• No results found

Global Existence and Extinction of Weak Solutions to a Class of Semiconductor Equations with Fast Diffusion Terms

N/A
N/A
Protected

Academic year: 2020

Share "Global Existence and Extinction of Weak Solutions to a Class of Semiconductor Equations with Fast Diffusion Terms"

Copied!
14
0
0

Loading.... (view fulltext now)

Full text

(1)

Volume 2008, Article ID 961045,14pages doi:10.1155/2008/961045

Research Article

Global Existence and Extinction of

Weak Solutions to a Class of Semiconductor

Equations with Fast Diffusion Terms

Bin Wu

Department of Information and Computing Sciences, College of Mathematics and Physics, Nanjing University of Information Science and Technology, Nanjing 210044, China

Correspondence should be addressed to Bin Wu,[email protected]

Received 2 March 2008; Accepted 13 August 2008

Recommended by Y. Giga

We consider the transient drift-diffusion model with fast diffusion terms. This problem is not only degenerate but also singular. We first present existence result for general nonlinear diffusivities for the Dirichlet-Neumann mixed boundary value problem. Then, the extinction phenomenon of weak solutions for the homogeneous Dirichlet boundary problem is studied. Sufficient conditions on the extinction and decay estimates of solutions are obtained by usingLp-integral model estimate method.

Copyrightq2008 Bin Wu. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

The scaled semiconductor drift-diffusion model reads

−∇·∇ψ pnCx, 1.1

nt− ∇ ·Jn rn, p1−np g, Jnnmnψ, 1.2

pt∇ ·Jp rn, p1−np g,Jppmpψ, 1.3

(2)

ifm 1exponents. The regimem > 1 describes a slow diffusion process in the electron

holedensity, whereas 0< m <1 is related to fast diffusion.

We supplement these equations with physically motivated boundary conditions 1:

ψ, n, p ψD, nD, pD, x, t∈ΣD≡ΓD×0, T, 1.4

∂ψ

∂η, ∂n ∂η,

∂p ∂η

0,0,0, x, t∈ΣN≡ΓN×0, T, 1.5

n, p n0, p0

, x∈Ω, t0, 1.6

where Ω splits into two disjoint subsets ΓN and ΓD.ΓN models the union of insulating

boundary segments andΓDthe union of ohmic contacts.

The standard drift-diffusion model corresponding tom 1 has been mathematically and numerically investigated in many paperssee 1–4. Existence and uniqueness of weak solutions have been shown. Recently, the existence analysis of the bipolar drift-diffusion problem in the adiabatic case has been studied by many authors 5–9. But, as far as we know, few works are concerned with the mixed boundary value problem of the drift-diffusion model with the fast diffusion terms.

The phenomenon of extinction is an important property of solutions for many evolutionary equations which have been studied extensively by many authors; see, for example, 10,11. Especially, there are some papers concerning the extinction for fast diffusive equations. For instance, in 12, the extinction and positivity for the evolutionaryp-Laplacian equation without sources were studied by some authors, and Li and Wu 13investigated the extinction for a fast diffusive filtration equation with nonlinear source terms. But, most of the work about extinction results is limited to single equation.

We prove first the global existence of weak solutions to problem1.1–1.6in the sense of the definition below. Wheng ≡ 0 and the boundary condition is homogeneous and of Dirichlet type, we show that the solutionsnx, t, px, tvanish in finite time in the following two cases.

iThe initial data are appropriately small.

iiThe doping profile is sufficiently small and generation rate grows very slowly even if the initial data are very large.

Simultaneously, the decay estimates of solutions are obtained. For the proof of our result, we employLp-integral model estimate method and a crucial lemma on differential inequality. This technique has been successfully applied to the porous medium equation 14.

We make the following assumptions.

H1 Ω⊂RN N1,2,3is bounded andΩ∈C0,1, whose outward normal vector isη.

H2Cx, gxL∞Ωandgx≥0 for a.e.x∈Ω.

H3rn, pis a locally Lipschitz continuous function defined forn, pand 0≤rn, p

r <∞.

H4nmD, pmD, ψDH1Ω∩L∞Ω, andnD, pD≥0 inΩ.

(3)

Remark 1.1. It is obvious thatH4implies thatnD, pDH1ΩLΩ.

Let

Y ωH1Ω|ω|

ΓD 0

1.7

and letY∗denote the dual space ofY.

Definition 1.2. ψ, n, p is called the weak solution to problem 1.1–1.6 if nnD L20, T;Y, ppDL20, T;Y, ψ ψDL20, T;Y, nt, ptL20, T;Y, n|t

0n0, p|t0 p0, and there hold

Ω∇ψ· ∇φdx

ΩpnCφdx,t∈0, T,φY,

nt, v

T

0

Ω

nmnψ· ∇vdxdt

T

0

Ω rn, p1−np gvdxdt,vL

20, T;Y,

pt, v

T

0

Ω

pmpψ· ∇vdxdt

T

0

Ω

rn, p1−np gvdxdt,vL20, T;Y.

1.8

2. Existence

This section is devoted to the proof of global existence of weak solutions to problem1.1–

1.6. We will prove the following existence theorem.

Theorem 2.1. Under hypotheses (H1)–(H4), there exists at least one weak solution of problem1.1

1.6.

The following lemma will be used in the proof of existence result.

Lemma 2.2 see 9. Let (H1) hold,α, ν, FLQT, and F, ν ≥ 0 for a.e.x, tQT, θ0 ∈ L

Ω, θH1Ω∩L∞Ω, γL2QTN, γ·η0, divγiLQTweakly, andαε >

0. Then, there exists a unique solutionθof the following problem:

θt− ∇ ·αθγθ νθF, x, tQT,

θD θ,

∂θ ∂η

ΣN

0, θ0 θ0, x∈Ω,

2.1

such thatθL20, T;Y, θtL20, T;Y

0, and

0≤θleλt a.e.x, tQT, 2.2

(4)

Our main difficulty in the proof is that problem 1.1–1.6 is degenerate at points wheren, p ∞, and is singular at points wheren, p0. This difficulty leads us to consider the following auxiliary regularized problem with the initial and boundary conditions1.4–

1.6:

−∇ ·∇ψ pknkCx, 2.3

nt− ∇ ·

m

1

nkτ

1−m

nnψ

rnk, pk1−npkg, 2.4

pt− ∇ ·

m

1

pkτ

1−m

ppψ

rnk, pk1−nkpg, 2.5

where 0< τ <1 andskmin{k,max{0, s}}.

Now, we use the Schauder fixed point theorem 15to prove the existence of weak solution to the above regularized problem.

Lemma 2.3. Under hypotheses (H1)–(H4), there exists at least one weak solution of problem2.3

2.5, under conditions1.41.6.

Proof. Define the setκas

κxL2QT|x0 a.e. xQT,x

L2QTR

. 2.6

It is obvious thatκis a closed convex set and weakly compact inL2QT. Givenn, pκ, we

consider the following problems:

−∇ ·∇ψ pknk Cx, 2.7

ψD ψD,

∂ψ ∂η

ΣN

0, 2.8

nt− ∇ ·

m 1

nkτ

1−m

nnψ

rnk,pkpknrnk,pkg, 2.9

nD nD,

∂n ∂η

ΣN

0, n0 n0, x∈Ω, 2.10

pt− ∇ ·

m 1

pkτ

1−m

ppψ

rnk,pknkprnk,pkg, 2.11

pD pD,

∂p ∂η

ΣN

0, p0 p0, x∈Ω. 2.12

We deduce the existence of a unique weak solution of 2.7 and 2.8 with the regularity

(5)

ν rnk, pk pk , andF rnk, pk g.Lemma 2.2ensures the existence and uniqueness of weak solution of problem2.9-2.10with the regularitynL20, T;H1ΩLQT, where

Lestimate ofnis dependent onk, τ and the known data, but independent ofR by2.2. Similar results hold for problem2.11-2.12.

Thus, the map

S:κ2−→L2QT2, n, p−→n, p 2.13

is well defined and compact. For the above given R, Sκ2 κ2 holds if we choose R

sufficiently large. By using the standard methodfor details, see 5or 9, we can obtain the continuity of mapS. Thus, existence of a fixed point ofS follows the weak solution of problem2.3–2.5, under conditions1.4–1.6.

Lemma 2.4. The weak solutions of problem2.32.5, under conditions1.41.6, satisfy the estimates

0≤nx, t, px, tk, a.e.x, tQT, 2.14

if one choosesksuch that

kM:max

sup

Ω

n0, p0

,sup

ΣD

nD, pDergL∞ΩCxL∞ΩT. 2.15

Proof. SetNneβtandP peβt; hereβ >0 is to be determined. Then,N, Psatisfies

NtβN− ∇ ·

m

1

eβtN

1−m

NNψ

eβtrn

k, pk1−npkg,

2.16

PtβP− ∇ ·

m

1

eβtP

1−m

PPψ

eβtrnk, pk1nkpg, x, tQT,

2.17

eβtN, eβtPnD, pD, x, tΣD, 2.18

∂N

∂η, ∂P ∂η

0,0, x, t∈ΣN, 2.19

N, P n0, p0

, x∈Ω, t0. 2.20

To obtain the upper bound, we compareNandPwithzt:k0βt, where

k0≥max

sup

Ω

n0, p0

,sup

ΣD

(6)

and μ such that μ > β is a large constant which can be determined by the following calculations. Then, we could takeNzandPzas test functions in2.16and2.17, respectively, and obtain

1 2

Ω

Nz2

t Pz2

t

0

Ω

Nz2

Pz2

t 0 Ωm 1

eβtN

1−m

Nz2t 0

Ωm

1

eβtP

1−m

Pz2

≤ −μ

t

0

Ωz

Nz Pz

t

0

Ω

N· ∇NzP· ∇Pz· ∇ψ

t

0

Ωe

βtrn

k, pkgNz Pz

≤ −μ

t

0

Ωz

Nz PN1

2 t 0 Ω

pknkCxNz2

Pz2

t 0 Ω

rgzNzPNt

0

Ωz

pknkCxNz Pz

≤−μrgLΩCxLΩ

t

0

Ωz

Nz Pz

1

2CxL∞Ω t

0

Ω

Nz2

Pz2

,

2.22

where in the last inequality we have used

pknkNzPzeβtpknknk0eμt

pk

0eμt

0. 2.23

Choosingβ 1/2CxLΩandμrgLΩCxLΩ, we get

1 2

Ω

Nz2

t Pz2

tm

1

1−mt

0

Ω

Nz2Pz20.

2.24

(7)

Thus, we obtain a weak solution of the following problem:

−∇ ·∇ψ pnCx, 2.25

nt− ∇ ·

m

1

1−m

nnψ

rn, p1−np g, 2.26

pt− ∇ ·

m

1

1−m

ppψ

rn, p1−np g, 2.27

with the initial and boundary conditions1.4–1.6.

In what follows, we give some estimates of the weak solutions to the above problem uniformly inτ, which are necessary in the proof ofTheorem 2.1.

Lemma 2.5. The solutions of problem 2.252.27, under conditions 1.41.6, satisfy the estimates m2 T 0 Ω 1

2−2m

n2 1

2−2m

p2

C, 2.28

whereCis independent ofτ.

Proof. First of all, a standard elliptic estimate gives

ψL2QTC

1nL2QTpL2QT

C. 2.29

Let

ϕn, nDm

n

nD

1

1−m

ds. 2.30

Sinceϕn, nD|ΓD 0 and T

0

Ω

ϕ

n, nD2

2m2 T 0 Ω 1

2−2m

n2 1

nDτ

2−2m

nD2

≤2m2

1

τ

2−2m

n2

L20,T;H1ΩnD

2

L20,T;H1Ω

Cτ,

2.31

we haveϕn, nDL20, T;Y. Then,

m2 T 0 Ω 1

2−2m

|∇n|2m2

T 0 Ω 1

1−m

n·

1

nDτ

1−m

nDT 0 Ωntϕ n, nD T 0

Ωnψ· ∇ϕ

n, nD T 0 Ω

rn, p1−np gϕn, nD

I1I2I3I4.

(8)

By Young’s inequality and2.29, we have

I1≤ m2 4 0 Ω 1

2−2m

|∇n|2

T 0 Ω nm

D2, 2.33

I3≤

1 8 T 0 Ω

ϕn, nD24

T

0

Ω|

nψ|2

m2 4 T 0 Ω 1

2−2m

|∇n|2 m2

4 T 0 Ω nm

D24M2∇ψ2L2QT

m2 4 T 0 Ω 1

2−2m

|∇n|2C.

2.34

Furthermore,

I4mr1M2gLΩ

T 0 Ω n 0 1

1−m

ds nD

0

1

1−m

ds

≤2mr1M2gLΩQT M

0

1

1−m

dsC.

2.35

For the termI2, we see that

I2≤

T 0 Ωnϕt n, nD

ΩnTϕ

n, nDT

ΩnTϕ

n, nD0

m T 0 Ω nnt 1

1−m

4M|Ω| M

0

1

1−m

dsm T 0 Ω ∂t n 0

s 1

1−m

ds Cm Ω

nx,T

n0x

s

1

1−m

ds

CC.

2.36

Inserting2.33–2.36into2.32, we finally get

m2 T 0 Ω 1

2−2m

|∇n|2C. 2.37

A similar estimate holds forp, and then we complete the proof.

Lemma 2.6. The weak solutions nand p of problem 2.252.27, under conditions1.41.6, satisfy the following estimate:

(9)

Proof. We will only prove that2.38holds forn. The rest can be derived similarly. Without loss of generality, we assume 0< τ <1. Then, we conclude that

T

0

Ω|∇

n|2

T

0

Ω

2−2m 1

1−m

|∇n| 2

M12−2m T

0

Ω

1

1−m

|∇n| 2

C.

2.39

Following Lemmas2.5and2.6, we can easily get the following lemma.

Lemma 2.7. The weak solutions nand p of problem 2.252.27, under conditions1.41.6, satisfy the following estimate:

ntL2

0,T;Y∗, ptL20,T;YC. 2.40

Proof ofTheorem 2.1. Let {ψτ, nτ, pτ} be the sequence of solutions of problem 2.25–2.27, under conditions 1.4–1.6. By passing to a subsequence, if necessary, from Lemmas 2.4– 2.7, we infer that

t,pεt−→nt, ptweakly inL20, T;Y,

ψε, nε, pε−→ψ, n, pstrongly inL2QT, a.e.inQT,

ψε, nε, pε−→ψ, n, pweakly inL20, T;H,

nτm,pτm−→∇nm,pmweakly inL2QT.

2.41

Now, we can conclude thatψ, n, p is a weak solution of problem1.1–1.6by standard method, and then complete the proof ofTheorem 2.1.

3. Extinction

Besides the assumptionsH1–H5, we need for our extinction result the following structural condition onrn, p.

H6rn, pρ0npfor some positive constantρ0.

Becausern, pdenotes the generation term in the recombination-generation rate, this condition means that the generation rate grows very slowly inn, pwhenρ0is very small.

Additionally, we replace boundary conditions1.4-1.5with homogeneous Dirichlet conditions

npψ0, xΩ×0, T. 3.1

It is obvious that the existence of weak solutions of problem1.1–1.3, under conditions

(10)

Before stating our main result in this section, we list the following important condi-tions inTheorem 3.1:

CxL

Ω2ρ0

1 2C

−2

0 |Ω|−2m/m1N−2/N

×n0mLm11Ωp0mLm11Ω

m−1/m1

,

3.2

n0m1

Lm1Ωp0

m1 Lm1Ω

C−2

0 |Ω|−2m/m1N−2/N

CxLΩ2ρ0

1m/1−m

, 3.3

CxLΩ2ρ0≤

2ms

ms2C

−2 0

n0sLs11Ωp0sLs11Ω

m−1/s1

, 3.4

n0s1

Ls1Ωp0

s1 Ls1Ω

4msC02

ms2Cx

LΩ2ρ0

1s/1−m

, 3.5

wheres N1−m/2−1 andC0is the embedding constant of the following inequality: uL2N/N−2ΩC0uL2Ω,uH0. 3.6 Theorem 3.1. Let (H1)–(H6) be fulfilled, letgx≡0, and letψ, n, pbe a weak solution of problem

1.11.3, under conditions3.1and1.6. Then, one has the following.

iIf 3.2or 3.3below is satisfied, then ux, tand px, tvanish in the finite time for

N−2/N2≤m <1, and n·, tm1

Lm1Ωp·, t

m1

Lm1ΩB1eα1t, t

0, T01

,

n·, tm1

Lm1Ωp·, t

m1 Lm1Ω

n·, T01Lmm11Ωp

·, T01mLm11Ω

1−m/1m

C11−m

1m t

1m/1−m

,

tT01, T1

,

n·, tm1

Lm1Ωp·, t

m1

Lm1Ω≡0, t

T1,

.

3.7

iiIf 3.4or 3.5below is satisfied, then ux, tand px, tvanish in the finite time for 0< m <N−2/N2, and

n·, ts1

Ls1Ωp·, t

s1

Ls1ΩB2eα2t, t

0, T02, n·, ts1

Ls1Ωp·, t

s1

Ls1Ωn·, t

s1

Ls1Ωp·, t

s1 Ls1Ω

n·, T02s1 Ls1Ωp

·, T02s1 Ls1Ω

1−m/1s

C21−m 1s t

1s/1−m

,

tT02, T2

,

n·, ts1

Ls1Ωp·, t

s1

Ls1Ω≡0, t

T2,

.

3.8

(11)

To obtain the above result, we will use the following lemma which is of crucial impor-tance to the proof.

Lemma 3.2. Let0< k < q≤1, and letyt≥0be a solution of the differential inequality

dy dt βy

kσyq t0, y0 y

0≥0, 3.9

whereβandσare positive constants. If

iy0< β

σ

1/qk

or (iiσβy

kq 0

2 3.10

holds, then there existα >0andB >0such that

0≤ytBeαt, t≥0. 3.11

Proof. We will use upper and lower solutions’ methods to complete the proof. The proof of caseiis basically the same as method of caseii. So, we are only devoted to the proof of casei. In fact, we only need to chooseα, Bproperly such thatBeαtis an upper solution of

3.9. That is,αandBmust satisfy

αBeαtβBkeαktσBqeαqt t0, By

0. 3.12

ChooseBsuch that lnB <lnβ/σ/qkandαsufficiently small. Then,

σ βBqk

α

βB1−k≤1, 3.13

which implies that

αqkt≥ln

σ βBqk

α βB1−k

,t≥0. 3.14

Noting that 0< q≤1, therefore we have

eαkqt σ βBqk

α

βB1−ke−1−qαt,t≥0, 3.15

which is equivalent to3.12ify0B <β/σ1/qk. The lemma is proved.

Proof ofTheorem 3.1. We consider first the caseN−2/N2≤m <1. Multiplying1.2and

1.3bynmandpm, respectively, and employing the Poisson equation1.1, we get

1

m1

d dt

Ω

nm1pm1dx

Ω

nm2pm2

dx

m

m1

Ω

pnCxnm1pm1dx

Ωrn, p1−np

nmpmdx

I1I2.

(12)

The embedding theorem gives

nm

Lm1Ω≤ |Ω|m/m1−N−2/2NnmL2N/N−2Ω

C0|Ω|m/m1−N−2/2N

nm

L2Ω,

3.17

and similarly forp.

By Young’s inequality, we have

I2≤ρ0

Ωnp

nmpmdx2ρ 0

Ω

nm1pm1dx. 3.18

By3.16–3.18, taking into accountpnnm1pm10, we obtain

d

dtf1t m1C −2

0 |Ω|−2m/m1N−2/Nf1t

2m/m1m1Cx

LΩ2ρ0

f1t,

3.19

where

f1t nmLm11ΩpmLm11Ω. 3.20

ByLemma 3.2, there existα1andB1such that

0≤f1tB1eα1t, t≥0, 3.21

if3.2or3.3holds.

Furthermore, there existT01such that

1mC−2

0 |Ω|−2m/m1N−2/NCxLΩ2ρ0

f1t1−m/1m

≥1mC0−2|Ω|−2m/m1N−2/NCxLΩ2ρ0

B1eα1T01

1−m/1m

:C1 >0

3.22

holds fortT01,∞. Thus, whentT01,∞,3.19turns to

d

dtf1t C1f1t

2m/m10, tT

01. 3.23

By a direct calculation, we deduce that

f1t

f1

T01

1−m/1m C11−m

1m t

1m/1−m

, tT01, T1

,

f1t≡0, t

T1,

,

3.24

where

T1

1m

C11−m

n·, T01mLm11Ωp

·, T01mLm11Ω

1−m/1m

(13)

For the case 0< m < N−2/N2, usingns andpsas test functions in1.2and

1.3, respectively, we obtain

1

s1

d dt

Ω

ns1ps1dx 4ms

ms2

Ω

nms/22pms/22

dx

ss

1

Ω

pnCxns1ps1dx

Ωrn, p1−np

nspsdx.

3.26

To simplify, we denote

f2t ns1

Ls1ΩpLss11Ω. 3.27

Hence, we have

d dtf2t

4s1ms

ms2 C− 2 0 f2t

ms/1ss1Cx

LΩ2ρ0

f2t, 3.28

where we have used

uLms1Ωs/2

Ωu

ms/2·2N/N−2dx

N−2/2N

C0∇

ums/2

L2Ω,uH0.

3.29

ByLemma 3.2, there existα2andB2such that

0≤f2tB2eα2t, t≥0, 3.30

if3.4or3.5holds.

By a simple analysis similar to that of the caseN−2/N2≤m <1, we obtain

f2t

f2

T02

1−m/1s C21−m

1s t

1s/1−m

, tT02, T2

,

f2t≡0, t

T2,

,

3.31

where

C2 s1

4ms

ms2C− 2

0 −CxLΩ2ρ0

B2eα2T021−m/1s

>0, 3.32

T2

1s

C21−m

n·, T02sLs11Ωp

·, T02sLs11Ω

1−m/1s

. 3.33

References

1 P. A. Markowich, C. A. Ringhofer, and C. Schmeiser, Semiconductor Equations, Springer, Vienna, Austria, 1990.

(14)

3 W. F. Fang and K. Ito, “On the time-dependent drift-diffusion model for semiconductors,”Journal of Differential Equations, vol. 117, no. 2, pp. 245–280, 1995.

4 W. F. Fang and K. Ito, “Global solutions of the time-dependent drift-diffusion semiconductor equations,”Journal of Differential Equations, vol. 123, no. 2, pp. 523–566, 1995.

5 A. J ¨ungel, “On the existence and uniqueness of transient solutions of a degenerate nonlinear drift-diffusion model for semiconductors,”Mathematical Models & Methods in Applied Sciences, vol. 4, no. 5, pp. 677–703, 1994.

6 A. J ¨ungel, “Qualitative behavior of solutions of a degenerate nonlinear drift-diffusion model for semiconductors,”Mathematical Models & Methods in Applied Sciences, vol. 5, no. 4, pp. 497–518, 1995. 7 J. I. D´ıaz, G. Galiano, and A. J ¨ungel, “On a quasilinear degenerate system arising in semiconductors

theory—part I: existence and uniqueness of solutions,”Nonlinear Analysis: Real World Applications, vol. 2, no. 3, pp. 305–336, 2001.

8 J. I. D´ıaz, G. Galiano, and A. J ¨ungel, “On a quasilinear degenerate system arising in semiconductor theory—part II: localization of vacuum solutions,”Nonlinear Analysis: Theory, Methods & Applications, vol. 36, no. 5, pp. 569–594, 1999.

9 P. Guan and B. Wu, “Existence of weak solutions to a degenerate time-dependent semiconductor equations with temperature effect,”Journal of Mathematical Analysis and Applications, vol. 332, no. 1, pp. 367–380, 2007.

10 S. N. Antontsev and S. I. Shmarev, “A model porous medium equation with variable exponent of nonlinearity: existence, uniqueness and localization properties of solutions,”Nonlinear Analysis: Theory, Methods & Applications, vol. 60, no. 3, pp. 515–545, 2005.

11 R. Ferreira and J. L. Vazquez, “Extinction behaviour for fast diffusion equations with absorption,”

Nonlinear Analysis: Theory, Methods & Applications, vol. 43, no. 8, pp. 943–985, 2001.

12 H. Yuan, S. Lian, W. Gao, X. Xu, and C. Cao, “Extinction and positivity for the evolutionp-Laplacian equation in Rn,”Nonlinear Analysis: Theory, Methods & Applications, vol. 60, no. 6, pp. 1085–1091, 2005. 13 Y. Li and J. Wu, “Extinction for fast diffusion equations with nonlinear sources,”Electronic Journal of

Differential Equations, vol. 2005, no. 23, pp. 1–7, 2005.

14 W. Liu, M. Wang, and B. Wu, “Extinction and decay estimates of solutions for a class of porous medium equations,”Journal of Inequalities and Applications, vol. 2007, Article ID 87650, 8 pages, 2007. 15 O. A. Ladyˇzhenskaya and N. N. Vral’tseva,Linear and Quasilinear Elliptic Equations, Academic Press,

References

Related documents