Continuity of solutions to the G-Laplace equation involving measures
Yan Zhang
1and Jun Zheng
B11School of Mathematics, Southwest Jiaotong University, Chengdu 611756, China Received 7 January 2019, appeared 9 June 2019
Communicated by Dimitri Mugnai
Abstract. We establish local continuity of solutions to theG-Laplace equation involving measures, i.e.,
−div
g(|∇u|)
|∇u| ∇u
=µ,
where µ is a nonnegative Radon measure satisfying µ(Br(x0)) ≤ Crm for any ball Br(x0) ⊂⊂ Ω with r ≤ 1 and m > n−1−δ ≥ 0. The function g is supposed to be nonnegative andC1-continuous on[0,+∞), satisfyingg(0) =0 and
δ≤ tg
0(t)
g(t) ≤g0,∀t>0
with positive constants δ and g0, which generalizes the structural conditions of Ladyzhenskaya–Ural’tseva for an elliptic operator.
Keywords: G-Laplace, Radon measure, Orlicz space, Hölder continuity.
2010 Mathematics Subject Classification: 35J60, 35B65, 35J70, 35J75.
1 Introduction
Let Ω be an open bounded domain of Rn(n ≥ 2), and µ a nonnegative Radon measure in Ωwith µ(Br(x0)) ≤ Crm for some constant C > 0 whenever Br(x0) ⊂⊂ Ω. We consider the equation
−∆Gu =−div
g(|∇u|)
|∇u| ∇u
=µ inD0(Ω), (1.1)
where gis a nonnegativeC1-function on [0,+∞), satisfyingg(0) =0 and the following struc- tural condition
0<δ ≤ tg
0(t)
g(t) ≤ g0, ∀ t>0, δ,g0 are positive constants. (1.2)
BCorresponding author. Email: zhengjun@swjtu.edu.cn
The structural condition of gwas introduced by Tolksdorf in 1983 [14], which is a natural generalization of the natural conditions of Ladyzhenskaya and Ural’tseva for elliptic equations (see [10]). The conditions of g imply that the operator ∆G includes not only the p-Laplace operator∆p where g(t) = tp−1 andδ = g0 = p−1, but also the case of a variable exponent p= p(t)>0:
−∆Gu= −div(|∇u|p(|∇u|)−2∇u),
corresponding to setg(t) =tp(t)−1, for which (1.2) holds if δ≤t(lnt)p0(t) +p(t)−1 ≤g0for allt > 0. Another typical example of gis g(t) = tplog(at+b)with p,a,b > 0 where in this caseδ= p andg0= p+1. More examples can be found in [2,3,6,17] etc.
Let G(t) = Rt
0g(s)ds. Under assumption (1.2), G is an increasing, C2-continuous and convex function, which is an N-function satisfying ∆2-condition (see [1]). Thus our class of operators will be considered in the setting of Orlicz spaces. We recall the definitions of Orlicz and Orlicz–Sobolev spaces together with their respective norms (see [1])
LG(Ω) ={u∈L1(Ω); Z
ΩG(|u(x)|)dx <+∞}, kukLG(Ω)=inf
k>0;
Z
ΩG
|u(x)|
k
dx≤1
, W1,G(Ω) ={u∈LG(Ω);|∇u| ∈LG(Ω)},
kukW1,G(Ω)= kukLG(Ω)+k∇ukLG(Ω).
Under the assumption (1.2),W1,G(Ω)is a reflexive and separable Banach space (see [1]).
We shall call a solution of (1.1) any functionu∈Wloc1,G(Ω)that satisfies Z
Ω
g(|∇u|)
|∇u| ∇u· ∇ϕdx=
Z
Ωϕdµ ∀ϕ∈ D(Ω). Ifµ≡0 in a domainD⊂Ω, we say thatuisG-harmonic inD.
We now introduce regularities of related elliptic equations involving measures. In 1994, Kilpeläinen considered the situation of thep-Laplacian and proved that ifµsatisfiesµ(Br)≤ Crn−p+α(p−1) for some positive constantsC andα∈ (0, 1], then any solution to the p-Laplace equation
−∆pu=−div(|∇u|p−2∇u) =µ (1.3) isC0,βloc-continuous for each β ∈ (0,α)(see [7]). This result was improved by Kilpeläinen and Zhong in 2002, showing that every solution of (1.3) is in fact Hölder continuous with the same exponent α as the one in the assumption µ(Br) ≤ Crn−p+α(p−1) (see [8]). In 2010, the p-Laplace problem (1.3) was extended by Lyaghfouri to the case with variable exponents, i.e., considering
−div(|∇u|p(x)−2∇u) =µ. (1.4) Under certain assumptions on the functionp(x)and the assumptionµ(Br)≤Crn−p(x)+α(p(x)−1) for some positive constantsCandα∈ (0, 1], the author proved that any bounded solution of (1.4) isCloc0,α-continuous with the same exponentα(see [11]).
When focusing on the problem governed by G-Laplacian, if µ(Br(x0)) ≤ Crm with m ∈ [n−1,n), Challal and Lyaghfouri proved that any solution of (1.1) is C0,αloc-continuous with
α= m−1n++g1+δ
0 (see [3]). Particularly, ifm=n−1, any bounded solution isCloc0,α-continuous with anyα ∈ 0,gδ
0
(see Theorem 3.3 in [3]). In 2011, these regularities were improved by Challal and Lyaghfouri in [5], showing that any local bounded solution of (1.1) is Cloc0,α-continuous with any α ∈ 0,m−ng+1+δ
0
provided m > n−1−δ. Note that under the assumption of non-decreasing monotonicity on g(tt), Zheng, Feng and Zhang obtained local C1,α-continuity of solutions form> n and local Hölder continuity with a small exponent for somem< nin 2015 (see [15]).
In this paper, we continue the work of Challal, Lyaghfouri and Zheng et al. by improving the regularity of solutions of the equation (1.1). Particularly, we prove theC0,αloc-continuity of solutions with anyα∈(0, 1)ifm=n−1. More precisely, for anym>n−1−δ and without any monotonicity assumption on g(tt), we have the following results.
Theorem 1.1. Assume thatµsatisfies(1.1)with m>n−1−δ ≥0. For any local bounded solution u∈Wloc1,G(Ω)of (1.1), we have the following regularities:
(i) If m > n, then u ∈ C1,αloc(Ω)with anyα ∈ (0, min{1+σg
0,2(m1+−gn
0)}), where σ is the same as in Lemma2.5.
(ii) If m∈[n−1,n), then u∈ Cloc0,α(Ω)with anyα∈(0, 1).
(ii) If n−1−δ<m<n−1, then u∈ Cloc0,α(Ω)with anyα∈ (0,m−n+1+δ
δ ).
Remark 1.2. In [7], the author proved for thep-Laplacian problem thatu ∈C0,αloc(Ω)with any α∈(0, 1)providedm=n−1. In this paper we not only improve the results of [3,5] and [15], but also extend the problem in [7] to general equations governed by a large class of degenerate and singular elliptic operators.
Throughout this paper, without special states, byBR andBrwe denote the balls contained in Ωwith the same center. Moreover, Br ⊂⊂BR⊂⊂ΩandkukL∞(BR)≤ M for some constant M >0. (u)r = |B1
r|
R
Brudxbe the average value ofuon the ball Br.
2 Preliminary
In this section, we state some auxiliary results which will be used throughout this paper. We begin with some properties of the function G.
Lemma 2.1([13, Lemma 2.1, Remark 2.1]). The function G has the following properties:
(G1) G is convex and C2-continuous.
(G2) 1tg+(gt)
0 ≤ G(t)≤ tg(t), ∀t≥0.
(G3) min{sδ+1,sg0+1}1G+(tg)
0 ≤G(st)≤(1+g0)max{sδ+1,sg0+1}G(t), ∀s,t≥0.
(G4) G(a+b)≤2g0(1+g0)(G(a) +G(b)), ∀a,b≥0.
For much more properties of G and problems governed by the operator ∆G, please see [2–6,13,15,16,18,19] etc.
Lemma 2.2([9, Lemma 2.7]). Letφ(s)be a non-negative and non-decreasing function. Suppose that φ(r)≤C1r
R α
φ(R) +C1Rβ,
for all r≤ R≤ R0, with positive constantsα,βand C1. Then, for anyτ <min{α,β}, there exists a constant C2=C2(C1,α,β,τ)such that for any r ≤R≤R0we have
φ(r)≤C2rτ.
The following lemmas are some properties ofG-harmonic functions.
Lemma 2.3([13, Theorem 2.3]). Assume u∈Wloc1,G(Ω). Let h be a weak solution of
∆Gh=0 in BR, h−u ∈W01,G(BR), then
Z
BR
(G(|∇u|)−G(|∇h|))dx≥C
Z
A2G(|∇u− ∇h|)dx+
Z
A1
g(|∇u|)
|∇u| |∇u− ∇h|2dx
, where A1 = {x ∈ BR;|∇u− ∇h| ≤ 2|∇u|}, A2 = {x ∈ BR;|∇u− ∇h| > 2|∇u|}, and C = C(δ,g0)>0.
Lemma 2.4 ([13, Lemma 2.7]). Let h ∈ W1,G(BR) be a weak solution of ∆Gh = 0 in BR. Then h∈Cloc1,α(BR). Moreover, for everyλ∈(0,n), there exists C=C(λ,n,δ,g0)>0such that
Z
Br
G(|∇h|)dx≤Crλ, ∀r∈(0,R]. Proof. Indeed, we have (see [10, p. 345])
Z
Br
G(|∇h|)dx≤ C
r
R nZ
BR
G(|∇h|)dx
≤ C
r
R nZ
BR
G(|∇h|)dx+CRn, ∀r∈(0,R]. Then for anyλ∈(0,n), we obtain by Lemma2.3
Z
Br
G(|∇h|)dx ≤Crλ, ∀r ∈(0,R], which completes the proof.
Lemma 2.5(Comparison withG-harmonic functions [15, Lemma 3.1]). Assume u∈W1,G(BR). Let h ∈ W1,G(BR) be a weak solution of ∆Gh = 0 in BR. Then there exist σ ∈ (0, 1) and C = C(n,δ,g0)>0such that
Z
Br
G(|∇u−(∇u)r|)dx≤C
r
R n+σZ
BRG(|∇u−(∇u)R|)dx+C Z
BRG(|∇u−∇h|)dx, ∀r ∈(0,R]. Lemma 2.6. Assume u∈Wloc1,G(Ω). Let BR⊂⊂Ωand h∈W1,G(BR)be a weak solution of
∆Gh= 0 in BR, h−u∈W01,G(BR).
Then for anyλ∈(0,n), there exists C =C(λ,n,δ,g0,kukL∞(BR))>0such that Z
BR
G(|∇u− ∇h|)dx≤CRm+CRm+2λ.
Proof. Firstly, convexity ofGgives Z
BR
(G(|∇u|)−G(|∇h|))dx≤
Z
BR
g(|∇u|)
|∇u| ∇u(∇u− ∇h)dx
=
Z
BR
(u−h)dµ (2.1)
≤Cµ(BR)
≤CRm, (2.2)
where we used the boundedness ofuwhich forces hto be bounded too.
Let A1 and A2 be defined as in Lemma 2.3. By Lemma 2.3, there exists a constant C = C(δ,g0)>0 such that
Z
BR
(G(|∇u|)−G(|∇h|))dx≥C Z
A2G(|∇u− ∇h|)dx (2.3) and
Z
BR
(G(|∇u|)−G(|∇h|))dx≥C Z
A1
g(|∇u|)
|∇u| |∇u− ∇h|2dx. (2.4) By (G2), G(tt) is increasing int>0. It follows from (G2), (G3), (2.2), (2.3), (2.4), Lemma2.3and 2.4 that
Z
A1
G(|∇u−∇h|)dx=
Z
A1
G(|∇u−∇h|)
|∇u−∇h| (|∇u−∇h|)dx
≤
Z
A1
G(2|∇u|)
2|∇u| |∇u−∇h|dx
≤C Z
A1
G(|∇u|)
|∇u| |∇u−∇h|dx
≤C
Z
A1
G(|∇u|)
|∇u|2 |∇u−∇h|2dx 12Z
A1
G(|∇u|)dx 12
≤C
Z
A1
g(|∇u|)|∇u|
|∇u|2 |∇u−∇h|2dx 12Z
A1
G(|∇u|)dx 12
=C
Z
A1
g(|∇u|)
|∇u| |∇u−∇h|2dx 12Z
BRG(|∇u|)dx 12
≤C
Z
BR
(G(|∇u|)−G(|∇h|))dx 12Z
BR
G(|∇u|)dx 12
=C
Z
BR
(G(|∇u|)−G(|∇h|))dx 12
·
Z
BR
(G(|∇u|)−G(|∇h|) +G(|∇h|))dx 12
≤C Z
BR
(G(|∇u|)−G(|∇h|))dx +C
Z
BR
(G(|∇u|)−G(|∇h|))dx 12Z
BR
G(|∇h|)dx 12
,
≤CRm+CRm+2λ, (2.5)
where in the last inequality but one we used (a+b)γ ≤ aγ+bγ for any a ≥ 0,b ≥ 0 and γ∈(0, 1). By (2.2), (2.3) and (2.5), we have
Z
BR
G(|∇u−∇h|)dx=
Z
A2
G(|∇u−∇h|)dx+
Z
A1
G(|∇u−∇h|)dx
≤ C Z
BR
(G(|∇u|)−G(|∇h|))dx+CRm+CRm+2λ
≤ CRm+CRm+2λ.
3 Proof of Theorem 1.1
Proof of Theorem1.1. Lethbe aG-harmonic function inBRthat agrees withuon the boundary, i.e.,
div g(|∇h|)
|∇h| ∇h=0 in BR and h−u∈W01,G(BR). By Lemma2.5and Lemma 2.6, for anyr ≤Rthere holds
Z
Br
G(|∇u−(∇u)r|)dx≤ C
r
R n+σZ
BR
G(|∇u−(∇u)R|)dx+C Z
BR
G(|∇u− ∇h|)dx
≤ C
r
R n+σZ
BR
G(|∇u−(∇u)R|)dx+CRm+CRm+2λ, whereλis an arbitrary constant in(0,n).
(i) Ifm>n, then we have Z
Br
G(|∇u−(∇u)r|)dx ≤C
r
R n+σZ
BR
G(|∇u−(∇u)R|)dx+CRm+2λ.
Sincem> nandλis an arbitrary constant in(0,n), one may chooseλsatisfying m+2λ > n. In view of Lemma2.2, we conclude that for anyτ<min{σ,m+2λ−n}there holds
Z
Br
G(|∇u−(∇u)r|)dx≤Crn+τ, ∀r≤ R. (3.1) Now we claim that
Z
Br
|∇u−(∇u)r|dx≤ Crn+1+τg0, ∀r ≤R. (3.2) Indeed, forr satisfyingr−nR
Br|∇u−(∇u)r|dx ≤ r1+τg0, (3.2) holds with C = 1. Now for rsatisfyingr−nR
Br |∇u−(∇u)r|dx >r1+τg0, we infer from the increasing monotonicity of G(tt) int>0,
G r−nR
Br|∇u−(∇u)r|dx r−nR
Br|∇u−(∇u)r|dx ≥ G r
τ 1+g0
r1+τg0 .
It follows from (G2) and (G3) Z
Br
|∇u−(∇u)r|dx≤ r
n+1+τg
0
G r1+τg0G
r−n Z
Br
|∇u−(∇u)r|dx
≤ Cr
n+1+τg
0
rτG(1) G
r−n Z
Br
|∇u−(∇u)r|dx
≤ Cr
n+1+τg
0
rτg(1) G
r−n Z
Br
|∇u−(∇u)r|dx
. (3.3)
Note that convexity ofGand (3.1) imply that G
1
|Br|
Z
Br
|∇u−(∇u)r|dx
≤ 1
|Br|
Z
Br
G(|∇u−(∇u)r|)dx≤Crτ. (3.4) By (G3), (3.3) and (3.4), one may get
Z
Br
|∇u−(∇u)r|dx ≤Crn+1+τg0,
where Cdepends only ong(1),g0 and the volume of the unit ball. Now we have proven that (3.2) holds for anyr ≤ R. Thusu ∈ C1,
τ 1+g0
loc (Ω)by Campanato’s embedding theorem. Due to the arbitrariness ofλ∈(0,n),τ>0 can be arbitrary withτ<min{σ,m−2n}, which guarantees that Theorem1.1(i) holds true.
(ii) Ifm∈[n−1,n], we only prove form= n−1 due to the fact thatµ(Br)≤ Crm ≤Crn−1 with smallr. By (G4), Lemma2.4and Lemma2.6, we get
Z
Br
G(|∇u|)dx≤ C Z
Br
G(|∇u− ∇h|)dx+C Z
Br
G(|∇h|)dx
≤ Crm+Crm+2λ +Crλ
≤ Crm,
where in the last inequality we letn> λ> n−1=m.
We claim that for anyr≤ R<1 withBR ⊂⊂Ωand some positive constantCindependent ofr, there holds
Z
Br
|∇u|dx ≤Crn−1+α0, (3.5)
with someα0 ∈(0, 1).
Indeed, forr≤ Rsatisfying
r−n+1−α0 Z
Br
|∇u|dx≤1, (3.6)
(3.5) holds withC =1. Forr≤ Rsatisfying r−n+1−α0
Z
Br
|∇u|dx≥1,
due to the increasing monotonicity of F(t) =G(t)−G(1)t int≥1, it follows G
r−n+1−α0 Z
Br
|∇u|dx
≥ G(1)·r−n+1−α0 Z
Br
|∇u|dx.
Then we have
Z
Br
|∇u|dx≤Crn−1+α0(r1−α0)1+δG
r−n Z
Br
|∇u|dx
≤Crn−1+α0·(r1−α0)1+δ 1
|Br|
Z
Br
G(|∇u|)dx
≤Crn−1+α0+(1−α0)(1+δ)·r−n·rm
=Crn−1+α0+(1−α0)(1+δ)+m−n. (3.7) Combining (3.6) and (3.7), we may chooseα0 =α0+ (1−α0)(1+δ) +m−n, i.e.,α0 =1−n1−+m
δ
such that (3.5) holds for allr ≤R.
Form=n−1, we conclude thatu∈ Cloc0,α0(Ω)by Morrey Theorem (see page 30, [12]) with α0= 1+δ
δ.
Note that infBru ≤ infBrh (see the proof of Theorem 3.3 in [3]). Then by (2.1) and Lemma2.4, forλlarger thanm+α0, we have
Z
Br
G(|∇u|)dx≤
Z
Br
(u−h)dµ+
Z
Br
G(|∇h|)dx
≤(sup
Br
u−inf
Br
h)µ(Br) +
Z
Br
G(|∇h|)dx
≤(sup
Br
u−inf
Br
u)µ(Br) +
Z
Br
G(|∇h|)dx
≤Cosc(u,Br)rm+Crλ
≤Crα0+m+Crλ
≤Crm+α0, where osc(u,Br) =supB
ru−infBru. Arguing as (3.5), we getu∈ Cloc0,α1(Ω)with α1 =1− n−(m+α0)
1+δ = δ
1+δ + α0 1+δ. Repeating this process, we getu∈C0,αlock(Ω)with
αk = δ
1+δ + αk−1 1+δ. Finally, we haveαk = α0
(1+δ)k +δ∑kj=1 1
(1+δ)j, which leads to limk→∞αk=1, and the result follows.
(iii) If n−1−δ < m < n−1, checking the proof and repeating the process as above, we may get α0 = 1− n1−+m
δ, α1 = 1+δ1++m−n
δ + 1α+0
δ, . . . , αk = 1+δ1++m−n
δ + α1k+−1
δ. Finally, one has u∈C0,αloc(Ω)for anyα∈(0,1+δ+δm−n).
References
[1] R. A. Adams, J. J. F. Fournier, Sobolev spaces, Pure and Applied Mathematics (Amster- dam), Vol. 140, Elsevier/Academic Press, Amsterdam, 2003.MR2424078
[2] J. E. M. Braga, D. R. Moreira, Uniform Lipschitz regularity for classes of minimizers in two phase free boundary problems in Orlicz spaces with small density on the negative phase, Ann. Inst. H. Poincaré, Anal. Non Linéaire.31(2014), No. 4, 823–850.https://doi.
org/10.1016/j.anihpc.2013.07.006;MR3249814;Zbl 1301.49097
[3] S. Challal, A. Lyaghfouri, Hölder continuity of solutions to the A-Laplace equation involving measures, Common. Pure Appl. Anal. 8(2009), No. 5, 1577–1583. https://doi.
org/10.3934/cpaa.2009.8.1577;MR2505287;Zbl 1179.35336
[4] S. Challal, A. Lyaghfouri, Porosity of free boundaries in A-obstacle problems, Non- linear Anal. 70(2009), No. 7, 2772–2778. https://doi.org/10.1016/j.na.2008.04.002;
MR2499745;Zbl 1166.35385
[5] S. Challal, A. Lyaghfouri, Removable sets for A-harmonic functions, Z. Anal.
Anwend. 30(2011), No. 4, 421–433. https://doi.org/10.4171/ZAA/1442; MR2853964;
Zbl 1260.35046
[6] S. Challal, A. Lyaghfouri, J. F. Rodrigues, On the A-obstacle problem and the Haus- dorff measure of its free boundary, Ann. Mat. Pura Appl. (4) 191(2012), No. 1, 113–165.
https://doi.org/10.1007/s10231-010-0177-7;MR2886164;Zbl 1235.35285
[7] T. Kilpeläinen, Hölder continuity of solutions to quasilinear elliptic equations involving measures,Potential Anal.3(1994), No. 3, 265–272.https://doi.org/10.1007/bf01468246;
MR1290667;Zbl 0813.35016
[8] T. Kilpeläinen, X. Zhong, Removable set for continuous solutions of quasilinear elliptic equations,Proc. Amer. Math. Soc.130(2002), No. 6, 1681–1688.https://doi.org/10.2307/
2699762;MR1887015
[9] R. Leitão, O. S. de Queiroz, E. V. Teixeira, Regularity for degenerate two-phase free boundary problems, Ann. Inst. H. Poincaré Anal. Non Linéaire 32(2015), No. 4, 741–762.
https://doi.org/10.1016/j.anihpc.2014.03.004;MR3390082
[10] G. M. Lieberman, The natural generalization of the natural conditions of Ladyzhensaya and Ural’tseva for elliptic equations,Comm. Partial Differ. Equ.16(1991), No. 2–3, 311–361.
https://doi.org/10.1080/03605309108820761;MR1104103
[11] A. Lyaghfouri, Hölder continuity of p(x)-superharmonic functions, Nonlinear Anal.
73(2010), No. 8, 2433–2444. https://doi.org/10.1016/j.na.2010.06.016; MR2674081;
Zbl 1194.35482
[12] J. Malý, W. P. Ziemer, Fine regularity of solutions of elliptic partial differential equations, Mathematical Surveys and Monographs, Vol. 51, Providence (RI): Amer. Math. Soc., 1997.
https://doi.org/10.1090/surv/051;MR1461542
[13] S. Martínez, N. Wolanski, A minimum problem with free boundary in Orlicz spaces, Adv. Math.218(2008), No. 6, 1914–1971. https://doi.org/10.1016/j.aim.2008.03.028;
MR2431665;Zbl 1170.35030
[14] P. Tolksdorf, On the Dirichlet problem for quasilinear equations in domains with conical boundary points,Comm. Partial Differ. Equ.8(1983), No. 7, 773–817.https://doi.org/10.
1080/03605308308820285;MR0700735
[15] J. Zheng, B. Feng, Z. Zhang, Regularity of solutions to theG-Laplace equation involving measures, Z. Anal. Anwend. 34(2015), No. 2, 165–174. https://doi.org/10.4171/ZAA/
1534;MR3336258;Zbl 1323.35191
[16] J. Zheng, B. Feng, P. Zhao, Regularity of minimizers in the two-phase free boundary problems in Orlicz–Sobolev spaces, Z. Anal. Anwend. 36(2017), No. 1, 37–47. https://
doi.org/10.4171/ZAA/1578;MR3638967;Zbl 1359.35066
[17] J. Zheng, X. Guo, Lyapunov-type inequalities for ψ-Laplacian equations, chi- naXiv:201805.00171, 2018.https://doi.org/10.12074/201805.00171
[18] J. Zheng, L. S. Tavares, C. O. Alves, A minimum problem with free boundary and subcritical growth in Orlicz spaces, preprint published onarXiv:1809.08518v2, 2018.
[19] J. Zheng, Z. Zhang, P. Zhao, A minimum problem with two-phase free boundary in Orlicz spaces, Monatsh. Math. 172(2013), No. 3–4, 441–475. https://doi.org/10.1007/
s00605-013-0557-3;MR3128005;Zbl 1285.35135