Abstract
We investigate the mixed local and nonlocal parabolic p-Laplace equation
where \(\Delta _p\) is the usual local p-Laplace operator and \(\mathcal {L}\) is the nonlocal p-Laplace type operator. Based on the combination of suitable Caccioppoli-type inequality and Logarithmic Lemma with a De Giorgi–Nash–Moser iteration, we establish the local boundedness and Hölder continuity of weak solutions for such equations.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this paper, we are concerned with the local behaviour of weak solutions to the following mixed problem:
where \(Q_T:=\Omega \times (0,T)\) with \(T>0\) and \(\Omega \) is a bounded domain in \({\mathbb {R}}^N\). This kind of evolution equations arises from the Lévy process, image processing etc; see [16] and references therein. The local p-Laplace operator \(\Delta _p\) is defined as follows:
and \({\mathcal {L}}\) is a nonlocal p-Laplace operator given by
where the symbol \(\mathrm {P.V.}\) stands for the Cauchy principal value. Here, K is a symmetric kernel fulfilling
and
with \(\Lambda \ge 1\) and \(0<s<1\) for all \(x,y\in {\mathbb {R}}^N\) and \(t\in (0,T)\).
Before stating our main results, let us mention some known results. For the nonlocal parabolic equations of p-Laplacian type,
the existence and uniqueness of strong solutions were verified by Vázquez [27], where the author studied the long-time behaviours as well. Mazón–Rossi–Toledo [24] established the well-posedness of solutions to Eq. (1.4) together with the asymptotic property. When it comes to regularity theory of this equation, Strömqvist [26] obtained the existence and local boundedness of weak solutions provided \(p\ge 2\). Hölder regularity with specific exponents in the case \(p\ge 2\) was proved by Brasco–Lindgren–Strömqvist [5]. Furthermore, Ding–Zhang–Zhou [12] showed the local boundedness and Hölder continuity of weak solutions to the nonhomogeneous case under the conditions that \(1<p<\infty \) and \(2<p<\infty \), respectively. We refer the readers to [6, 18, 22, 28, 29] and references therein for more results.
In the mixed local and nonlocal setting, for the case \(p=2\),
Foondun [19] has derived Harnack inequality and interior Hölder estimates for nonnegative solutions, see also [8] for a diverse approach. In addition, the Harnack inequality regarding the parabolic version of (1.5) was established in [2, 7], where, however, the authors only proved such inequality for globally nonnegative solutions. Very recently, Garain–Kinnunen [21] proved a weak Harnack inequality with a tail term for sign changing solutions to the parabolic problem of (1.5). For what concerns maximum principles, interior sobolev regularity along with symmetry results among many other quantitative and qualitative properties for solutions to (1.5), one can see for instance [3, 4, 13,14,15]. In the nonlinear framework (i.e. \(p\ne 2\)), Garain–Kinnunen [20] developed the local regularity theory for
with \(K(x,y)\simeq |x-y|^{-(N+sp)}\), involving boundedness, Hölder continuity, Harnack inequality, as well as lower/upper semicontinuity of weak supersolutions/subsolutions. Nonetheless, to the best of our knowledge, there are few results concerning on the mixed local and nonlocal nonlinear parabolic problems. To this end, influenced by the ideas developed in [10, 12, 20], we aim to establish the local boundedness and interior Hölder regularity of weak solutions to Eq. (1.1). It is noteworthy that our results are new even for the case \(p=2\).
Before giving the notion of weak solutions to (1.1), let us recall the tail space
Then, we define the tail appearing in estimates throughout this article,
where \((x_0, t_0)\in {\mathbb {R}}^N\times (0,T)\) and the interval \(I=\left[ t_0-T_1, t_0+T_2\right] \subseteq (0,T)\). This is a parabolic counterpart to the tail introduced in [10]. It is easy to check that \(\mathrm {Tail}_\infty (v;x_0,r,I)\) is well-defined for any \(v \in L^{\infty }(I; L_{sp}^{p-1}({\mathbb {R}}^{N}))\).
For any \(1<p<\infty \) and \(0<s<1\), the fractional Sobolev space is defined by
endowed with the norm
which is a reflexive Banach space, see [11, 25]. From [11, Proposition 2.2], we know that the classical Sobolev space \(W^{1,p}(\Omega )\) is continuously embedded in the fractional Sobolev space \(W^{s,p}(\Omega )\).
The notion of weak solutions to (1.1) is stated as follows.
Definition 1.1
A function \(u\in L^p(I;W_\mathrm{{loc}}^{1,p}(\Omega )) \cap C(I;L_\mathrm{{loc}}^2(\Omega )) \cap L^{\infty }(I;L_{sp}^{p-1}({\mathbb {R}}^N))\) is a local weak subsolution (super-) to (1.1) if for any closed interval \(I:=[t_1, t_2] \subseteq (0,T)\), there holds that
for every nonnegative test function \(\varphi \in L^p (I;W^{1,p}(\Omega ))\cap W^{1,2}(I;L^2(\Omega ))\) with the property that \(\varphi \) has spatial support compactly contained in \(\Omega \), where
A function u is a local weak solution to (1.1) if and only if u is a local weak subsolution and supersolution.
We now are in a position to state the main contribution of this work. First, we provide the local boundedness of weak solutions in the cases that \(p>\frac{2N}{N+2}\) and \(1<p\le \frac{2N}{N+2}\). For two real numbers, set
Theorem 1.2
(Local boundedness) Let \(p> 2N/(N+2)\) and \(q:=\max \{p,2\}\). Assume that u is a local weak subsolution to (1.1). Let \((x_0,t_0) \in Q_T\), \(R \in (0,1)\) and \(Q_R^- \equiv B_R(x_0) \times (t_0-R^p,t_0)\) such that \({\overline{B}}_R(x_0)\subseteq \Omega \) and \([t_0-R^p, t_0] \subseteq (0,T)\). Then it holds that
where \(\kappa :=1+2/N\) and \(C>0\) only depends on N, p, s and \(\Lambda \).
In the scenario that \(1<p\le 2N/(N+2)\), assuming that the weak subsolution has the following constructions: for \(m>\max \left\{ 2,\frac{N(2-p)}{p}\right\} \), there exists a sequence of \(\{u_k\}_{k\in {\mathbb {N}}}\) whose components are bounded subsolutions of (1.1) fulfilling
and
Theorem 1.3
(Local boundedness) Let \(1<p\le 2N/(N+2)\), \(\kappa =1+2/N\) and \(m>\max \left\{ 2,\frac{N(2-p)}{p}\right\} \). Suppose that \(u\in L_{\mathrm{loc}}^m(Q_T)\) with the properties (1.8) and (1.9) is a local weak subsolution to (1.1). Let \((x_0,t_0) \in Q_T\), \(R \in (0,1)\), and \(Q_R^- \equiv B_R(x_0) \times (t_0-R^p,t_0)\) such that \({\overline{B}}_R(x_0)\subseteq \Omega \) and \([t_0-R^p, t_0] \subseteq (0,T)\). Then it holds that
where \(\beta =N(m-p\kappa )/(N+p)\) and \(C>0\) only depends on N, p, s, m and \(\Lambda \).
Based on the boundedness result (Theorem 1.2), we are able to deduce that the weak solutions are locally Hölder continuous for \(p>2\).
Theorem 1.4
(Hölder continuity) Let \(p>2\). Assume that u is a local weak solution to (1.1). Let \((x_0,t_0) \in Q_T\), \(R \in (0,1)\) and \(Q_R \equiv B_R(x_0) \times (t_0-R^p,t_0+R^p)\) such that \({\overline{Q}}_R\subseteq Q_T\). Then there is a constant \(\alpha \in (0,p/(p-1))\) such that for every \(\rho \in (0,R/2)\),
with some \(d\in (0,1)\) and \(C\ge 1\) depending on N, p, s and \(\Lambda \).
The paper is organized as follows. In Sect. 2, we collect some notations and auxiliary inequalities. Necessary energy estimates are showed in Sect. 3. Sections 4 and 5 are devoted to proving the local boundedness and Hölder regularity of weak solutions, respectively.
2 Preliminaries
In this section, we first give some notations for clarity and then provide some important inequalities to be used later.
2.1 Notation
Let \(B_\rho (x)\) be the open ball with radius \(\rho \) and centred at \(x\in {\mathbb {R}}^N\). We denote the parabolic cylinders by \(Q_{\rho ,r}(x, t):=B_\rho (x) \times (t-r,t+r)\), \(Q_\rho (x,t):=Q_{\rho ,\rho ^p}(x,t)=B_\rho (x) \times (t-\rho ^p,t+\rho ^p)\) and \(Q_\rho ^-(x,t):=Q_{\rho ,\rho ^p}^-(x, t)=B_\rho (x) \times (t-\rho ^p,t)\) with \(r,\rho >0\) and \((x,t) \in {\mathbb {R}}^{N} \times (0,T)\). If not important, or clear from the context, we simply write these symbols by \(B_\rho =B_\rho (x)\), \(Q_{\rho ,r}=Q_{\rho ,r}(x,t)\), \(Q_\rho =Q_\rho (x,t)\) and \(Q_\rho ^-=Q_\rho ^-(x,t)\). Moreover, for \(g\in L^1(V)\), we denote the integral average of g by
Define
for any \(a, b \in {\mathbb {R}}\). We also use the notation
It is worth mentioning that the constant C represents a general positive constant which may differ from each other.
Next, we will show several fundamental but very useful Sobolev inequalities. The similar results can be found in [1, 9]. For the sake of readability and completeness, we give the proof of the last two lemmas.
2.2 Sobolev Inequalities
Lemma 2.1
Let \(1 \le p\), \(\ell \le q<\infty \) satisfy \(\frac{N}{p}-\frac{N}{q}\le 1\) and
with \(\theta \in (0,1)\). Then there exists a constant \(C>0\) only depending on \(N, p, q, \ell \) such that
for all \(u \in W^{1,p}(B_1) \cap L^\ell (B_1)\).
Lemma 2.2
Let \(1<p<N\). Then for every \(u\in W^{1,p}(B_1)\), there holds that
where \(C>0\) only depends on N and p.
Lemma 2.3
Let \(0<t_1<t_2\) and \(p\in (1,\infty )\). Then for every
it holds that
where \(C>0\) only depends on p and N.
Proof
We divide the proof into the following two cases.
Case 1 \(1<p<N\). Applying Lemma 2.2 and Hölder inequality, we infer that
By the scaling argument, we get
Case 2: \(p\ge N\). We can easily find that
and
with \(\theta =\frac{N}{N+2}\). Thus by Lemma 2.1, it follows that
for any \(t\in (t_1,t_2)\). Using the rescaling argument, we have
for any \(t\in (t_1,t_2)\). Integrating the above inequality over \((t_1,t_2)\), we get
\(\square \)
Lemma 2.4
Let \(0<t_1<t_2\) and \(p\in (1,\infty )\). Then for every
it holds that
where \(C>0\) only depends on p and N.
Proof
It follows from Lemma 2.1 that
for all \(t\in (t_1,t_2)\), where \(C>0\) only depends on p and N. Using the rescaling argument, we have
Integrating the above inequality over \((t_1,t_2)\), we get
\(\square \)
3 Energy Estimates
In this section, we will establish the Caccioppoli inequality and Logarithmic form inequality for Eq. (1.1). The first step of the proof should be the regularization procedure with respect to the time variable, which can be performed by straightforward adaptation of standard reasonings as used in [5, 12, 23]. We omit this step here.
Lemma 3.1
(Caccioppoli-type inequality) Let \(p>1\) and u be a local subsolution to (1.1). Let \(B_r \equiv B_r(x_0)\) satisfy \({\overline{B}}_r \subseteq \Omega \) and \(0<\tau _1<\tau _2\), \(\ell >0\) satisfy \(\left[ \tau _1-\ell , \tau _2\right] \subseteq (0,T).\) For any nonnegative functions \(\psi \in C_0^\infty (B_r)\) and \(\eta \in C^\infty ({\mathbb {R}})\) such that \(\eta (t) \equiv 0\) if \(t \le \tau _1-\ell \) and \(\eta (t) \equiv 1\) if \(t \ge \tau _1\), it holds that
where \(w_+(x,t):=\left( u(x,t)-k\right) _+\) and \(C>0\) depends on \(N,p,s,\Lambda \).
Proof
Taking \(\varphi (x,t):=w_+(x,t)\psi ^p(x)\eta ^2(t)=\left( u(x,t)-k\right) _+\psi ^p(x)\eta ^2(t)\) as a test function in (1.7), for \(s\in \left[ \tau _1,\tau _2\right] \), we can obtain
Then we are going to estimate \(I_1, I_2, I_3\) and \(I_4\). First, we evaluate
where in the last line we note that \(\eta (\tau _1-\ell )=0\) and \(\eta (s)=1\) when \(s\ge \tau _1\). We proceed with estimating \(I_2\). By Young’s inequality with \(\varepsilon \), it yields that
with \(\varepsilon =\frac{1}{2}\). As the same proof of Lemma 3.3 in [12], we have
and
Merging the estimates on (3.2)–(3.5), we get
which leads to the desired result. \(\square \)
From the forthcoming lemma, one can interpret the reason why we only derive the Hölder continuity of weak solutions in the case \(p>2\). For the subquadratic case, we at present cannot show a similar logarithmic estimate.
Lemma 3.2
(Logarithmic estimates) Let \(p>2\) and u be a local weak solution to (1.1). Let \(B_r\equiv B_r(x_0)\) and \((x_0,t_0)\in Q_{T}, T_0>0\), \(0<r\le R/2\). We also denote \({\tilde{Q}} \equiv B_R(x_0) \times (t_0-2T_0, t_0+2T_0)\) such that \({\overline{B}}_{R}(x_0) \subseteq \Omega \) and \(\left[ t_0-2T_0,t_0+2T_0\right] \subseteq (0,T)\). If \(u\in L^\infty ({\tilde{Q}})\) and \(u\ge 0\) in \({\tilde{Q}}\), then for any \(d>0\), it holds that
where \(C>0\) depends on N, p, s and \(\Lambda \).
Proof
Let \(\psi \in C_0^\infty (B_{3r/2})\) and \(\eta \in C_0^\infty (t_0-2T_0,t_0+2T_0)\) fulfil
and
Choosing \(\varphi (x,t):=(u(x,t)+d)^{1-p} \psi ^p(x) \eta ^2(t)\) to test the weak formulation of (1.1), we get
It follows from the proof of [12, Lemma 3.5] that
and
By Young’s inequality with \(\varepsilon \), we estimate the integral \(I_2\) as
with \(\varepsilon \) satisfying that \(\varepsilon <p-1\). Combining with (3.7)–(3.10), we can obtain the Logarithmic estimates. \(\square \)
Next, we will give a corollary of Lemma 3.2, which plays a crucial role in obtaining the Hölder continuity.
Corollary 3.3
Let \(p>2\) and u be a local weak solution to (1.1). Let \(B_r\equiv B_r(x_0)\) and \((x_0,t_0)\in Q_{T}, T_0>0\), \(0<r\le R/2\). We also denote \({\tilde{Q}} \equiv B_R(x_0) \times (t_0-2T_0, t_0+2T_0)\) such that \({\overline{B}}_{R}(x_0) \subseteq \Omega \) and \(\left[ t_0-2T_0,t_0+2T_0\right] \subseteq (0,T)\). Suppose that \(u\in L^\infty ({\tilde{Q}})\) and \(u\ge 0\) in \({\tilde{Q}}\). Let \(a,d>0\), \(b>1\) and define
Then it holds that
where \(C>0\) depends on N, p, s and \(\Lambda \).
Proof
By the Poincaré inequality (see for example [17, p. 276]), it yields that
for any \(t\in \left( t_0-T_0,t_0+T_0\right) \). Integrating the above inequality over \(\left( t_0-T_0,t_0+T_0\right) \) leads to
where \(C>0\) depends only on N, p. Observing that v is a truncation function of the sum of a constant and \(\log (u+d)\), which gives that
We can get the results from Lemma 3.2 along with (3.12) and (3.13). \(\square \)
4 Local Boundedness
In this part, we are ready to study the local boundedness of weak solutions. To this end, we first introduce some notations. For \(\sigma \in [1/2,1)\), set
and
Denote
with
Let
We now provide a Caccioppoli-type inequality in a special cylinder, which leads to the recursive inequalities (see Lemmas 4.2 and 4.3).
Lemma 4.1
Let \(p>1\) and u be a local weak subsolution to (1.1). Let \((x_0, t_0) \in Q_T\), \(0<r<1\) and \(Q_r^-=B_r(x_0) \times (t_0-r^p,t_0)\) such that \({\overline{B}}_{r}(x_0) \subseteq \Omega \) and \(\left[ t_0-r^p,t_0\right] \subseteq (0,T)\). Suppose q is a parameter satisfying \(q\ge \max \{p,2\}\), it holds that
where \(C>0\) only depends on \(N,p,s,\Lambda \) and q.
Proof
First we give a trivial but very useful inequality
where \(0\le \tau < q\). We take the cut-off functions \(\psi _j \in C_0^\infty ({\tilde{B}}_j)\) and \(\eta _j \in C_0^\infty ({\tilde{\Gamma }}_j)\) such that
and
Let \(r=r_j\), \(\tau _2=t_0\), \(\tau _1=t_0-r_{j+1}^p\) and \(\ell ={\tilde{r}}_j^p-r_{j+1}^p\) in Lemma 3.1. Then we arrive at
Using (4.2) and the definition of \(\psi _j\), we estimate \(I_1\) as
Analogous to the proof of Lemma 4.1 in [12], we have
and
By virtue of the fact that \(\psi _j\equiv 1\) in \(B_{j+1}\), \(\eta _j \equiv 1\) in \(\Gamma _{j+1}\) and (1.3), merging inequalities (4.3)–(4.7), we get
After rearrangement, we get the desired result. \(\square \)
The following two lemmas are the consequences of Lemmas 2.3 and 4.1 .
Lemma 4.2
Let \(p>2N/(N+2)\) and \(\max \{p,2\}\le q<p(N+2)/N\). Let \((x_0,t_0) \in Q_T\), \(0<r<1\) and \(Q_r^-=B_r(x_0) \times (t_0-r^p,t_0)\) such that \({\overline{B}}_{r}(x_0) \subseteq \Omega \) and \(\left[ t_0-r^p,t_0\right] \subseteq (0,T)\). Then for a local subsolution u to (1.1), we infer that
for \(j\in {\mathbb {N}}\), where \(b:=(1+p/N)(N+p+q)\), \(\kappa :=1+2/N\) and \(C>0\) only depends on \(N,p,s,\Lambda \) and q.
Proof
Since \(q<p\kappa \), it follows from Hölder inequality that
From (4.2), we can get
and
Now by (4.11), Lemmas 2.3 and 4.1 we estimate
with \(b=(1+p/N)(N+p+q)\). Combining (4.9), (4.10) and (4.12), we get the desired result. \(\square \)
Lemma 4.3
Let \(1<p\le 2N/(N+2)\) and \(m>\max \left\{ 2,\frac{N(2-p)}{p}\right\} \). Let \((x_0,t_0) \in Q_T\), \(0<r<1\) and \(Q_r^-=B_r(x_0) \times (t_0-r^p,t_0)\) such that \({\overline{B}}_{r}(x_0) \subseteq \Omega \) and \(\left[ t_0-r^p,t_0\right] \subseteq (0,T)\). Suppose that \(u\in L_{\mathrm{loc}}^\infty (Q_T)\) is a local weak subsolution to (1.1). Then for any \(j\in {\mathbb {N}}\), we have
where \(b:=(1+p/N)(N+p+m)\), \(\kappa :=1+2/N\) and \(C>0\) only depends on N, p, s, m and \(\Lambda \).
Proof
Based on the assumptions, we have
By utilizing Lemma 2.3, Lemma 4.1 with \(q=m\) and the inequality (4.11) with \(q=m\), it yields that
with \(b:=(1+p/N)(N+p+m)\). Thus combining (4.14) and (4.15), we get the desired inequality (4.13). \(\square \)
Remark 4.4
In Lemma 4.3, the quantity \(\sigma ^{\frac{p(N+p)}{N}}\) can be removed, since \(\sigma \in [1/2,1)\). In addition,
Thus, we can get
Next, we introduce an analysis lemma which will be used later.
Lemma 4.5
[9, Lemma 4.1] Let \(\{Y_j\}_{j=0}^\infty \) be a sequence of positive numbers such that
for some constants K, \(b>1\) and \(\delta >0\). Then we have \(\lim _{j\rightarrow \infty }Y_j=0\).
Finally, we end this section by proving the results of local boundedness.
Proof of Theorem 1.2
Let \(r=R\), \(\sigma =\frac{1}{2}\), then \(r_j=\frac{R}{2}+2^{-j-1}R\). Fix \(q=\max \{2,p\}\). We denote
Supposing \({\tilde{k}}\ge 1\) and recalling \(r<1\), we derive from Lemma 4.2 that
where \(b:=(1+p/N)(N+p+q)\), \(\kappa :=1+2/N\) and \(C>0\) only depends on \(N,p,s,\Lambda \). For any \(j\in {\mathbb {N}}\), define \(W_j=Y_j/r^p\). Thus we get
where \({\tilde{k}}\) is such that
with C only depending on N, p, s and \(\Lambda \). Thus along with Lemma 4.5 we can deduce that \(\lim _{j\rightarrow \infty }W_j=0\). Consequently, we get the desired result
We now finish the proof. \(\square \)
Proof of Theorem 1.3
Under the hypotheses (1.8) and (1.9), we may suppose, by a proper approximation procedure that u is locally bounded in advance. Indeed, observe that the approximation subsolutions \(u_k\) are bounded (i.e. \({\text {ess}} \sup _{Q_n^-} (u_k)_+<\infty \)). Therefore, we could substitute u by \(u_k\) to perform the proof below. In other words, the estimate (4.20) below still holds true for u replaced by \(u_k\), which along with (1.8) and (1.9) leads to a k-independent bound on \(u_k\) in \(L^\infty \). We eventually find that u is qualitatively locally bounded via the a.e. convergence of \(u_k\). That is, \({\text {ess}} \sup _{Q_n^-} u_+\) is finite for \(n=0,1,2,3,\ldots \).
Let \(R_0=R/2\) and \(R_n=R/2+\sum _{i=1}^n 2^{-i-1}R\) for \(n\in {\mathbb {N}}^+\), and \(Q_n^-=B_{R_n}(x_0) \times \left( t_0-R_n^p, t_0\right) \). Set
Choosing \(r=R_{n+1}\) and \(\sigma r=R_{n}\), then
We denote
Due to Lemma 4.3, we obtain
with \(b:=(1+p/N)(N+p+m)\) and \(d=(N+p)^2/N\). For any \(j\in {\mathbb {N}}\), define \(W_j=Y_j/R_n^p\). Then we get
According to Lemma 4.5, we can see that
if
To ensure the above inequality, we need choose \({\tilde{k}}\) properly large. Indeed,
Namely,
which implies that
Therefore, we take
which makes (4.18) hold true. Here the constant C only depends on N, p, s, m and \(\Lambda \). Under this choice, it follows from Lemma 4.5 that
Obverse \(m>\max \left\{ 2,\frac{N(2-p)}{p}\right\} \) and \(\kappa =1+2/N\), which indicates that
Now we apply the Young’s inequality with \(\varepsilon \) to (4.19), arriving at
with \(\beta =(m-p\kappa )N/(p+N)\), where we used the fact \(R/2\le R_n<R\). Via the induction argument, we can derive
It is easy to see that the sum on the right-hand side could be revised by a convergent series, provided that we take
Finally, letting \(n\rightarrow \infty \), we deduce that
where \(C>0\) depends only on N, p, s, m and \(\Lambda \). \(\square \)
5 Local Hölder Continuity
In this section, we aim at establishing the Hölder continuity of weak solutions to (1.1) in the case that \(p>2\), based on the local boundedness results. Before verifying this conclusion, we introduce some notations.
Let \(({\overline{x}}_0,{\overline{t}}_0)\in Q_T\) and \(r\in (0,R]\) for some \(R\in (0,1)\). Let also \(\alpha \in (0,\frac{p}{p-1})\) and \(\sigma \in (0,\sigma _0)\) with \(\sigma _0^\frac{p}{p-1}\le \frac{1}{4}\) be two constants to be determined later. Set
and
with C depending on \(N,p,s,\Lambda \). Define
where
Thus, it is easy to obtain
Denote
and
Hence, for \(j\ge 1\), we have
The above inequalities combine with the definitions of \(d_j\) and \(t_j\) gives that
Now we are going to deduce an oscillation reduction on weak solutions.
Lemma 5.1
Let \(p>2\) and u be a local weak solution to (1.1). Set \(\left( {\overline{x}}_0,{\overline{t}}_0\right) \in Q_T\), \(r\in (0,R]\) for some \(R\in (0,1)\) and \(Q_R\equiv B({\overline{x}}_0) \times \left( {\overline{t}}_0-R^p, {\overline{t}}_0+R^p\right) \) such that \({\overline{Q}}_R \subseteq Q_T\). Then
Proof
We prove this Lemma by the induction argument. It follows from Theorem 1.2 and the definition of \(\omega (r_0)\) that (5.5) holds true for \(j=0\). We may assume that (5.5) is valid for each \(i\in \{0,\ldots ,j\}\) with some \(j\ge 0\). Then we devote to proving it holds for \(i=j+1\).
Denote
It is obvious that either
or
must hold. In the case of (5.6), we set \(u_j:=u-\underset{Q_j}{{\text {ess}}\,\inf } u\). In the case of (5.7), we set \(u_j:=\omega (r_j)-\left( u-\underset{Q_j}{{\text {ess}}\,\inf } u\right) \). In all cases, we have
and
Now we provide an important estimate to be used later,
where C only depends on N, p, s, the difference of \(p/(p-1)\) and \(\alpha \). Indeed, it is easy to see the claim is true when \(j=0\). For \(j\ge 1\), we have
where in the last line we used (5.9) and the definition of \(u_j\). Since \(\sigma \le 1/4\) and \(\alpha <p/(p-1)\), we estimate the right-hand side as follows:
Thus, we have proved (5.10) with the constant C depending on N, p, s, the difference of \(p/(p-1)\) and \(\alpha \). Next, we define
It follows from Corollary 3.3 with \(a\equiv \omega (r_j)/2\) and \(b\equiv \exp (k)\) that
where in the last line we used the fact \(4 t_{j+1} \le t_j\) and (5.10). Choosing \(d=\varepsilon \omega (r_j)\) in (5.11), we get
Since \(\alpha <p/(p-1)\), we can verify
Thus, we arrive at
where C depends on \(N,p,s,\Lambda \), the difference of \(p/(p-1)\) and \(\alpha \). Following the calculation in pp. 37–38 in [12], there holds that
where \({\overline{C}}>0\) depends on \(N,p,s,\Lambda \), the difference of \(p/(p-1)\) and \(\alpha \).
In what follows, we will proceed by a suitable iteration to infer the desired oscillation decay over the domain \(Q_{j+1}\). For any \(i=\) \(0,1,2, \ldots \), we define
Then we take the cut-off function \(\psi _i \in C_0^\infty ({\tilde{B}}^i)\) and \(\eta _i \in C_0^\infty ({\tilde{\Gamma }}_i)\) such that
and
Define
Taking \(\ell =\theta _i-\theta _{i+1}, \tau _1={\overline{t}}_0-\theta _{i+1}\) and \(\tau _2={\overline{t}}_0+\theta _{i+1}\) in Lemma 3.1, we get
We estimate \(I_1\) as
where we used the properties of \(\psi _i\). As the computations in pp. 39–40 in [12], we derive
and
Since \(\psi _i\equiv 1\) in \(B^{i+1}\), we can deduce that
From (5.3), we get
Combining (5.13) with (5.14) gives that
where \(A_i\) is denoted by
In view of Lemma 2.4, we can see
Merging (5.15) and (5.16) leads to
We can readily get the recursive inequality
where \({\tilde{C}}\) depends on \(N,p,s,\Lambda \), the difference of \(p/(p-1)\) and \(\alpha \). Set
Then we take
where we need to note that \(\sigma _0\) only depends on p. Utilizing the definition of \(A_i\), we obtain
It follows from Lemma 4.5 that
which implies that
Thereby, recalling (5.1) and the definition of \(u_j\), we have
Now, we pick \(\alpha \in (0,p/(p-1))\) such that
which together with (5.17) ensures that
Now we finish the proof. \(\square \)
Proof of Theorem 1.4
Let \(p>2\). Suppose that u is a local weak solution of (1.1). Let \((x_0,t_0) \in Q_T\), \(R \in (0,1)\) and \(Q_R \equiv B_R(x_0) \times (t_0-R^p,t_0+R^p)\) such that \({\overline{Q}}_R\subseteq Q_T\). Taking \(r=R\) in Lemma 5.1 we can get
with \(\alpha <p/(p-1)\), \(\sigma <1/4\), where \(C\ge 1\) depends on \(N,p,s,\Lambda ,\) and
For every \(\rho \in (0,R/2]\), we have \(\rho \in \left( r_{j_0+1}, r_{j_0}\right] \) for \(j_0 \in {\mathbb {N}}\). Choosing \(d=[C\omega (R/2)]^{2-p}\), it follows that \(Q_{\rho , d \rho ^p} \subseteq Q_{j_0}\). By applying (5.18), we get
We can obtain the Hölder continuity from the above inequality along with (5.19). \(\square \)
References
Adams, R.A.: Sobolev Spaces. Academic Press, New York (1975)
Barlow, M.T., Bass, R.F., Chen, Z., Kassmann, M.: Non-local Dirichlet forms and symmetric jump processes. Trans. Am. Math. Soc. 361(4), 1963–1999 (2009)
Biagi, S., Dipierro, S., Valdinoci, E., Vecchi, E.: Mixed local and nonlocal elliptic operators: regularity and maximum principles. Commun. Partial Differ. Equ. (2021). https://doi.org/10.1080/03605302.2021.1998908
Biagi, S., Dipierro, S., Valdinoci, E., Vecchi, E.: Semilinear elliptic equations involving mixed local and nonlocal operators. Proc. R. Soc. Edinburgh Sect. A 151(5), 1611–1641 (2021)
Brasco, L., Lindgren, E., Strömqvist, M.: Continuity of solutions to a nonlinear fractional diffusion equation. J. Evol. Equ. (2021). https://doi.org/10.1007/s00028-021-00721-2
Caffarelli, L., Chan, C., Vasseur, A.: Regularity theory for parabolic nonlinear integral operators. J. Am. Math. Soc. 24(3), 849–869 (2011)
Chen, Z., Kumagai, T.: A priori Hölder estimate, parabolic Harnack principle and heat kernel estimates for diffusions with jumps. Rev. Mat. Iberoam. 26(2), 551–589 (2010)
Chen, Z., Kim, P., Song, R., Vondraček, Z.: Boundary Harnack principle for \(\Delta +\Delta ^{\alpha /2}\). Trans. Am. Math. Soc. 364(8), 4169–4205 (2012)
DiBenedetto, E.: Degenerate Parabolic Equations. Universitext. Springer, New York (1993)
Di Castro, A., Kuusi, T., Palatucci, G.: Local behavior of fractional \(p\)-minimizers. Ann. Inst. H. Poincaré Anal. Non Linéaire 33, 1279–1299 (2016)
Di Nezza, E., Palatucci, G., Valdinoci, E.: Hitchhiker’s guide to the fractional Sobolev spaces. Bull. Sci. Math. 136, 521–573 (2012)
Ding, M., Zhang, C., Zhou, S.: Local boundedness and Hölder continuity for the parabolic fractional \(p\)-Laplace equations. Calc. Var. Partial Differ. Equ. 60, 38 (2021)
Dipierro, S., Proietti Lippi, E., Valdinoci, E.: Linear theory for a mixed operator with Neumann conditions. arXiv:2006.03850
Dipierro, S., Lippi, E. Proietti., Valdinoci, E.: (Non)local logistic equations with Neumann conditions. arXiv:2101.02315
Dipierro, S., Ros-Oton, X., Serra, J., Valdinoci, E.: Non-symmetric stable operators: regularity theory and integration by parts. arXiv:2012.04833
Dipierro, S., Valdinoci, E.: Description of an ecological niche for a mixed local/nonlocal dispersal: an evolution equation and a new Neumann condition arising from the superposition of Brownian and Lévy processes. Phys. A 575, 126052 (2021)
Evans, L.C.: Partial Differential Equations, Graduate Studies in Mathematics, vol. 19. American Mathematical Society, Providence (1998)
Felsinger, M., Kassmann, M.: Local regularity for parabolic nonlocal operators. Commun. Partial Differ. Equ. 38(9), 1539–1573 (2013)
Foondun, M.: Heat kernel estimates and Harnack inequalities for some Dirichlet forms with non-local part. Electron. J. Probab. 14(11), 314–340 (2009)
Garain, P., Kinnunen, J.: On the regularity theory for mixed local and nonlocal quasilinear elliptic equations. arXiv:2102.13365
Garain, P., Kinnunen, J.: Weak Harnack inequality for a mixed local and nonlocal parabolic equation. arXiv:2105.15016
Kim, Y.: Nonlocal Harnack inequalities for nonlocal heat equations. J. Differ. Equ. 267(11), 6691–6757 (2019)
Kinnunen, J., Lindqvist, P.: Pointwise behaviour of semicontinuous supersolutions to a quasilinear parabolic equation. Ann. Mat. Pura Appl. 185(3), 411–435 (2006)
Mazón, J.M., Rossi, J.D., Toledo, J.: Fractional \(p\)-Laplacian evolution equations. J. Math. Pures Appl. 105(6), 810–844 (2016)
Molica Bisci, G., Radulescu, V.D., Servadei, R.: Variational Methods for Nonlocal Fractional Problems, Encyclopedia of Mathematics and Its Applications, vol. 162. Cambridge University Press, Cambridge (2016)
Strömqvist, M.: Local boundedness of solutions to nonlocal parabolic equations modeled on the fractional \(p\)-Laplacian. J. Differ. Equ. 266, 7948–7979 (2019)
Vázquez, J.L.: The Dirichlet problem for the fractional \(p\)-Laplacian evolution equation. J. Differ. Equ. 260(7), 6038–6056 (2016)
Vázquez, J.L.: The evolution fractional \(p\)-Laplacian equation in \({\mathbb{R}}^N\). Fundamental solution and asymptotic behaviour. Nonlinear Anal. 199, 112034 (2020)
Vázquez, J.L.: The fractional \(p\)-Laplacian evolution equation in \({\mathbb{R}}^N\) in the sublinear case. Calc. Var. Partial Differ. Equ. 60, 140 (2021)
Acknowledgements
The authors wish to thank the anonymous reviewers for many valuable comments and suggestions to improve the manuscript. This work was supported by the National Natural Science Foundation of China (No. 12071098).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Fang, Y., Shang, B. & Zhang, C. Regularity Theory for Mixed Local and Nonlocal Parabolic p-Laplace Equations. J Geom Anal 32, 22 (2022). https://doi.org/10.1007/s12220-021-00768-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12220-021-00768-0