We extend the so-called universal potential estimates of Kuusi-Mingione type (J. Funct. Anal. 262: 4205–4269, 2012) to the singular case 1<p≤2−1/n for the quasilinear equation with measure data
−div(A(x,∇u))=μ
in a bounded open subset Ω of Rn, n≥2, with a finite signed measure μ in Ω. The operator div(A(x,∇u)) is modeled after the p-Laplacian Δpu:=div(|∇u|p−2∇u), where the nonlinearity A(x,ξ) (x,ξ∈Rn) is assumed to satisfy natural growth and monotonicity conditions of order p, as well as certain additional regularity conditions in the x-variable.
Citation: Quoc-Hung Nguyen, Nguyen Cong Phuc. Universal potential estimates for 1<p≤2−1n[J]. Mathematics in Engineering, 2023, 5(3): 1-24. doi: 10.3934/mine.2023057
[1] | María Ángeles García-Ferrero, Angkana Rüland . Strong unique continuation for the higher order fractional Laplacian. Mathematics in Engineering, 2019, 1(4): 715-774. doi: 10.3934/mine.2019.4.715 |
[2] | Rolando Magnanini, Giorgio Poggesi . Interpolating estimates with applications to some quantitative symmetry results. Mathematics in Engineering, 2023, 5(1): 1-21. doi: 10.3934/mine.2023002 |
[3] | Mikyoung Lee, Jihoon Ok . Local Calderón-Zygmund estimates for parabolic equations in weighted Lebesgue spaces. Mathematics in Engineering, 2023, 5(3): 1-20. doi: 10.3934/mine.2023062 |
[4] | Boumediene Abdellaoui, Kheireddine Biroud, Ana Primo, Fernando Soria, Abdelbadie Younes . Fractional KPZ equations with fractional gradient term and Hardy potential. Mathematics in Engineering, 2023, 5(2): 1-36. doi: 10.3934/mine.2023042 |
[5] | Boumediene Abdellaoui, Ireneo Peral, Ana Primo . A note on the Fujita exponent in fractional heat equation involving the Hardy potential. Mathematics in Engineering, 2020, 2(4): 639-656. doi: 10.3934/mine.2020029 |
[6] | Huyuan Chen, Laurent Véron . Weak solutions of semilinear elliptic equations with Leray-Hardy potentials and measure data. Mathematics in Engineering, 2019, 1(3): 391-418. doi: 10.3934/mine.2019.3.391 |
[7] | Edgard A. Pimentel, Miguel Walker . Potential estimates for fully nonlinear elliptic equations with bounded ingredients. Mathematics in Engineering, 2023, 5(3): 1-16. doi: 10.3934/mine.2023063 |
[8] | Catharine W. K. Lo, José Francisco Rodrigues . On the obstacle problem in fractional generalised Orlicz spaces. Mathematics in Engineering, 2024, 6(5): 676-704. doi: 10.3934/mine.2024026 |
[9] | Silvia Cingolani, Marco Gallo, Kazunaga Tanaka . On fractional Schrödinger equations with Hartree type nonlinearities. Mathematics in Engineering, 2022, 4(6): 1-33. doi: 10.3934/mine.2022056 |
[10] | Virginia Agostiniani, Lorenzo Mazzieri, Francesca Oronzio . A geometric capacitary inequality for sub-static manifolds with harmonic potentials. Mathematics in Engineering, 2022, 4(2): 1-40. doi: 10.3934/mine.2022013 |
We extend the so-called universal potential estimates of Kuusi-Mingione type (J. Funct. Anal. 262: 4205–4269, 2012) to the singular case 1<p≤2−1/n for the quasilinear equation with measure data
−div(A(x,∇u))=μ
in a bounded open subset Ω of Rn, n≥2, with a finite signed measure μ in Ω. The operator div(A(x,∇u)) is modeled after the p-Laplacian Δpu:=div(|∇u|p−2∇u), where the nonlinearity A(x,ξ) (x,ξ∈Rn) is assumed to satisfy natural growth and monotonicity conditions of order p, as well as certain additional regularity conditions in the x-variable.
We are concerned here with the quasilinear elliptic equation with measure data
−div(A(x,∇u))=μ, | (1.1) |
in a bounded open subset Ω of Rn, n≥2. Here μ is a finite signed measure in Ω and the nonlinearity A=(A1,…,An):Rn×Rn→Rn is vector valued function. Throughout the paper, we assume that there exist Λ≥1 and p>1 such that
|A(x,ξ)|≤Λ|ξ|p−1,|DξA(x,ξ)|≤Λ|ξ|p−2, | (1.2) |
⟨DξA(x,ξ)η,η⟩≥Λ−1|ξ|p−2|η|2 | (1.3) |
for every x∈Rn and every (ξ,η)∈Rn×Rn∖{(0,0)}. More regularity assumptions on function x↦A(x,ξ) will be needed later.
A typical example of (1.1) is the p-Laplace equation with measure data
−Δpu:=−div(|∇u|p−2∇u)=μinΩ. | (1.4) |
Since the seminal work of Kilpeläinen and Malý [7] (see also [16] for a different approach), the study of pointwise behaviors of solutions to quasilinear equations with measure data (1.1) has undergone substantial progress. In particular, the series of works [4,5,9] (see also [12]) provide interesting pointwise bounds for gradients of solutions to the seemingly unwieldy Eq (1.1), at least for p>2−1n. These pointwise gradient bounds have been extended recently in [3,14,15] for the more singular case 1<p≤2−1n.
On the other hand, a more unified approach to pointwise bounds for solutions and their gradients was presented in [8]. The results of [8] give pointwise bounds not only for the size but also for the oscillation of solutions and their derivatives expressed in terms of bounds by linear or nonlinear potentials in certain Calderón spaces. These cover different kinds of pointwise fractional derivative estimates as well as estimates for (sharp) fractional maximal functions of the solutions and their gradients.
However, the treatment of [8] is still confined to the range p>2−1n, and the purpose of this note is to extend it to the singular case 1<p≤2−1n. Note that, for 1<p≤2−1n, by looking at the fundamental solution we see that in general distributional solutions of (1.4) may not even belong to W1,1loc(Ω).
Thus in this paper we shall restrict ourselves only to the case
1<p≤2−1n, |
and note that the main results obtained here also hold in the case 2−1n<p<2 thanks to [8]. Moreover, except for the comparison estimates obtained earlier in [13,15], the methods used in this paper are very much guided by those of [8]. We would also like to point out that there are analogous results in the case p≥2 that we refer to [8] for the precise statements.
In some sense our pointwise regularity for the non-homogeneous equation (1.1) is obtained from perturbation/interpolation arguments involving the associated homogeneous equations. Thus information on the regularity of associated homogeneous equations will play an important role. In this direction, we first recall a quantitative version of the well-known De Giorgi's result that established Cα0, α0∈(0,1), regularity for solutions of div(A(x,∇w))=0. Henceforth, by Qr(x0) we mean the open cube Qr(x0):=x0+(−r,r)n with center x0∈Rn and side-length 2r. In other words,
Qr(x0)={x∈Rn:|x−x0|∞:=max1≤i≤n|xi−x0i|<r}. |
Lemma 1.1. Under (1.2)–(1.3), let w∈W1,p(Ω), p>1, be a solution of the equation div(A(x,∇w))=0 in Ω. Then there exists α0∈(0,1), depending only on n,p and Λ, such that for any cubes Qρ(x0)⊂QR(x0)⊂Ω, and ϵ∈(0,1], we have
−∫Qρ(x0)|w−(w)Qρ(x0)|pdx≲(ρR)α0p−∫QR(x0)|w−(w)QR(x0)|pdx, | (1.5) |
and
infq∈R−∫Qρ(x0)|w−q|ϵdx≲(ρR)α0ϵinfq∈R−∫QR(x0)|w−q|ϵdx. | (1.6) |
We point out that the proof of (1.5) follows from [6,Chapter 7], whereas the proof of (1.6) follows from (1.5) and the reverse Hölder property of w.
In the case the nonlinearity A(x,ξ) is independent of x, we actually have C1,β0, β0∈(0,1), regularity the homogeneous equation (see, e.g., [2,10,11]). For our purpose, we shall use the following quantitative version of this regularity result (see [3,5]).
Lemma 1.2. Let v∈W1,p(Ω), p>1, be a solution of div(A0(∇v))=0 in Ω, where A0(ξ) satisfies (1.2)–(1.3) and is independent of x. Then there exists β0∈(0,1), depending only on n,p and Λ, such that for any cubes Qρ(x0)⊂QR(x0)⊂Ω and ϵ∈(0,1], we have
−∫Qρ(x0)|∇w−(∇w)Qρ(x0)|dx≲(ρR)β0−∫QR(x0)|∇w−(∇w)QR(x0)|dx, |
and
infq∈Rn−∫Qρ(x0)|∇v−q|ϵdx≲(ρR)β0ϵinfq∈Rn−∫QR(x0)|∇v−q|ϵdx. | (1.7) |
In what follows, we shall use the (maximal) constants α0 in Lemma 1.1 and β0 in Lemma 1.2 as certain thresholds in our regularity theory. Also, henceforth, we reserve the letter κ for the following constant
κ:=(p−1)2/2. | (1.8) |
Our first result provides a De Giorgi's theory for non-homogeneous equations with measure data, which also includes [15,Theorem 1.4] as an end-point case. For the case p>2−1/n, see [8,Theorem 1.1].
Theorem 1.1. Under (1.2)–(1.3), with 1<p≤2−1n, let κ be as in (1.8), and suppose that u∈C0(Ω)∩W1,ploc(Ω) is a solution of (1.1). Let QR(x0)⊂Ω and ˉα∈(0,α0), where α0 is as in Lemma 1.1. Then for any x,y∈QR/8(x0) we have
|u(x)−u(y)|≲[WR1−α(p−1)/p,p(|μ|)(x)+WR1−α(p−1)/p,p(|μ|)(y)]|x−y|α+(−∫QR(x0)|u|κdx)1κ(|x−y|R)α | (1.9) |
uniformly in α∈[0,ˉα]. Here the implicit constant depends only on n,p,Λ, and ˉα.
In (1.9), the function WR1−α(p−1)/p,p(|μ|)(⋅) is a truncated Wolff's potential of |μ|. In general, given a nonnegative measure ν and ρ>0, the Wolff's potential Wρα,sν, α>0,s>1, is defined by
Wρα,s(ν)(x):=∫ρ0[ν(Qt(x))tn−αs]1s−1dtt,x∈Rn. |
Note that Wρα,2(ν)=Iρ2α(ν), where Iργ(ν), γ>0, is a truncated Riesz's potential defined by
Iργ(ν)(x):=∫ρ0ν(Qt(x))tn−γdtt,x∈Rn. |
We remark that, except for (1.2)–(1.3), no further regularity assumption is needed in Theorem 1.1. However, this will force the constant ˉα to be small in general.
On the other hand, it is possible to allow ˉα to be arbitrarily close to 1 as long as we further impose a 'small BMO' condition on the map x↦A(x,ξ). This condition entails the smallness of the limit lim supρ→0ω(ρ), where
ω(ρ):=supy∈Rn[−∫Qr(y)Υ(A,Qr(y))(x)2dx]12,ρ>0, | (1.10) |
and for each cube Qr(y) we set
Υ(A,Qr(y))(x):=supξ∈Rn∖{0}|A(x,ξ)−¯AQr(y)(ξ)||ξ|p−1, |
with ¯AQr(y)(ξ)=−∫Qr(y)A(x,ξ)dx. The precise statement is as follows.
Theorem 1.2. Under (1.2)–(1.3), with 1<p≤2−1n, let κ be as in (1.8), and suppose that u∈C0(Ω)∩W1,ploc(Ω) is a solution to (1.1). Let QR(x0)⊂Ω. Then for any positive ˉα<1 there exists a small δ=δ(n,p,Λ,ˉα)>0 such that if
lim supρ→0ω(ρ)≤δ, | (1.11) |
then for any x,y∈QR/8(x0)⊂Ω, we have
|u(x)−u(y)|≲[WR1−α(p−1)/p,p(|μ|)(x)+WR1−α(p−1)/p,p(|μ|)(y)]|x−y|α+(−∫QR(x0)|u|κdx)1κ(|x−y|R)α | (1.12) |
uniformly in α∈[0,ˉα]. Here the implicit constant depends on n,p,Λ,ˉα,ω(⋅), and diam(Ω).
Under a certain Dini-VMO condition, we could also allow ˉα=1 in the above theorem. However, in this case the Wolff's potential is replaced with a Riesz's potential raised to the power of 1p−1.
Theorem 1.3. Under (1.2)–(1.3), with 1<p≤2−1n, let κ be as in (1.8), and suppose that u∈C1(Ω) is a solution to (1.1). Let QR(x0)⊂Ω. If for some σ1∈(0,1) such that ω(⋅)σ1 is Dini-VMO, i.e.,
∫10ω(ρ)σ1dρρ<+∞, | (1.13) |
then for any x,y∈QR/8(x0)⊂Ω, we have
|u(x)−u(y)|≲[(IRp−α(p−1)(|μ|)(x))1p−1+(IRp−α(p−1)(|μ|)(y))1p−1]|x−y|α+(−∫QR(x0)|u|κdx)1κ(|x−y|R)α |
uniformly in α∈[0,1]. Here the implicit constant depends on n,p,Λ,ˉα,σ1,ω(⋅), and diam(Ω).
We remark that, when α=1, Theorem 1.3 recovers the pointwise gradient estimates of [3] and [15] that were obtained under a slightly different Dini condition.
Finally, under a stronger Dini-Hölder condition we can also bound solution gradients in appropriate Calderón spaces.
Theorem 1.4. Under (1.2)–(1.3), with 1<p≤2−1n, let κ be as in (1.8), and suppose that u∈C1(Ω) is a solution to (1.1). Let QR(x0)⊂Ω. If for some σ1∈(0,1) such that ω(⋅)σ1 is Dini-Hölder of order ˉα, i.e.,
∫10ω(ρ)σ1ρˉαdρρ<+∞ | (1.14) |
for some ˉα∈[0,β0), then for any x,y∈QR/4(x0)⊂Ω, we have
|∇u(x)−∇u(y)|≲[(IR1−α(|μ|)(x))1p−1+(IR1−α(|μ|)(y))1p−1]|x−y|α+(−∫QR(x0)|∇u|κdx)1κ(|x−y|R)α |
uniformly in α∈[0,ˉα]. Here β0 is as in Lemma 1.2, and the implicit constant depends on n,p,Λ,ˉα,σ1,ω(⋅), and diam(Ω).
The study of regularity problems for Eq (1.1) is based on the following comparison estimate that connects the solution of measure datum problem to a solution of a homogeneous problem.
To describe it, we let u∈W1,ploc(Ω) be a solution of (1.1). Then for a cube Q2R=Q2R(x0)⋐Ω, we consider the unique solution w∈W1,p0(Q2R(x0))+u to the local interior problem
{−div(A(x,∇w))=0inQ2R(x0),w=uon∂Q2R(x0). | (2.1) |
Lemma 2.1. Suppose that Q3R(x0)⊂Ω for some R>0. Let u and w be as in (2.1) and let κ be as in (1.8), where 1<p≤2−1n. Then it holds that
(−∫Q2R(x0)|∇(u−w)|κdx)1κ≲(|μ|(Q3R(x0))Rn−1)1p−1+|μ|(Q3R(x0))Rn−1(−∫Q3R(x0)|∇u|κdx)2−pκ. | (2.2) |
Proof. For 1<p≤3n−22n−1, inequality (2.2) was obtained in [15,Theorem 1.2]. For 3n−22n−1<p≤2−1n, by [13,Lemma 2.2], we have
(−∫Q2R(x0)|∇(u−w)|γ0dx)1γ0≲(|μ|(Q2R(x0))Rn−1)1p−1+|μ|(Q2R(x0))Rn−1(−∫Q2R(x0)|∇u|γ0dx)2−pγ0 |
for some γ0∈[2−p2,n(p−1)n−1). In fact, an inspection of the proof of [13,Lemma 2.2] reveals that we can take any γ0∈(n2n−1,n(p−1)n−1). Thus we may assume that κ=(p−1)2/2<γ0. To conclude the proof, it is therefore enough to show that
(−∫Q2R(x0)|∇u|γ0dx)1γ0≲(|μ|(Q3R(x0))Rn−1)1p−1+(−∫Q3R(x0)|∇u|κdx)1κ. | (2.3) |
To this end, let γ1∈(γ0,n(p−1)n−1). By [15,Corollay 2.4], we have
(−∫Qρ(x)|∇u|γ1dy)1γ1≲(|μ|(Q9ρ/8(x))ρn−1)1p−1+1ρ(−∫Q9ρ/8(x)|u−λ|γ0dy)1γ0 | (2.4) |
for any λ∈R and any cube Qρ(x) such that Q9ρ/8(x)⊂Ω.
Now, with Q8ρ/7(x)⊂Ω, let w1 be the unique solution w1∈W1,p0(Q8ρ/7(x))+u to the problem
{−div(A(x,∇w1))=0inQ8ρ/7(x),w1=uon∂Q8ρ/7(x). |
Then from the proof of [13,Lemma 2.2] (using (2.8) and (2.18) in [13]), we can deduce that
1ρ(−∫Q8ρ/7(x)|u−w1|γ0dy)1γ0≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+|μ|(Q8ρ/7(x))ρn−1(−∫Q8ρ/7(x)|∇u|γ0dy)2−pγ0. | (2.5) |
By Young's inequality, this yields
1ρ(−∫Q8ρ/7(x)|u−w1|γ0dy)1γ0≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+(−∫Q8ρ/7(x)|∇u|γ0dy)1γ0. | (2.6) |
Thus by quasi-triangle and Hölder's inequalities we get
1ρ(−∫Q9ρ/8(x)|u−λ|γ0dy)1γ0≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+(−∫Q8ρ/7(x)|∇u|γ0dy)1γ0+1ρ(−∫Q9ρ/8(x)|w1−λ|nn−1dy)n−1n, | (2.7) |
where we choose λ=−∫Q9ρ/8(x)w1dz.
We now use Poincaré and the reverse Hölder's inequalities for ∇w1 to obtain that
1ρ(−∫Q9ρ/8(x)|w1−λ|nn−1dy)n−1n≲−∫Q9ρ/8(x)|∇w1|dy≲(−∫Q8ρ/7(x)|∇w1|γ0dy)1γ0≲(−∫Q8ρ/7(x)|∇u−∇w1|γ0dy)1γ0+(−∫Q8ρ/7(x)|∇u|γ0dy)1γ0≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+(−∫Q8ρ/7(x)|∇u|γ0dy)1γ0, |
where we used [13,Lemma 2.2] and Young's inequality in the last bound.
Thus combining this result with (2.7) we find
1ρ(−∫Q9ρ/8(x)|u−λ|γ0dy)1γ0≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+(−∫Q8ρ/7(x)|∇u|γ0dy)1γ0. |
At this point, plugging this into (2.4) we arrive at
(−∫Qρ(x)|∇u|γ1dy)1γ1≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+(−∫Q8ρ/7(x)|∇u|γ0dy)1γ0, |
which holds for any cube Qρ(x) such that Q8ρ/7(x)⊂Ω. Recall that γ1>γ0, and thus by a covering/iteration argument as in [6,Remark 6.12], we have
(−∫Qρ(x)|∇u|γ1dy)1γ1≲(|μ|(Q8ρ/7(x))ρn−1)1p−1+(−∫Q8ρ/7(x)|∇u|ϵdy)1ϵ | (2.8) |
for any ϵ>0. This obviously yields (2.3) as desired and the proof is complete.
Remark 2.1. Using the above argument, in particular (2.5), we can also show the following comparison estimate for the functions u and w: for any 3n−22n−1<p≤2−1n,
(−∫Q2R(x0)|u−w|κdx)1κ≲(|μ|(Q3R(x0))Rn−p)1p−1+|μ|(Q3R(x0))Rn−2(−∫Q3R(x0)|∇u|κdx)2−pκ, |
and
(−∫Q2R(x0)|u−w|κdx)1κ≲(|μ|(Q3R(x0))Rn−p)1p−1+|μ|(Q3R(x0))Rn−p(−∫Q3R(x0)|u−λ|κdx)2−pκ |
for any λ∈R. For 1<p≤3n−22n−1, these inequalities have been obtained in [15,Theorem 1.2].
The following Poincaré type inequality was obtained in the case 1<p≤3n−22n−1 in [15,Corollary 1.3]. A similar proof using Lemma 2.1 and inequalities of the form (2.6) and (2.8) also yields the result in the case 3n−22n−1<p≤2−1n.
Corollary 2.1. Suppose that Q3r/2(x0)⊂Ω for some r>0. Let u∈W1,ploc(Ω), 1<p≤2−1n, be a solution of (1.1). Then for any ϵ>0 we have
infq∈R(−∫Qr(x0)|u−q|ϵ)1ϵ≲(|μ|(Q3r/2(x0))rn−p)1p−1+r(−∫Q3r/2(x0)|∇u|ϵ)1ϵ. |
With u and w as in (2.1), we now consider another auxiliary function v such that v∈W1,p0(QR(x0))+w is the unique solution to the equation
{−div(¯AQR(x0)(∇v))=0inQR(x0),v=won∂QR(x0), | (2.9) |
where ¯AQR(x0)(ξ)=−∫QR(x0)A(x,ξ)dx.
The following result can be deduced from [8,Lemma 2.3] and an appropriate reverse Hölder's inequality.
Lemma 2.2. Let p>1, 0<ϵ≤p, and u,w, and v be as in (2.1) and (2.9), where Q2R(x0)⋐Ω. Then there exists a small positive constant σ0>0 such that
(−∫QR(x0)|∇v−∇w|ϵdx)1ϵ≲ω(R)σ0(−∫Q2R(x0)|∇w|ϵdx)1ϵ, |
where ω(⋅) is as defined in (1.10).
Likewise, following lemma follows from [8,Lemma 2.5].
Lemma 2.3. Let 1<p<2, 0<ϵ≤p, and u,w, and v be as in (2.1) and (2.9), where Q2R(x0)⋐Ω. Then for any σ1∈(0,1) such that ω(⋅)σ1 is Dini-VMO, i.e., (1.13) holds, it follows that
(−∫QR(x0)|∇v−∇w|ϵdx)1ϵ≲ω(R)σ1(−∫Q2R(x0)|∇w|ϵdx)1ϵ. |
As in [8], our proofs of Theorems 1.1–1.4 are based on the corresponding pointwise estimates for the associate fractional and sharp fractional maximal functions, which are interesting in their own right. This section is devoted to such pointwise fractional maximal function bounds.
Given R>0 and q>0, following [1], we define the following truncated sharp fractional maximal function of a function f∈Lqloc(Rn):
M#,Rα,q(f)(x):=sup0<ρ≤Rinfm∈Rρ−α(−∫Qρ(x)|f−m|qdx)1q,α≥0. |
Also, we define a truncated fractional maximal function by
MRβ,q(f)(x):=sup0<ρ≤Rρβ(−∫Qρ(x)|f|qdx)1q,β∈[0,n/q]. |
In the case q=1, we usually drop the index q in the above notation, i.e., we set M#,Rα,1(f)=M#,Rα(f) and MRβ,1(f)=MRβ(f). Moreover, the definition of MRβ(f) can also be naturally extended to the case where f=μ is a locally finite signed measure in Rn:
MRβ(μ)(x):=sup0<ρ≤Rρβ|μ|(Qρ(x))|Qρ(x)|,β∈[0,n/q]. |
Note that by Poincaré inequality we have
M#,Rβ(f)(x)≲MR1−β(∇f)(x),β∈[0,1], |
for any f∈W1,1loc(Rn).
On the other hand, if u∈W1,ploc(Ω), 1<p≤2−1n, then it follows from Corollary 2.1 that
M#,Rβ,ϵ(u)(x)≲[M3R/2p−β(p−1)(μ)(x)]1p−1+M3R/21−β,ϵ(∇u)(x),β∈[0,1], | (3.1) |
for any ϵ∈(0,1) and any cube Q3R/2(x)⊂Ω.
The following fractional maximal function bound will be needed in the proof of Theorem 1.1.
Theorem 3.1. Under (1.2)–(1.3), let 1<p≤2−1n, and suppose that u∈W1,ploc(Ω) is a solution of (1.1). Let Q3R(x)⊂Ω and ˉα∈(0,α0), where α0∈(0,1) is as in Lemma 1.1. Then we have
M#,2Rα,κ(u)(x)+M3R1−α,κ(∇u)(x)≲[M3Rp−α(p−1)(μ)(x)]1p−1+R1−α(−∫Q3R(x)|∇u|κdy)1κ | (3.2) |
uniformly in α∈[0,ˉα]. Here the implicit constant depends on n,p,Λ, and ˉα.
Proof. The main idea of the proof of (3.2) lies the proof of [8,Proposition 3.1] that treated the case p>2−1n. Note that by (3.1) it is enough to show
MϵR1−α,κ(∇u)(x)≲[M3Rp−α(p−1)(μ)(x)]1p−1+R1−α(−∫Q3R(x)|∇u|κdy)1κ, | (3.3) |
for some ϵ=ϵ1(n,p,Λ,ˉα)∈(0,1).
Let 0<ρ≤r≤R, and choose w as in (2.1) with Q2r(x) in place of Q2R(x0). We have
−∫Qρ(x)|∇u|κdy≲−∫Qρ(x)|∇w|κdy+(rρ)n−∫Q2r(x)|∇u−∇w|κdy≲(ρr)(α0−1)κ−∫Q2r(x)|∇w|κdy+(rρ)n−∫Q2r(x)|∇u−∇w|κdy≲(ρr)(α0−1)κ−∫Q2r(x)|∇u|κdy+{(ρr)(α0−1)κ+(rρ)n}−∫Q2r(x)|∇u−∇w|κdy≲(ρr)(α0−1)κ−∫Q2r(x)|∇u|κdy+(rρ)n−∫Q2r(x)|∇u−∇w|κdy, |
where we used the inequality
−∫Qρ(x)|∇w|κdy≲(ρr)(α0−1)κ−∫Q2r(x)|∇w|κdy, |
which is a modified version of [8,Theorem 2.2], in the second inequality.
Thus by Lemma 2.1 we get
(−∫Qρ(x)|∇u|κdy)1/κ≲(ρr)α0−1(−∫Q2r(x)|∇u|κdy)1/κ+(rρ)n/κ[|μ|(Q3r(x))rn−1]1p−1+(rρ)n/κ(|μ|(Q3r(x))rn−1)(−∫Q3r(x)|∇u|κdy)(2−p)/κ. |
Let ϵ∈(0,1), and choose ρ=ϵr. Then by Young's inequality we have
(−∫Qϵr(x)|∇u|κdy)1/κ≤C(ϵ)[|μ|(Q3r(x))rn−1]1p−1+[Cϵα0−1+1](−∫Q3r(x)|∇u|κdy)1/κ. |
Multiplying both sides by (ϵr)1−α, 0<α≤ˉα<α0, and taking the supremum with respect to r∈(0,R], we find
sup0<r≤ϵRr1−α(−∫Qr(x)|∇u|κdy)1/κ≤C(ϵ)sup0<r≤R[|μ|(Q3r(x))rn−p+α(p−1)]1p−1+[Cϵα0−1+1](ϵ/3)1−αsup0<r≤R(3r)1−α(−∫Q3r(x)|∇u|κdy)1/κ. |
We now choose ϵ∈(0,1) such that
[Cϵα0−1+1](ϵ/3)1−ˉα≤1/2, |
to deduce that
sup0<r≤ϵRr1−α(−∫Qr(x)|∇u|κdy)1/κ≤C(ϵ)sup0<r≤R[|μ|(Q3r(x))rn−p+α(p−1)]1p−1+supϵR<r≤3Rr1−α(−∫Q3r(x)|∇u|κdy)1/κ≲[M3Rp−α(p−1)(μ)(x)]1p−1+R1−α(−∫Q3R(x)|∇u|κdy)1κ. |
This is (3.3) and the proof is complete.
The following result will be needed for the proof of Theorem 1.2.
Theorem 3.2. Let 1<p≤2−1n and u∈C0(Ω) be a solution to (1.1). Suppose that Q3R(x)⊂Ω. Then for any positive ˉα<1 there exists a small δ=δ(n,p,Λ,ˉα)>0 such that if (1.11) holds, then the estimate
M#,2Rα,κ(u)(x)+M3R1−α,κ(∇u)(x)≲[M3Rp−α(p−1)(μ)(x)]1p−1+R1−α(−∫Q3R(x)|∇u|κdy)1κ |
holds uniformly in α∈[0,ˉα]. Here the implicit constant depends on n,p,Λ,ˉα,ω(⋅), and diam(Ω).
Proof. The proof is similar to that of Theorem 3.1, but this time we need to use Lemma 2.2. As above, by (3.1) it is enough to show (3.3) for some ϵ=ϵ1(n,p,Λ,ˉα)∈(0,1). Let 0<ρ≤r≤R, and choose w as in (2.1) with Q2r(x) in place of Q2R(x0). Then choose v as in (2.9) with Qr(x) in place of QR(x0). This time we have
−∫Qρ(x)|∇u|κdy≲−∫Qρ(x)|∇v|κdy+(rρ)n−∫Qr(x)|∇v−∇w|κdy+(rρ)n−∫Q2r(x)|∇u−∇w|κdy≲−∫Qr(x)|∇v|κdy+(rρ)n−∫Qr(x)|∇v−∇w|κdy+(rρ)n−∫Q2r(x)|∇u−∇w|κdy≲−∫Qr(x)|∇u|κdy+{1+(rρ)n}(−∫Qr(x)|∇v−∇w|κdy+−∫Q2r(x)|∇u−∇w|κdy)≲−∫Qr(x)|∇u|κdy+(rρ)n−∫Qr(x)|∇v−∇w|κdy+(rρ)n−∫Q2r(x)|∇u−∇w|κdy. |
Here we used
−∫Qρ(x)|∇v|κdy≲−∫Qr(x)|∇v|κdy, | (3.4) |
which is a a modified version of (2.6) in [8,Theorem 2.1] in the second inequality.
Then by Lemma 2.2 we get
(−∫Qρ(x)|∇u|κdy)1/κ≲(−∫Qr(x)|∇u|κdy)1/κ+(rρ)n/κω(r)σ0(−∫Q2r(x)|∇w|κdy)1/κ+(rρ)n/κ(−∫Q2r(x)|∇u−∇w|κdy)1/κ≲{1+(rρ)n/κω(r)σ0}(−∫Q2r(x)|∇u|κdy)1/κ+{(rρ)n/κω(r)σ0+(rρ)n/κ}(−∫Q2r(x)|∇u−∇w|κdy)1/κ, |
for a small constant σ0>0. Thus using Lemma 2.1 and the fact that ω(r)≤2Λ, we find
(−∫Qρ(x)|∇u|κdy)1/κ≲{1+(rρ)n/κω(r)σ0}(−∫Q2r(x)|∇u|κdy)1/κ+(rρ)n/κ[|μ|(Q3r(x))rn−1]1p−1+(rρ)n/κ(|μ|(Q3r(x))rn−1)(−∫Q3r(x)|∇u|κdy)(2−p)/κ. | (3.5) |
Let ϵ∈(0,1), and choose ρ=ϵr. Then by Young's inequality we have
(−∫Qϵr(x)|∇u|κdy)1/κ≤Cϵ[|μ|(Q3r(x))rn−1]1p−1+[c1ϵ−n/κω(r)σ0+c2](−∫Q3r(x)|∇u|κdy)1/κ. |
Multiplying both sides by (ϵr)1−α, 0<α≤ˉα<1, and taking the supremum with respect to r∈(0,R], we find
sup0<r≤ϵRr1−α(−∫Qr(x)|∇u|κdy)1/κ≤Cϵsup0<r≤R[|μ|(Q3r(x))rn−p+α(p−1)]1p−1+[c1ϵ−n/κsup0<r≤Rω(r)+c2](ϵ/3)1−αsup0<r≤R(3r)1−α(−∫Q3r(x)|∇u|κdy)1/κ. |
We now choose ϵ∈(0,1) such that
c2(ϵ/3)1−ˉα≤1/4, |
and then choose ˉR=ˉR(n,p,Λ,ˉα,ω(⋅))>0 and a small δ=δ(n,p,Λ,ˉα)>0 in (1.11) such that
c1ϵ−n/κsup0<r≤ˉRω(r)(ϵ/3)1−ˉα≤c1ϵ−n/κ(2δ)(ϵ/3)1−ˉα≤1/4. |
Then it follows that
[c1ϵ−n/κsup0<r≤Rω(r)+c2](ϵ/3)1−α≤1/2, |
provided R≤ˉR. Hence, for R≤ˉR, we deduce that
sup0<r≤ϵRr1−α(−∫Qr(x)|∇u|κdy)1/κ≤C(ϵ)sup0<r≤3R[|μ|(Qr(x))rn−p+α(p−1)]1p−1+supϵR<r≤3Rr1−α(−∫Qr(x)|∇u|κdy)1/κ≲[M3Rp−α(p−1)(μ)(x)]1p−1+R1−α(−∫Q3R(x)|∇u|κdy)1κ. |
This proves (3.3) in the case R≤ˉR. For R>ˉR, we observe that
MϵR1−α,κ(∇u)(x)≤MϵˉR1−α,κ(∇u)(x)+(RˉR)n/κ(ϵR)1−α(−∫QϵR(x)|∇u|κdy)1/κ. |
Thus we also obtain (3.3) in the case R>ˉR as long as we allow the implicit constant to depend on diam(Ω), and n,p,Λ,ˉα,ω(⋅).
In order to prove Theorem 1.3, we need the following pointwise fractional maximal function bound.
Theorem 3.3. Let 1<p≤2−1n and u∈C1(Ω) be a solution to (1.1). Suppose that Q3R(x)⊂Ω. If for some σ1∈(0,1) such that ω(⋅)σ1 is Dini-VMO, i.e., (1.13) holds, then the estimate
M#,Rα,κ(u)(x)+M3R1−α,κ(∇u)(x)≲[I3Rp−α(p−1)(|μ|)(x)]1p−1+R1−α(−∫Q3R(x)|∇u|κdy)1κ |
holds uniformly in α∈[0,1]. Here the implicit constant depends on n,p,Λ,ˉα,ω(⋅), σ1, and diam(Ω).
Proof. As in the proof of Theorem 3.2, it is enough to show
MR1−α,κ(∇u)(x)≲[I2Rp−α(p−1)(|μ|)(x)]1p−1+R1−α(−∫QR(x)|∇u|κdy)1κ. |
Moreover, we may assume that R≤ˉR, where ˉR=ˉR(n,p,Λ,σ1,ω(⋅))>0 is to be determined.
Arguing as in the proof of (3.5), but this time using (1.7) (in Lemma 1.2) instead of (3.4) and Lemma 2.3 instead of Lemma 2.2, we have for Qρ(x)⊂Qr(x)⊂Q3r(x)⊂Ω,
(−∫Qρ(x)|∇u−qQρ(x)|κdy)1/κ≲(ρr)β0(−∫Q3r(x)|∇u−qQ3r(x)|κdy)1/κ+(rρ)n/κω(r)σ1(−∫Q3r(x)|∇u|κdy)1/κ+(rρ)n/κ[|μ|(Q3r(x))rn−1]1p−1+(rρ)n/κ(|μ|(Q3r(x))rn−1)(−∫Q3r(x)|∇u|κdy)(2−p)/κ. | (3.6) |
Here qQρ(x)∈Rn is defined by
qQρ(x):=argminq∈Rn(−∫Qρ(x)|∇u−q|κdy)1/κ,Qρ(x)⋐Ω. |
That is, qQρ(x) is a vector such that
infq∈Rn(−∫Qρ(x)|∇u−q|κdy)1/κ=(−∫Qρ(x)|∇u−qQρ(x)|κdy)1/κ. |
Note that for Qρ(x)⊂Qs(x)⋐Ω, one has
|qQs(x)|=(−∫Qs(x)|qQs(x)|κdy)1/κ≲(−∫Qs(x)|∇u−qQs(x)|κdy)1/κ+(−∫Qs(x)|∇u|κdy)1/κ≲(−∫Qs(x)|∇u|κdy)1/κ, | (3.7) |
and also
|qQρ(x)−qQs(x)|=(−∫Qρ(x)|qQρ(x)−qQs(x)|κdy)1/κ≲(−∫Qρ(x)|∇u−qQρ(x)|κdy)1/κ+(−∫Qρ(x)|∇u−qQs(x)|κdy)1/κ≲(sρ)n/κ(−∫Qs(x)|∇u−qQs(x)|κdy)1/κ. | (3.8) |
For brevity, for any j=0,1,2,…, and Q3R(x)⊂Ω, we now define
Qj=QRj(x),Rj=ϵjR, |
where ϵ∈(0,1/3) is to be determined, and
Aj=(−∫Qj|∇u−qj|κdy)1/κ,qj=qQj. |
Then applying (3.6) with ρ=ϵRj<r=Rj/3 we have
Aj+1≤c1ϵβ0Aj+c2ϵ−n/κω(Rj/3)σ1(−∫Qj|∇u|κdy)1/κ+Cϵ[|μ|(Qj)Rn−1j]1p−1+Cϵ(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. | (3.9) |
By quasi-triangle inequality, this yields
Aj+1≤c1ϵβ0Aj+c2ϵ−n/κω(Rj/3)σ1Aj+c2ϵ−n/κω(Rj/3)σ1|qj|+Cϵ[|μ|(Qj)Rn−1j]1p−1+Cϵ(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
We now choose ϵ sufficiently small so that c1ϵβ0≤1/4 and then restrict R≤ˉR, where ˉR=ˉR(n,p,Λ,σ1,ω(⋅))>0 is such that
c2ϵ−n/κsup0<ρ≤ˉRω(ρ)σ1≤1/4. |
Then we have
Aj+1≤12Aj+Cω(Rj/3)σ1|qj|+C[|μ|(Qj)Rn−1j]1p−1+C(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. | (3.10) |
Summing this up over j∈{0,1,,…,m−1}, m∈N, we get
m∑j=1Aj≤12m−1∑j=0Aj+Cm−1∑j=0ω(Rj/3)σ1|qj|+Cm−1∑j=0[|μ|(Qj)Rn−1j]1p−1+Cm−1∑j=0(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
Hence,
m∑j=1Aj≤A0+Cm−1∑j=0ω(Rj/3)σ1|qj|+Cm−1∑j=0[|μ|(Qj)Rn−1j]1p−1+Cm−1∑j=0(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
On the other hand, for any m∈N, by (3.8) we can write
|qm+1|=m∑j=0(|qj+1|−|qj|)+|q0|≤Cm∑j=0Aj+|q0|, |
and therefore in view of (3.7),
|qm+1|≤cA0+|q0|+Cm−1∑j=0ω(Rj/3)σ1|qj|+Cm−1∑j=0[|μ|(Qj)Rn−1j]1p−1+Cm−1∑j=0(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ≤C(−∫QR(x)|∇u|κdy)1/κ+Cm−1∑j=0ω(Rj/3)σ1|qj|+Cm−1∑j=0[|μ|(Qj)Rn−1j]1p−1+Cm−1∑j=0(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
At this point, multiplying both sides of the above inequality by R1−αm+1, m∈N, we deduce that
R1−αm+1|qm+1|≲R1−α(−∫QR(x)|∇u|κdy)1/κ+Cm−1∑j=0ω(Rj/3)σ1R1−αj|qj|+m−1∑j=0[|μ|(Qj)Rn−p+α(p−1)j]1p−1+m−1∑j=0(|μ|(Qj)Rn−p+α(p−1)j)R(1−α)(2−p)j(−∫Qj|∇u|κdy)(2−p)/κ. |
Thus,
R1−αm+1|qm+1|≤c3R1−α(−∫QR(x)|∇u|κdy)1/κ+c3m−1∑j=0ω(Rj/3)σ1R1−αj|qj|+c3[I2Rp−α(p−1)(|μ|)(x)]1p−1+c3I2Rp−α(p−1)(|μ|)(x)[MR1−α,κ(∇u)(x)]2−p. | (3.11) |
We next further restrict ˉR so that for any R≤ˉR,
m−1∑j=0ω(Rj/3)σ1≤12c3. |
This is possible because we have
m−1∑j=0ω(Rj/3)σ1=ω(R/3)+m−1∑j=1ω(Rj/3)σ1≤c∫RR/3ω(ρ)σ1dρρ+cm−1∑j=1∫Rj−1/3Rj/3ω(ρ)σ1dρρ≤c∫R0ω(ρ)σ1dρρ, | (3.12) |
where we used the fact that ω(ρ1)≤cω(ρ2) provided ρ1≤ρ2≤Cρ1, C>1.
Then by an induction argument we deduce from (3.11) that
R1−αm|qm|≲R1−α(−∫QR(x)|∇u|κdy)1/κ+[I2Rp−α(p−1)(|μ|)(x)]1p−1+I2Rp−α(p−1)(|μ|)(x)[MR1−α,κ(∇u)(x)]2−p, | (3.13) |
for every integer m≥0.
Let us call the right-hand side of (3.13) by Q. Then from (3.10) and simple manipulations we obtain
Am+1≤12Am+c|qm|+cRα−1mQ, |
which by (3.13) yields
R1−αm+1Am+1≤12R1−αmAm+cR1−αm|qm|+cQ≤12R1−αmAm+cQ. |
As R1−α0A0≤cQ, by iteration we get
R1−αmAm≤CQ, | (3.14) |
for every integer m≥0.
To conclude the proof, we observe that
MR1−α,κ(∇u)(x)≤Csupm≥0R1−αm(−∫Qm|∇u|κdy)1/κ≤Csupm≥0[R1−αmAm+R1−αmqm]≤CQ, |
where we used (3.13) and (3.14) in the last inequality. Then recalling the definition of Q and using Young's inequality we obtain
MR1−α,κ(∇u)(x)≤CR1−α(−∫QR(x)|∇u|κdy)1κ+C[I2Rp−α(p−1)(|μ|)(x)]1p−1+12MR1−α,κ(∇u)(x). |
This completes the proof of the theorem. The following pointwise sharp fractional maximal function bound will be used in the proof of Theorem 1.4.
Theorem 3.4. Let 1<p≤2−1n and u∈C1(Ω) be a solution to (1.1). Suppose that Q3R(x)⊂Ω. If for some σ1∈(0,1) such that
sup0<ρ≤1ω(ρ)σ1ρˉα≤K, | (3.15) |
for some ˉα∈[0,β0), then the estimate
M#,3Rα,κ(∇u)(x)≲[M3R1−α,κ(μ)(x)]1p−1+[I3R1(|μ|)(x)]1p−1+R−α(−∫Q3R(x)|∇u|κdy)1κ |
holds uniformly in α∈[0,ˉα]. Here β0 is as in Lemma 1.2, and the implicit constant depends on n,p,Λ,ˉα,ω(⋅),σ1,K, and diam(Ω).
Remark 3.1. Condition (3.15) implies the Dini-VMO condition (1.13). In turns, (1.13) implies (1.11), whereas (3.15) is implied by the Dini-Hölder condition (1.14).
Proof. It suffices to show
M#,Rα,κ(∇u)(x)≲[MR1−α,κ(μ)(x)]1p−1+[IR1(|μ|)(x)]1p−1+R−α(−∫QR(x)|∇u|κdy)1κ, |
for R≤1, where the implicit constant depends on n,p,Λ,ˉα,ω(⋅),σ1,K, and diam(Ω).
With the notation used in proof of Theorem 3.3, multiplying both sides of (3.9) by R−αj+1, j≥0, we have
R−αj+1Aj+1≤c1ϵβ0−αR−αjAj+CϵR−αjω(Rj/3)σ1(−∫Qj|∇u|κdy)1/κ+CϵR−αj[|μ|(Qj)Rn−1j]1p−1+CϵR−αj(|μ|(Qj)Rn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
This time we choose ϵ∈(0,1/3) such that
c1ϵβ0−α≤c1ϵβ0−ˉα≤12, |
and employ (3.15) together with the restriction Rj≤1, to deduce
R−αj+1Aj+1≤12R−αjAj+CK(−∫Qj|∇u|κdy)1/κ+C[|μ|(Qj)Rn−1+αj]1p−1+C(|μ|(Qj)Rn−1+αj)(−∫Qj|∇u|κdy)(2−p)/κ. | (3.16) |
On the other hand, applying Theorem 3.3 in the case α=1, we can bound
(−∫Qj|∇u|κdy)1/κ≲[IR1(|μ|)(x)]1p−1+(−∫QR(x)|∇u|κdy)1κ |
for every integer j≥0. Thus, using (3.16) and Young's inequality we get
R−αj+1Aj+1≤12R−αjAj+C[MR1−α,κ(μ)(x)]1p−1+C[IR1(|μ|)(x)]1p−1+C(−∫QR(x)|∇u|κdy)1κ. |
Iterating this inequality, we find for any m∈N,
R−αmAm≤2−mR−α0A0+C[MR1−α,κ(μ)(x)]1p−1+C[IR1(|μ|)(x)]1p−1+C(−∫QR(x)|∇u|κdy)1κ≤C[MR1−α,κ(μ)(x)]1p−1+C[IR1(|μ|)(x)]1p−1+CR−α(−∫QR(x)|∇u|κdy)1κ. |
In view of the fact that
M#,Rα,κ(∇u)(x)≲supm≥0R−αmAm, |
this completes the proof of the theorem.
Proof of Theorem 1.1. For any cube Qρ(x)⋐Ω, let qQρ(x)∈R be defined by
qQρ(x):=argminq∈R(−∫Qρ(x)|u−q|κdy)1/κ, |
i.e., qQρ(x) is a real number such that
infq∈Rn(−∫Qρ(x)|u−q|κdy)1/κ=(−∫Qρ(x)|u−qQρ(x)|κdy)1/κ. |
Then using quasi-triangle inequality a few times and Lemma 1.1, we have for Qρ(x)⊂Qr(x)⊂Q3r(x)⊂Ω,
(−∫Qρ(x)|u−qQρ(x)|κdy)1/κ≲(ρr)α0(−∫Q2r(x)|u−qQ2r(x)|κdy)1κ+(ρr)−n/κ(−∫Q2r(x0)|u−w|κdy)1κ. |
Here we choose w as in (2.1) with Q2r(x) in place of Q2R(x0).
We now apply Remark 2.1 to bound the second term on the right-hand side of the above inequality. This yields that
(−∫Qρ(x)|u−qQρ(x)|κdy)1/κ≲(ρr)α0(−∫Q2r(x)|u−qQ2r(x)|κdy)1κ+(ρr)−n/κ(|μ|(Q3r(x))rn−p)1p−1+(ρr)−n/κ|μ|(Q3r(x))rn−p(−∫Q3r(x)|u−qQ3r(x)|κdy)2−pκ. |
Letting ρ=ϵr, ϵ∈(0,1), and using Young's inequality we find
(−∫Qϵr(x)|u−qQϵr(x)|κdy)1/κ≲Cϵα0(−∫Q3r(x)|u−qQ3r(x)|κdy)1κ+Cϵ(|μ|(Q3r(x0))rn−p)1p−1. | (4.1) |
Next, we choose ϵ∈(0,1/3) small enough so that Cϵα0≤12, where C is the constant in (4.1). Let QR(x0)⊂Ω be as given in the theorem. Then for any cube Qδ(x)⊂QR(x0) we set δj=ϵjδ, Qj=Qδj(x), qj=qQj, j≥0, and define
Bj:=(−∫Qj|u−qQj|κdy)1/κ. |
Applying (4.1) with r=δj/3 yields
Bj+1≤12Bj+C(|μ|(Qj)δn−pj)1p−1. |
Summing this up over j∈{1,3,...,m−1}, we obtain
m∑j=1Bj≤CB1+Cm−1∑j=1(|μ|(Qj)δn−pj)1p−1. |
As in (3.8), we have
|qj+1−qj|≤CBj |
for all integers j≥1, and thus
|qm|≤|qm−q1|+q1≤q1+Cm−1∑j=1Bj≤q1+CB1+Cm−1∑j=1(|μ|(Qj)δn−pj)1p−1≤C(−∫Q1|u|κdx)1κ+Cm−1∑j=1(|μ|(Qj)δn−pj)1p−1 | (4.2) |
holds for every integer m≥2. Here we use the simple fact (see (3.7)) that
B1+q1≤C(−∫Q1|u|κdx)1κ. |
Now for x,y∈QR/8(x0) we choose
δ=12|x−y|∞=12max1≤i≤n|xi−yi|. |
Note that δ<R/8 and Qδ(y)⊂Q3δ(x)⊂QR/2(x0). Then applying (4.2), we have
|qm|≤C(−∫Qδ(x)|u|κdz)1κ+Cδαm−1∑j=1(|μ|(Qδj(x))δn−p+α(p−1)j)1p−1. |
Sending m→∞ and using [1,Lemma 4.1], we get
|u(x)|≤C(−∫Qδ(x)|u|κdz)1κ+CδαWR1−α(p−1)/p,p(|μ|)(x). |
Since u−m,m∈R, is also a solution of (1.1), it follows that
|u(x)−m|≤C(−∫Qδ(x)|u−m|κdz)1κ+CδαWR1−α(p−1)/p,p(|μ|)(x)≤C(−∫Q3δ(x)|u−m|κdz)1κ+CδαWR1−α(p−1)/p,p(|μ|)(x). |
Likewise, we have
|u(y)−m|≤C(−∫Qδ(y)|u−m|κdz)1κ+CδαWR1−α(p−1)/p,p(|μ|)(y)≤C(−∫Q3δ(x)|u−m|κdz)1κ+CδαWR1−α(p−1)/p,p(|μ|)(y). |
Now choosing m=qQ3δ(x) we find
|u(x)−u(y)|≤C(−∫Q3δ(x)|u−qQ3δ(x)|κdz)1κ+Cδα[WR1−α(p−1)/p,p(|μ|)(x)+WR1−α(p−1)/p,p(|μ|)(y)]. | (4.3) |
On the other hand, by Theorem 3.1 and the fact that 3δ<3R/8, we have
(−∫Q3δ(x)|u−qQ3δ(x)|κdz)1κ≲δα[M9R/16p−α(p−1)(μ)(x)]1p−1+(δR)αR(−∫Q9R/16(x)|∇u|κdz)1κ≲δαWR1−α(p−1)/p,p(|μ|)(x)+(δR)α(−∫BR(x)|u|κdz)1κ, | (4.4) |
where we used a Caccioppoli type inequality of [15,Corollary 2.4] in the last bound.
Combining inequalities (4.3) and (4.4), we complete the proof of the theorem.
Proof of Theorems 1.2. The main idea of the proof of Theorem 1.2 lies in the proof of [8,Theorem 1.2]. First, in view of Theorem 1.1, it suffices to prove (1.12) uniformly in α∈[α0/2,ˉα], ˉα<1, for all x,y∈QR/8(x0).
On the other hand, for a.e. x,y∈QR/8(x0) and f∈Lκ(QR(x0)), we have the inequality
|f(x)−f(y)|≤(cα)|x−y|α[M#,R/2α,κ(f)(x)+M#,R/2α,κ(f)(y)], |
provided α∈(0,1]. See inequalities (4.9) and (4.10) in [1]. Applying this with f=u and α∈[α0/2,ˉα], and using Theorem 3.2, we obtain
|u(x)−u(y)|≤(cα0)|x−y|α[M3R/4p−α(p−1)(μ)(x)+M3R/4p−α(p−1)(μ)(y)]1p−1+(cα0)|x−y|αR1−α{(−∫Q3R/4(x)|∇u|κdy)1κ+(−∫Q3R/4(y)|∇u|κdy)1κ}. |
Then invoking the Caccioppoli type inequality of [15,Corollary 2.4] we obtain (1.12) uniformly in α∈[α0/2,ˉα] as desired.
Proof. (Proof of Theorem 1.3) The proof of Theorem 1.3 is similar to that of Theorems 1.2, but this time we use Theorem 3.3 instead of Theorem 3.2.
Proof of Theorem 1.4. Let QR(x0)⊂Ω be as given in the theorem. For any x,y∈QR/4(x0), we set δ=12|x−y|∞. Note that δ<R/4 and Qδ(y)⊂Q3δ(x)⊂QR(x0). We shall keep the notation in the proof of Theorem 3.3 except that we replace R with δ so that Rj=δj=ϵjδ, Qj=Qϵjδ(x), qj=qQϵjδ(x), and
Aj=(−∫Qϵjδ(x)|∇u−qQϵjδ(x)|κdy)1/κ=(−∫Qj|∇u−qj|κdy)1/κ |
for all integers j≥0.
Then by choosing ϵ in (3.9) such that c1ϵβ0≤1/2, we have
Aj+1≤12Aj+Cϵω(δj/3)σ1(−∫Qj|∇u|κdy)1/κ+Cϵ[|μ|(Qj)δn−1j]1p−1+Cϵ(|μ|(Qj)δn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
Summing this up over j∈{0,1,,…,m−1}, m∈N, and then simplifying, we get
m∑j=1Aj≤A0+Cm−1∑j=0ω(δj/3)σ1(−∫Qj|∇u|κdy)1/κ+Cm−1∑j=0[|μ|(Qj)δn−1j]1p−1+Cm−1∑j=0(|μ|(Qj)δn−1j)(−∫Qj|∇u|κdy)(2−p)/κ. |
On the other hand, by (3.8),
|qm+1−m|=m∑j=0(|qj+1−m|−|qj−m|)+|q0−m|≤m∑j=0(|qj+1−qj|+|q0−m|≤Cm∑j=0Aj+|q0−m|, |
which holds for any m∈Rn and integer m≥0.
Hence, it follows that
|qm+1−m|≤Cm−1∑j=0ω(δj/3)σ1(−∫Qj|∇u|κdy)1/κ+Cm−1∑j=0[|μ|(Qj)δn−1j]1p−1+Cm−1∑j=0(|μ|(Qj)δn−1j)(−∫Qj|∇u|κdy)(2−p)/κ+CA0+|q0−m|. |
Then using
|q0−m|≲(−∫Qδ(x)|∇u−m|κdy)1/κ, |
which can be proved as in (3.7), we get
|qm+1−m|≲m−1∑j=0ω(δj/3)σ1(−∫Qj|∇u|κdy)1/κ+m−1∑j=0[|μ|(Qj)δn−1j]1p−1+m−1∑j=0(|μ|(Qj)δn−1j)(−∫Qj|∇u|κdy)(2−p)/κ+(−∫Qδ(x)|∇u−qQδ(x)|κdy)1/κ+(−∫Qδ(x)|∇u−m|κdy)1/κ. |
We next set
M(x,r):=[Ir1(|μ|)(x)]1p−1+(−∫Qr(x)|∇u|κdz)1/κ,r>0. |
Then applying Theorem 3.3 with α=1 and 3R=δ, we have
(−∫Qj|∇u|κdy)1/κ≲M(x,δ),∀j≥0. |
Plugging this into the last bound for |qm+1−m| we deduce that
|qm+1−m|≲δαm−1∑j=0δ−αjω(δj/3)σ1M(x,δ)+δαm−1∑j=0[|μ|(Qj)δn−1+αj]1p−1δα(2−p)p−1+δαm−1∑j=0(|μ|(Qj)δn−1+αj)M(x,δ)2−p+(−∫Qδ(x)|∇u−qQδ(x)|κdy)1/κ+(−∫Qδ(x)|∇u−m|κdy)1/κ. |
Also, note that as in (3.12) we have
m−1∑j=0δ−αjω(δj/3)σ1≲m−1∑j=0δ−ˉαjω(δj/3)σ1≲∫δ0ω(ρ)σ1ρˉαdρρ≲∫R/40ω(ρ)σ1ρˉαdρρ. |
At this point, using the Dini-Hölder condition (1.14), we obtain, after some simple manipulations,
|qm+1−m|≲δαM(x,δ)+δα[I2δ1−α(|μ|)(x)]1p−1+δαI2δ1−α(|μ|)(x)M(x,δ)2−p+(−∫Qδ(x)|∇u−qQδ(x)|κdy)1/κ+(−∫Qδ(x)|∇u−m|κdy)1/κ. |
Here we also used that δ<R/4<diam(Ω) and the implicit constants are allowed to depend on diam(Ω).
Thus letting m→∞ and using Young's inequality we obtain
|∇u(x)−m|≲δαM(x,δ)+δα[IR1−α(|μ|)(x)]1p−1+(−∫Qδ(x)|∇u−qQδ(x)|κdz)1/κ+(−∫Qδ(x)|∇u−m|κdz)1/κ. |
Likewise, we also have
|∇u(y)−m|≲δαM(y,δ)+δα[IR1−α(|μ|)(y)]1p−1+(−∫Qδ(y)|∇u−qQδ(y)|κdz)1/κ+(−∫Qδ(y)|∇u−m|κdz)1/κ≲δαM(y,δ)+δα[IR1−α(|μ|)(y)]1p−1+(−∫Q3δ(x)|∇u−qQ3δ(x)|κdz)1/κ+(−∫Q3δ(x)|∇u−m|κdz)1/κ, |
where we used that Qδ(y)⊂Q3δ(x).
Combining these two estimates and choosing m=qQ3δ(x), we find
|∇u(x)−∇u(y)|≲δα{[IR1−α(|μ|)(x)]1p−1+[IR1−α(|μ|)(y)]1p−1}+δα[M(x,δ)+M(y,δ)]+(−∫Q3δ(x)|∇u−qQ3δ(x)|κdz)1/κ. | (6.1) |
As δ<R/4 and QR/4(x)∪QR/4(y)⊂QR(x0), we can apply Theorem 3.3 with α=1 to have the bound
M(x,δ)+M(y,δ)≲[IR/41(|μ|)(x)]1p−1+(−∫QR/4(x)|∇u|κdz)1/κ+[IR/41(|μ|)(y)]1p−1+(−∫QR/4(y)|∇u|κdz)1/κ≲[IR1−α(|μ|)(x)]1p−1+[IR1−α(|μ|)(y)]1p−1+R−α(−∫QR(x0)|∇u|κdz)1/κ. | (6.2) |
Similarly, we can use Theorem 3.4 to bound the last term on the right-hand of (6.1) as follows:
(−∫Q3δ(x)|∇u−qQ3δ(x)|κdz)1/κ≲δαM#,3R/4α,κ(∇u)(x)≲δα[M3R/41−α,κ(μ)(x)]1p−1+δα[I3R/41(|μ|)(x)]1p−1+(δR)α(−∫Q3R/4(x)|∇u|κdz)1κ≲δα[IR1−α(|μ|)(x)]1p−1+(δR)α(−∫QR(x0)|∇u|κdz)1κ. | (6.3) |
We now plug estimates (6.2) and (6.3) into (6.1) to arrive at
|∇u(x)−∇u(y)|≲δα{[IR1−α(|μ|)(x)]1p−1+[IR1−α(|μ|)(y)]1p−1}+(δR)α(−∫QR(x0)|∇u|κdz)1κ. |
This completes the proof because δ≤12|x−y|.
Remark 6.1. In Theorems 1.3–1.4 and 3.3–3.4 we may take σ1=1 in (1.13), (1.14) and (3.15), provided we replace ω with a non-decreasing function ˜ω:[0,1]→[0,∞) such that
limρ→0˜ω(ρ)=0,and|A(x,ξ)−A(y,ξ)|≤˜ω(|x−y|)|ξ|p−1 |
for all x,y,ξ∈Rn, |x−y|≤1. The reason is that in this case solutions to (2.1) are locally Lipschitz, and we can also take σ1=1 in Lemma 2.3; see [8,Section 8].
Q. H. N. is supported by the Academy of Mathematics and Systems Science, Chinese Academy of Sciences startup fund, and the National Natural Science Foundation of China (No. 12050410257 and No. 12288201) and the National Key R & D Program of China under grant 2021YFA1000800. N. C. P. is supported in part by Simons Foundation (award number 426071).
The authors declare no conflict of interest.
[1] |
R. A. DeVore, R. C. Sharpley, Maximal functions measuring smoothness, Mem. Amer. Math. Soc., 47 (1984), 293. http://doi.org/10.1090/memo/0293 doi: 10.1090/memo/0293
![]() |
[2] |
E. DiBenedetto, C1+α local regularity of weak solutions of degenerate elliptic equations, Nonlinear Anal. Theor., 7 (1983), 827–850. https://doi.org/10.1016/0362-546X(83)90061-5 doi: 10.1016/0362-546X(83)90061-5
![]() |
[3] |
H. Dong, H. Zhu, Gradient estimates for singular p-Laplace type equations with measure data, Calc. Var., 61 (2021), 86. https://doi.org/10.1007/s00526-022-02189-5 doi: 10.1007/s00526-022-02189-5
![]() |
[4] |
F. Duzaar, G. Mingione, Gradient estimates via linear and nonlinear potentials, J. Funct. Anal., 259 (2010), 2961–2998. https://doi.org/10.1016/j.jfa.2010.08.006 doi: 10.1016/j.jfa.2010.08.006
![]() |
[5] | F. Duzaar, G. Mingione, Gradient estimates via non-linear potentials, Amer. J. Math., 133 (2011), 1093–1149. |
[6] | E. Giusti, Direct methods in the calculus of variations, River Edge, NJ: World Scientic Publishing Co., Inc., 2003. https://doi.org/10.1142/5002 |
[7] |
T. Kilpeläinen, J. Malý, The Wiener test and potential estimates for quasilinear elliptic equations, Acta Math., 172 (1994), 137–161. https://doi.org/10.1007/BF02392793 doi: 10.1007/BF02392793
![]() |
[8] |
T. Kuusi, G. Mingione, Universal potential estimates, J. Funct. Anal., 262 (2012), 4205–4269. https://doi.org/10.1016/j.jfa.2012.02.018 doi: 10.1016/j.jfa.2012.02.018
![]() |
[9] |
T. Kuusi, G. Mingione, Linear potentials in nonlinear potential theory, Arch. Rational Mech. Anal., 207 (2013), 215–246. https://doi.org/10.1007/s00205-012-0562-z doi: 10.1007/s00205-012-0562-z
![]() |
[10] | P. Lindqvist, Notes on the p-Laplace equation, Univ. Jyväskylä, 2006. |
[11] | J. J. Manfredi, Regularity of the gradient for a class of nonlinear possibly degenerate elliptic equations, Ph.D. Thesis of University of Washington, St. Louis, 1986. |
[12] |
G. Mingione, Gradient potential estimates, J. Eur. Math. Soc., 13 (2011), 459–486. http://doi.org/10.4171/JEMS/258 doi: 10.4171/JEMS/258
![]() |
[13] |
Q.-H. Nguyen, N. C. Phuc, Good-λ and Muckenhoupt-Wheeden type bounds in quasilinear measure datum problems, with applications, Math. Ann., 374 (2019), 67–98. https://doi.org/10.1007/s00208-018-1744-2 doi: 10.1007/s00208-018-1744-2
![]() |
[14] |
Q.-H. Nguyen, N. C. Phuc, Pointwise gradient estimates for a class of singular quasilinear equation with measure data, J. Funct. Anal., 278 (2020), 108391. https://doi.org/10.1016/j.jfa.2019.108391 doi: 10.1016/j.jfa.2019.108391
![]() |
[15] | Q.-H. Nguyen, N. C. Phuc, A comparison estimate for singular p-Laplace equations and its consequences, submitted for publication. |
[16] | N. S. Trudinger, X.-J. Wang, On the weak continuity of elliptic operators and applications to potential theory, Amer. J. Math., 124 (2002), 369–410. |
1. |
Ke Chen, Quoc-Hung Nguyen, Na Zhao,
Global Calderón–Zygmund theory for parabolic p-Laplacian system: the case 1 10.1016/j.jfa.2023.109852 |
|
2. | Giuseppe Mingione, 2024, Chapter 2, 978-3-031-67600-0, 65, 10.1007/978-3-031-67601-7_2 | |
3. | Kyeong Song, Gradient estimates for irregular obstacle problems with measure data, 2024, 0022247X, 129168, 10.1016/j.jmaa.2024.129168 |