urso preto”.. Analisando-se

Desejamos uma agradável leitura e discussão, esperando que este número da Revista EDaPECI possa contribuir para aprofundar o conhecimento e a reflexão crítica sobre a

É evidente que a participação de Osman Lins, mesmo que um tanto descentrada neste grupo, contribui enormemente na apropriação de um olhar sobre a arte da impressão do

Plasmodium infection. Regulation of iron homeostasis is a systemic response orchestrated in great part by hepcidin, and hemoglobinopathies have been linked to