Abstract
Local Hölder regularity is established for certain weak solutions to a class of parabolic fractional p-Laplace equations with merely measurable kernels. The proof uses DeGiorgi’s iteration and refines DiBenedetto’s intrinsic scaling method. The control of a nonlocal integral of solutions in the reduction of oscillation plays a crucial role and entails delicate analysis in this intrinsic scaling scenario. Dispensing with any logarithmic estimate and any comparison principle, the proof is new even for the linear case.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
We are interested in the local Hölder regularity of weak solutions to a class of parabolic equations involving a fractional p-Laplacian type operator:
where \(E_T=E\times (0,T]\) for some open set \(E\subset \mathbb {R}^N\) and some \(T>0\), and the nonlocal operator \(\mathscr {L}\) is defined by
for some \(p>1\). Here, \(\mathrm{P.V.}\) denotes the principle value of the integral, whereas the kernel \(K:\mathbb {R}^N\times \mathbb {R}^N\times (0,T]\rightarrow [0,\infty )\) is a measurable function satisfying the following condition uniformly in t:
for some positive \(C_o\), \(C_1\) and \(s\in (0,1)\).
Throughout this note, the parameters \(\{s, p, N, C_o, C_1\}\) are termed the data, and we use \(\varvec{\gamma }\) as a generic positive constant in various estimates that can be determined by the data only.
The formal definition of weak solution to (1.1)–(1.3) and notation can be found in Sect. 1.2. We proceed to present our main result as follows.
Theorem 1.1
Let u be a locally bounded, local, weak solution to (1.1)–(1.3) in \(E_T\) with \(p>1\).Then u is locally Hölder continuous in \(E_T\). More precisely, there exist constants \(\varvec{\gamma }>1\) and \(\beta \in (0,1)\) that can be determined a priori only in terms of the data, such that for any \(0<\varrho<R<\widetilde{R}\), there holds
provided the cylinders \((x_o,t_o)+Q_{R}(\varvec{\omega }^{2-p})\subset (x_o,t_o)+Q_{\widetilde{R}}\) are included in \(E_T\), where
Remark 1.1
Theorem 1.1 has been formulated independent of any initial/boundary data. While local, the oscillation estimate bears global information via the tail of u; see (1.9). In particular, a solution is allowed to grow at infinity. Whereas if u is globally bounded in \(\mathbb {R}^N\times (0,T)\), then \(\varvec{\omega }\) can be taken as the global bound. This occurs if, for instance, proper initial/boundary data are prescribed, cf. [4, 6]. In addition, if u is globally bounded in \(\mathbb {R}^N\times (-\infty ,T)\), then u is, a fortiori, a constant by the oscillation estimate.
Remark 1.2
Theorem 1.1 continues to hold for more general structures. For instance, one can consider the kernel satisfying
Also, proper lower order terms can be considered, cf. [8]. However, we will concentrate on the actual novelty and leave possible generalizations to the motivated reader.
Remark 1.3
Our approach only has a minimal requirement regarding fractional calculus; all relevant results are collected in Appendix A. However, DiBenedetto’s intrinsic scaling method is quite involved; monographs [9, 11, 25] provide a good account of the method. It would be instructive to first practice our arguments in the elliptic setting or in the linear parabolic setting.
1.1 Novelty and significance
The nonlocal elliptic operator \(\mathscr {L}\) as in (1.2) with a kernel like (1.3), especially when \(p=2\), has been a classical topic in Probability, Potential theory, Harmonic Analysis, etc. In addition, nonlocal partial differential equations arise from continuum mechanics and phase transition, from population dynamics, and from optimal control and game theory. We refer to [5, 14] for a source of motivations and applications.
Local regularity for the nonlocal elliptic operator with merely measurable kernels is well studied, cf. [8, 12, 13, 17, 22]—just to mention a few. In [12, 13], localization techniques are developed in order to establish Hölder regularity and Harnack’s inequality for the elliptic operator. A logarithmic estimate plays a key role in [12, 13]. Whereas [8] further improves these results to functions in certain DeGiorgi classes and the logarithmic estimate is dispensed with.
The parabolic nonlocal problem (1.1) has witnessed a growing interest recently; see [2,3,4, 6, 15, 16, 18, 19, 21, 23, 24, 26, 27]—just to mention a few. Coming to the local regularity, while the case \(p=2\) has been subject to extensive studies, the case \(p\ne 2\) is largely open. Local regularity of supersolutions is studied in [2, 3]. Some local boundedness estimates are reported in [15, 23]. Meanwhile, it is tried in [15] to adapt techniques of [12] and to show Hölder regularity for (1.1) with \(p>2\). However, we are unable to verify [15, (5.23)], which is based on a logarithmic estimate. When the kernel K(x, y, t) is exactly \( 2|x-y|^{-N-sp}\) for \(p\ge 2\), explicit Hölder exponents are obtained in [4].
Our contribution lies in establishing Hölder regularity for the parabolic fractional p-Laplace type equation with merely measurable kernels for all \(p>1\). The approach is structural, in the sense that we dispense with any kind of comparison principle and do not reply on solving PDEs. More generally, we find that the Hölder regularity is in fact encoded in a family of energy estimates in Corollary 2.1, and tools like logarithmic estimates or exponential change of variables play no role in our arguments.Footnote 1 This differs from the Moser approach, which keeps using the PDE with different testing functions. As such, the arguments are new even for \(p=2\) and hold the promise of a wider applicability, for instance, in Calculus of Variations.
Unlike the elliptic operator or the parabolic operator with \(p=2\), the local behavior of a solution to the parabolic p-Laplacian is markedly different: it has to be read in its own intrinsic geometry. This is the guiding idea in the local operator theory, cf. [7, 9,10,11, 25]. In terms of oscillation estimates, this idea leads to the construction of geometric sequences \(\{R_i\}\) and \(\{\varvec{\omega }_{i}\}\) connected by the intrinsic relation
The nonlocal theory developed here is no exception. However, the nonlocal character of (1.1) needs to be carefully handled in this intrinsic scaling scenario. A new component brought by the nonlocality of the operator is a proper control of the so-called tail—a nonlocal integral of the solution (see (1.9)). Precisely, we have
where \( \varvec{\mu }_i^{\pm }\) denotes the supremum/infimum of u over \(Q_{R_{i}}(\varvec{\omega }_{i}^{2-p})\). In other words, the nonlocal tail is controlled by the local oscillation, if the intrinsic relation (1.4) is verified. The tail estimate (1.5) in turn allows us to reduce the oscillation in the next step, and so on. This induction procedure can be illustrated by
The local regularity theory for the nonlocal parabolic problem (1.1) with \(p\ne 2\) is still at its inception. We believe the techniques developed in this note are flexible enough and provide a handy toolkit that can be used to fruitfully attack more general nonlocal parabolic equations.
1.2 Definitions and notation
1.2.1 Function spaces
For \(p>1\) and \(s\in (0,1)\), we introduce the fractional Sobolev space \(W^{s,p}(\mathbb {R}^N)\) by
which is endowed with the norm
Similarly, the fractional Sobolev space \(W^{s,p}(E)\) for a domain \(E\subset \mathbb {R}^N\) can be defined. Moreover, we denote
These spaces admit imbedding into proper Lebesgue spaces; we collect some in Appendix A.
1.2.2 Notion of weak solution
A measurable function \(u:\,\mathbb {R}^N\times (0,T]\rightarrow \mathbb {R}\) satisfying
is a local, weak sub(super)-solution to (1.1)–(1.3), if for every compact set \(\mathcal {K}\subset E\) and every sub-interval \([t_1,t_2]\subset (0,T]\), we have
and
where
for all non-negative testing functions
A function u that is both a local weak sub-solution and a local weak super-solution to (1.1)–(1.3) is a local weak solution.
Remark 1.4
As we are developing a local theory, the function space in (1.8) can be taken smaller. Namely, the notion of \(W_o^{s,p}(\mathcal {K})\) can be replaced by functions \(\varphi (\cdot , t)\in W^{s,p}(\mathbb {R}^N)\) with a compact support in \(\mathcal {K}\) for a.e. t.
Remark 1.5
To ensure the convergence of the global integral in (1.7), it suffices to weaken the \(L^\infty \) norm appearing in the condition (1.6) by the \(L^1\) norm. However, in deriving the energy estimate of Proposition 2.1, the condition (1.6) is needed already.
1.2.3 Some notation
Throughout this note, we will use \(K_\varrho (x_o)\) to denote the ball of radius \(\varrho \) and center \(x_o\) in \(\mathbb {R}^N\), and the symbols
to denote (backward) cylinders with the indicated positive parameters. When the context is unambiguous, we will omit the vertex \((x_o,t_o)\) from the symbols for simplicity. When \(\theta =1\), it is also omitted.
A nonlocal integral of u—termed the tail of u—inevitably appears in the theory, which we define as
For any \(Q(R,S)\subset E_T\), the finiteness of this tail is guaranteed by (1.6).
2 Energy estimates
This section is devoted to energy estimates satisfied by local weak sub(super)-solutions to (1.1)–(1.3). We first introduce, for any \(k\in \mathbb {R}\), the truncated functions
In what follows, when we state “u is a sub(super)-solution...” and use \(``\pm "\) or \(``\mp "\) afterwards, we mean the sub-solution corresponds to the upper sign and the super-solution corresponds to the lower sign in the statement.
Proposition 2.1
Let u be a local weak sub(super)-solution to (1.1)–(1.3) in \(E_T\). There exists a constant \(\varvec{\gamma } (C_o,C_1,p)>0\), such that for all cylinders \(Q(R,S) \subset E_T\), every \(k\in \mathbb {R}\), and every non-negative, piecewise smooth cutoff function \(\zeta (\cdot ,t)\) compactly supported in \( K_{R} \) for all \(t\in (t_o-S,t_o)\), there holds
Here, we have denoted \(w=u-k\) for simplicity.
Proof
We will only deal with the case of sub-solution as the other case is similar. Using \(\varphi =w_+\zeta ^p\) as a testing function in the weak formulation modulo a proper time mollification (cf. Appendix B), the last terms on the right/left-hand side of the energy estimate are rather standard. We only treat the integral resulting from the fractional diffusion part, which, due to the support of \(\zeta \) and symmetry of the integrand, can be split into two parts, that is,
where we have used \(\textrm{d}\mu =K(x,y,t)\textrm{d}y\textrm{d}x\textrm{d}t\) for simplicity.
Let us manipulate the first integral, which is the leading term. To this end, we denote \(A_k=\big \{u(\cdot , t)>k\big \}\cap K_R\) for fixed \(t\in (t_o-S,t_o)\). Observe that if \(x\in A_k\) while \(y\in K_R{\setminus } A_k\), one obtains
for some proper \(c=c(p)\). Whereas if \(x,y\in A_k\), we claim that
To prove the claim, we may assume that \(u(x,t)\ge u(y,t)\) due to symmetry and write
If \(\zeta (x,t)\ge \zeta (y,t)\), the above display (2.3) is estimated below by
and hence the claim follows. If, instead \(\zeta (x,t)<\zeta (y,t)\), then (2.3) can be written as
To proceed, we need an elementary inequality:
This simply follows from the mean value theorem and Young’s inequality:
We may apply (2.4) with \(a=\zeta (y,t)\), \(b=\zeta (x,t)\) and
to obtain
Combining the last two estimates in (2.3), we obtain (2.2) in the case \(\zeta (x,t)<\zeta (y,t)\) also.
Employing (2.1) and (2.2) and properly adjusting c if necessary, the first integral \(I_1\) is estimated by
Now let us treat the second integral \(I_2\), which yields the only nonlocal integral in the energy estimate. Indeed, we first estimate
As a result, we may estimate by the condition (1.3) on the kernel K,
Note that the finiteness of the above nonlocal integral is guaranteed by (1.6). This term will evolve into the tail term (1.9) in the forthcoming theory.
Finally, we can put all these estimates together and use the condition (1.3) on the kernel K to conclude. \(\square \)
The above energy estimate can be written in \(K_R\times (t_o- S, t)\) for any \(t\in (t_o- S, t_o)\). As usual, this will lead to an \(L^\infty \) estimate in the time variable on the left, due to the arbitrariness of t. Further, by choosing a proper cutoff function \(\zeta \), we derive the following two types of energy estimates from Proposition 2.1, which encode all the information needed to show Theorem 1.1.
Corollary 2.1
Let u be a local weak sub(super)-solution to (1.1)–(1.3) in \(E_T\). There exists a constant \(\varvec{\gamma } (C_o,C_1,p)>0\), such that for all cylinders \(Q(r,\tau )\subset Q(R,S) \subset E_T\), and every \(k\in \mathbb {R}\), there holds
and
Here, we have denoted \(w=u-k\) for simplicity.
3 Preliminary tools
In this section, we collect the main modules of the proof of Hölder regularity. An important feature is that the tail term appears in these modules via an either-or form. This feature clarifies the role of the tail and greatly facilitates the delicate intrinsic scaling arguments to be unfolded in the next two sections. To streamline, we derive them from the energy estimates in Proposition 2.1. Nevertheless, it will be clear from their proofs that Corollary 2.1 actually suffices. We also stress that the arguments in this section are given in a unified fashion for all \(p>1\).
Throughout this section, let \(\mathcal {Q}:=K_R(x_o)\times (T_1,T_2]\) be a cylinder included in \(E_T\). We introduce numbers \(\varvec{\mu }^{\pm }\) and \(\varvec{\omega }\) satisfying
The first result concerns a DeGiorgi type lemma. For simplicity, we omit the vertex \((x_o,t_o)\) from \(Q_\varrho (\theta )\).
Lemma 3.1
Let u be a locally bounded, local weak sub(super)-solution to (1.1)–(1.3) in \(E_T\). For some \( \delta ,\,\xi \in (0,1)\) set \(\theta =\delta (\xi \varvec{\omega })^{2-p}\) and assume \( Q_\varrho (\theta ) \subset \mathcal {Q}\). There exists a constant \(\nu \in (0,1)\) depending only on the data \(\{s, p, N, C_o, C_1\}\) and \(\delta \), such that if
then either
or
Moreover, we have the dependence \(\nu \approx \delta ^q\) for some \( q>1\) depending on p and N.
Proof
It suffices to show the case of super-solution with \(\varvec{\mu }^-=0\). As a restatement,Footnote 2 we will show that there exists \(\nu \), such that if
and if
then
We stress that \(\nu \) only depends on the data \(\{s, p, N, C_o, C_1\}\) and \(\delta \), but independent of \(\xi \).
Upon a translation, we may assume \((x_o,t_o)=(0,0)\) and define for \(n\in \mathbb {N}\cup \{0\}\),
It is helpful to observe that
Introduce the cutoff function \(\zeta \) in \(Q_n\), vanishing outside \(\widehat{Q}_{n}\) and equal to the identity in \(\widetilde{Q}_{n}\), such that
Let us examine the energy estimate of Proposition 2.1 in this setting:
Recalling \(w_-=(u-k_n)_-\), we treat the three terms on the right-hand side of the energy estimate as follows. For the first term, we estimate
where we have defined \(A_n:=\{u<k_n\}\cap Q_n\).
For the second term, we observe that for \(|y|\ge \varrho _n\) and \(|x|\le {\hat{\varrho }}_n\), there holds
consequently, recalling also \(u\ge \varvec{\mu }^-=0\) a.e. in \(\mathcal {Q}\) by assumption, we estimate
In the last line, we have enforced
For the third term, it is quite standard to obtain
Collecting these estimates on the right-hand side of the energy estimate, we arrive at
Now set \(0\le \phi \le 1\) to be a cutoff function in \(\widetilde{Q}_n\), which vanishes outside \(\overline{Q}_n\), equals the identity in \(Q_{n+1}\) and satisfies \(|D\phi |\le 2^n/\varrho \). An application of the Hölder inequality and the Sobolev imbedding (cf. Proposition A.3 with \(d=2^{-n-4}\)) gives that
for some \({\varvec{b}}={\varvec{b}}(p,N)>1\). To obtain the last line, we used the triangle inequality
for some \(c=c(p)\), such that
Plugging this into the second-to-last line and employing the above energy estimate, the last line follows. Notice also from Proposition A.3 there holds
In terms of \( {\varvec{Y}}_n=|A_n|/|Q_n|\), this estimate leads to the recursive inequality
for some generic constant \(\varvec{\gamma }\). Hence, by the fast geometric convergence, cf. [9, Chapter I, Lemma 4.1], there exists a positive constant \(\nu \) depending only on the data, such that \({\varvec{Y}}_n\rightarrow 0\) if we require that \({\varvec{Y}}_o\le \nu \). \(\square \)
The next lemma is a variant of the previous one, involving quantitative initial data.
Lemma 3.2
Let u be a locally bounded, local weak sub(super)-solution to (1.1)–(1.3) in \(E_T\). Let \(\xi \in (0,1)\). There exists a positive constant \(\nu _o\) depending only on the data \(\{s, p, N, C_o, C_1\}\) and independent of \(\xi \), such that if
then either
or
provided the cylinders are included in \(\mathcal {Q}\).
Proof
Assume \((x_o,t_o)=(0,0)\). It suffices to show the case of super-solutions with \(\varvec{\mu }^-=0\). Let us first examine the energy estimate of Proposition 2.1 in \(Q(R,S)\equiv K_{\varrho }\times (0,\theta \varrho ^{sp})\) for some \(\theta \) to be determined. Note that the time level \(t_o-S\) in Proposition 2.1 corresponds to \(t=0\) here. Let \(\zeta (x)\) be a time independent, piecewise smooth, cutoff function in \(K_\varrho \) that vanishes on \(\partial K_\varrho \). If we take the level \(k\le \xi \varvec{\omega }\), the spatial integral at \(t=0\) (i.e. the term at the time level \(t_o-S\) in Proposition 2.1) vanishes due to the assumption that \(u(\cdot , 0)\ge \xi \varvec{\omega }\) a.e. in \(K_{\varrho }\). The term involving \(\partial _t\zeta \) also vanishes since \(\zeta \) is independent of t. As a result, the energy estimate reads
Introduce \(k_n\), \(\varrho _n\), \(\tilde{\varrho }_n\), \(\hat{\varrho }_n\), \(\bar{\varrho }_n\), \(K_n\), \(\widetilde{K}_n\), \(\widehat{K}_n\) and \(\overline{K}_n\) as in Lemma 3.1. The only difference is that the cylinders \(Q_n\), \(\widetilde{Q}_n\), \(\widehat{Q}_n\) and \(\overline{Q}_n\) are now of forward type, i.e. \(Q_n=K_n\times (0,\theta \varrho ^{sp})\), \(\widetilde{Q}_n=\widetilde{K}_n\times (0,\theta \varrho ^{sp})\), \(\widehat{Q}_n=\widehat{K}_n\times (0,\theta \varrho ^{sp})\) and \(\overline{Q}_n=\overline{K}_n\times (0,\theta \varrho ^{sp})\). Note that while shrinking the base balls \(K_n\), \(\widehat{K}_n\), \(\widetilde{K}_n\) and \(\overline{K}_n\) along \(\varrho _n\), the height of the cylinders is fixed. For the piecewise smooth function \(\zeta (x)\) in \(K_n\), we choose it to vanish outside \(\widehat{K}_n\), be equal to 1 in \(\widetilde{K}_n\), and satisfy \(|D\zeta |\le 2^n/\varrho \). By a similar treatment of the right-hand side as in Lemma 3.1, after enforcing
we may obtain that
where \(A_n:=\{u<k_n\}\cap Q_n\).
Then we may proceed as in Lemma 3.1 to obtain the recursive inequality
where \( {\varvec{Y}}_n=|A_n|/|Q_n|\) and the constants \(\varvec{\gamma },\,{\varvec{b}}\) depend only on the data. Hence, by the fast geometric convergence, cf. [9, Chapter I, Lemma 4.1], there exists a positive constant \(\nu _o\) depending only on the data, such that if
then \({\varvec{Y}}_n\rightarrow 0\). To finish the proof, we choose \(\theta =\nu _o(\xi \varvec{\omega })^{2-p}\). \(\square \)
The following lemma propagates measure theoretical information forward in time.
Lemma 3.3
Let u be a locally bounded, local weak sub(super)-solution to (1.1)–(1.3) in \(E_T\). Introduce parameters \(\xi \) and \(\alpha \) in (0, 1). There exist \(\delta \) and \(\varepsilon \) in (0, 1), depending only on the data \(\{s, p, N, C_o, C_1\}\) and \(\alpha \), such that if
then either
or
provided this cylinder is included in \(\mathcal {Q}\). Moreover, we may trace the dependences by \(\varepsilon \approx \alpha \) and \(\delta \approx \alpha ^{p+N+1}\).
Proof
We only show the case of super-solutions with \(\varvec{\mu }^-=0\). Assume \((x_o,t_o)=(0,0)\) without loss of generality. Use the energy estimate in Proposition 2.1 in the cylinder \(Q=K_{\varrho }\times (0,\delta (\xi \varvec{\omega })^{2-p}\varrho ^{sp}]\), with the truncation
Choose a standard non-negative cutoff function \(\zeta (x,t)\equiv \zeta (x)\) independent of time that equals 1 on \(K_{(1-\sigma )\varrho }\) with \(\sigma \in (0,1)\) to be chosen later and vanishes outside \(K_{(1-\frac{1}{2}\sigma )\varrho }\) satisfying \(|D\zeta |\le 4(\sigma \varrho )^{-1}\); in such a case, the energy estimate presents that, for all \(0<t<\delta (\xi \varvec{\omega })^{2-p}\varrho ^{sp}\), there holds
The first term on the right-hand side is estimated by
As for the second term, we observe that for \(x\in {\text {supp}}\zeta \subset K_{(1-\frac{1}{2}\sigma )\varrho }\) and \(y\in \mathbb {R}^N{\setminus } K_{\varrho }\), there holds
therefore, we estimate
Here, in the last line we have enforced
Therefore, combining all these estimates in the energy estimate, we have for all \(0<t<\delta (\xi \varvec{\omega })^{2-p}\varrho ^{sp}\), that
The left-hand side in the last display is estimated from below by
where we have defined
Notice also a simple fact that
Using the last two estimates, together with the energy estimate, we have for all \(0<t<\delta (\xi \varvec{\omega })^{2-p}\varrho ^{sp}\), that
To conclude the proof, we choose various parameters \(\sigma \), \(\delta \) and \(\varepsilon \) to satisfy:
In this way, we obtain
for all \(0<t<\delta (\xi \varvec{\omega })^{2-p}\varrho ^{sp}\), which is a restatement of the desired measure estimate. \(\square \)
The following measure shrinking lemma is usually a delicate part in the theory of the parabolic p-Laplacian. However, the term that involves mixed positive/negative truncations in the energy estimate greatly simplifies the argument. For ease of notation, we omit the vertex \((x_o,t_o)\) from \(Q_\varrho (\theta )\).
Lemma 3.4
Let u be a locally bounded, local weak sub(super)-solution to (1.1)–(1.3) in \(E_T\). Suppose that for some \(\delta \), \(\sigma \) and \(\xi \) in \((0,\tfrac{1}{2})\), there holds
Let \(\theta =\delta (\sigma \xi \varvec{\omega })^{2-p}\). There exists \(\varvec{\gamma }>0\) depending only on the data \(\{s, p, N, C_o, C_1\}\) and independent of \(\{\alpha , \delta , \sigma ,\xi \}\), such that either
or
provided \(Q_{2\varrho }(\theta )\) is included in \(\mathcal {Q}\).
Proof
It suffices to show the case of super-solutions with \(\varvec{\mu }^-=0\). For simplicity, we assume \((x_o,t_o)=(0,0)\). Let us employ the energy estimate of Proposition 2.1 in \(K_{2\varrho }\times (-\theta \varrho ^{sp}, 0]\) with the truncation
and introduce a cutoff function \(\zeta \) in \(K_{ 2\varrho }\) (independent of t) that is equal to 1 in \(K_{\varrho }\) and vanishes outside \(K_{\frac{3}{2}\varrho }\), such that \(|D\zeta |\le 4/\varrho \). Then, we obtain from Proposition 2.1 that
The first term on the right-hand side is estimated by
noticing that \(|K_{2\varrho }|=2^N |K_\varrho |\) in the last line. As for the second term, observe that for \(|x|\le \frac{3}{2}\varrho \) and \(|y|\ge 2\varrho \), there holds
consequently, we estimate
Here, in the last line we have enforced
and used \(|Q_{2\varrho }(\theta )|=2^{N+p}|Q_{\varrho }(\theta )|\). Whereas the third term is standard:
recalling that \(\theta =\delta (\sigma \xi \varvec{\omega })^{2-p}\) and noticing that \(|K_{2\varrho }|=2^N |K_\varrho |\).
Combining the above estimate we see that the right-hand side of the energy estimate is bounded by
provided the tail estimate is enforced. We stress that \(\varvec{\gamma }\) depends only on the data \(\{s, p, N, C_o, C_1\}\) and is independent of \(\{\alpha , \delta , \sigma ,\xi \}\).
The left-hand side is estimated by extending the integrals over smaller sets and by using the given measure theoretical information:
Combining these estimates and properly adjusting relevant constants, we conclude the proof. \(\square \)
4 Proof of theorem 1.1: \(1<p\le 2\)
4.1 Expansion of positivity
Suppose the cylinder \(\mathcal {Q}\) and the numbers \(\varvec{\mu }^{\pm }\) and \(\varvec{\omega }\) are defined as in Sect. 3. The key ingredient of the reduction of oscillation lies in the following expansion of positivity, which is valid for \(1<p\le 2\).
Proposition 4.1
Let u be a locally bounded, local, weak sub(super)-solution to (1.1)–(1.3) in \(E_T\), with \(1<p\le 2\). Suppose for some constants \(\alpha ,\xi \in (0,1)\), there holds
Then there exist constants \(\delta ,\,\eta \in (0,1)\) depending only on the data \(\{s, p, N, C_o, C_1\}\) and \(\alpha \), such that either
or
provided
Moreover, we have \(\delta \approx \alpha ^{p+N+1}\) and \(\eta \approx \alpha ^q\) for some \(q>1\) depending on the data \(\{s, p, N, C_o, C_1\}\).
Proof
Assuming \((x_o,t_o)=(0,0)\) and \(\varvec{\mu }^{-}=0\) for simplicity, it suffices to deal with super-solutions. As a restatement, we need to show that there exist \(\delta \) and \(\eta \), such that if
and if
then
We stress that \(\delta \) and \(\eta \) depend only on the data \(\{s, p, N, C_o, C_1\}\) and \(\alpha \), but independent of \(\xi \).
Rewriting the measure theoretical information at the initial time \(t_o=0\) in the larger ball \(K_{4\varrho }\) and replacing \(\alpha \) by \(4^{-N}\alpha \), we can enforce
and apply Lemma 3.3 to obtain \(\delta , \varepsilon \in (0,1)\) depending only on the data \(\{s, p, N, C_o, C_1\}\) and \(\alpha \), such that
This measure theoretical information for each slice of the time interval in turn allows us to apply Lemma 3.4 in the cylinders \((0,\bar{t})+Q_{4\varrho }(\delta (\sigma \varepsilon \xi \varvec{\omega })^{2-p})\) with an arbitrary \(\bar{t}\in \big (\delta (\sigma \varepsilon \xi \varvec{\omega })^{2-p}(4\varrho )^{sp}, \delta (\xi \varvec{\omega })^{2-p}(4\varrho )^{sp}\big ]\), and with \(\xi \) and \(\alpha \) there replaced by \(\varepsilon \xi \) and \(\tfrac{1}{2} 4^{-N}\alpha \). This is viable because \(\sigma \in (0,1)\) and \(\delta (\sigma \varepsilon \xi \varvec{\omega })^{2-p}\le \delta (\xi \varvec{\omega })^{2-p}\); consequently, we have
when \(\bar{t}\) ranges over the given interval. Note also this step used the fact that \(p\le 2\).
Letting \(\nu \) be determined in Lemma 3.1 in terms of the data and \(\delta \), we further choose \(\sigma \) according to Lemma 3.4 to satisfy
This choice is possible because \(\varvec{\gamma }\) of Lemma 3.4 is independent of \(\sigma \). Further enforcing
such a choice of \(\sigma \) permits us to apply Lemma 3.1 in the cylinders \((0,\bar{t})+Q_{4\varrho }(\delta (\sigma \varepsilon \xi \varvec{\omega })^{2-p})\) with an arbitrary \(\bar{t}\in \big (\delta (\sigma \varepsilon \xi \varvec{\omega })^{2-p}(4\varrho )^{sp}, \delta (\xi \varvec{\omega })^{2-p}(4\varrho )^{sp}\big ]\), and with \(\xi \) there replaced by \(\sigma \varepsilon \xi \). Therefore, by arbitrariness of \(\bar{t}\) we conclude that
The proof is completed by defining \(\eta =\sigma \varepsilon \) and properly adjusting relevant constants in dependence of the data and \(\alpha \). \(\square \)
Based on Proposition 4.1, the remaining part is devoted to the proof of Theorem 1.1 for \(1<p\le 2\). All constants determined in the course of the proof are stable as \(p\rightarrow 2\).
4.2 The first step
For some \(Q_{\widetilde{R}}\subset E_T\) we introduce
and \(Q_o=Q_R(\varvec{\omega }^{2-p})\). By properly shrinking R, we may assume that \(Q_o\subset Q_{\widetilde{R}}\) and set
Without loss of generality, we take \((x_o,t_o)=(0,0)\).
Then the following intrinsic relation holds true:
The choice of the reference cylinder \(Q_{\widetilde{R}}\) is made to verify (4.1), on which the subsequent arguments are based.
Let \(\delta \in (0,1)\) be determined in Proposition 4.1 with \(\alpha =\tfrac{1}{2}\). For some \(c\in (0,\tfrac{1}{4})\) to be chosen, define
and consider two alternatives
Assuming \(\varvec{\mu }^+ - \varvec{\mu }^-\ge \tfrac{1}{2}\varvec{\omega }\), one of the two alternatives must hold. Whereas the case \(\varvec{\mu }^+ - \varvec{\mu }^-<\tfrac{1}{2}\varvec{\omega }\) will be trivially incorporated into the forthcoming oscillation estimate (4.3).
Let us suppose the first alternative holds for instance. An appeal to Proposition 4.1 with \(\alpha =\tfrac{1}{2}\), \(\xi =\tfrac{1}{4}\) and \(\varrho =cR\) determines \(\eta \in (0,\tfrac{1}{2})\) and yields that either
or
which, thanks to (4.1), gives the reduction of oscillation
The number c is chosen to ensure that (4.2) does not happen. Indeed, we may first estimate
This can be seen by the definitions of \(\varvec{\omega }\) and the tail,
Then, using (4.4) we choose
such that (4.2) does not occur. Note that (4.5) is not the final choice of c yet and it is subject to a further smallness requirement.
Next, we set \(R_1=\lambda R\) for some \(\lambda \le c\) to verify the set inclusion
As a result of this inclusion and (4.3) we obtain
which plays the role of (4.1) in the next stage.
4.3 The induction
Now we may proceed by induction.
Suppose up to \(i=1,\cdots , j\), we have built
The induction argument will show that the above oscillation estimate continues to hold for the \((j+1)\)-th step.
Let \(\delta \) be fixed as before, whereas \(c\in (0,1)\) is subject to a further choice. To reduce the oscillation in the next stage, we basically repeat what has been done in the first step, now with \(\varvec{\mu }^{\pm }_j\), \(\varvec{\omega }_j\), \(R_j\), \(Q_j\), etc. In fact, we define
and consider two alternatives
Like in the first step, we may assume \(\varvec{\mu }^+_j - \varvec{\mu }^-_j\ge \tfrac{1}{2}\varvec{\omega }_j\), so that one of the two alternatives must hold. Otherwise, the case \(\varvec{\mu }^+_j - \varvec{\mu }^-_j<\tfrac{1}{2}\varvec{\omega }_j\) can be trivially incorporated into the forthcoming oscillation estimate (4.8).
Let us suppose the first case holds for instance. An application of Proposition 4.1 in \(Q_j\), with \(\alpha =\tfrac{1}{2}\), \(\xi =\tfrac{1}{4}\) and \(\varrho =cR_j\) yields (for the same \(\eta \) as before) that either
or
which, thanks to the j-th induction assumption, gives the reduction of oscillation
The final choice of c is made to ensure that (4.7) does not happen, independent of j. This hinges upon the following tail estimate
To prove this, we first rewrite the tail as follows:
Here, we denoted \(K_i=K_{R_i}\) for short. The first integral is estimated by using the definition of \(\varvec{\omega }\). Namely, for any \(t\in (-\varvec{\omega }_j^{2-p}R_j^{sp},0)\),
Whereas the second integral is estimated by using the simple fact that, for \(i=1,2,\cdots ,j\),
Namely, for any \(t\in (-\varvec{\omega }_j^{2-p}R_j^{sp},0)\),
Combine them and further estimate the tail by
The summation in the last display is bounded by \((1-\eta )^{1-p}\) if we restrict the choice of \(\lambda \) by
Consequently, the tail estimate (4.9) is proven and (4.7) does not happen, if we choose c to verify
The final choice of c is made out of the smaller one of (4.5) and (4.10).
Let \(R_{j+1}=\lambda R_j\) for some \(\lambda \in (0,1)\) to verify the set inclusion
Note that the choice of \(\lambda \) in the last display may have been adjusted from the previous one in (4.6) due to the possible change of c made in (4.10). The final choice of \(\lambda \) is
As a result of the inclusion (4.11) and (4.8) we obtain
which completes the induction argument. From now on, the deduction of a Hölder modulus of continuity becomes quite standard; cf. [9, Chapter III, Proposition 3.1].
5 Proof of theorem 1.1: \(p> 2\)
As in the previous section, we first introduce \(Q_{\widetilde{R}}\subset E_T\),
and \(Q_o=Q_R(a\theta )\) for \(\theta =(\frac{1}{4}\varvec{\omega })^{2-p}\) and some \(a\in (0,1)\) to be determined. By properly shrinking R, we may assume that \(Q_o\subset Q_{\widetilde{R}}\) and set
Without loss of generality, we take \((x_o,t_o)=(0,0)\).
Then the following intrinsic relation holds:
As before, the choice of \(Q_{\widetilde{R}}\) is made to ensure (5.1), on which the subsequent arguments are based.
Unlike the case \(1<p\le 2\), an expansion of positivity for the case \(p>2\) requires additional technical complication. To deal with this case, we instead refine DiBenedetto’s argument in [10]: On one hand, the tail needs a great care in this intrinsic scaling scenario; on the other hand, we dispense with any kind of logarithmic estimate and just rely on the energy estimate.
5.1 The first alternative
In this section, we work with u as a super-solution near its infimum. Furthermore, we assume
The other case \(\varvec{\mu }^+ -\varvec{\mu }^- \le \frac{1}{2}\varvec{\omega }\), will be considered later.
Suppose \(a,\, c\in (0,1)\) verify that \(a>2c^{sp}\) for the moment (which will be confirmed in (5.11) and (5.12)), and for some \(\bar{t}\in \big (-a\theta R^{sp}+\theta (cR)^{sp},0\big ]\), there holds
where \(\nu \) is the constant determined in Lemma 3.1 (with \(\delta =1\)) in terms of the data. According to Lemma 3.1 with \(\delta =1\), \(\xi =\frac{1}{4}\) and \(\varrho =cR\), we have either
or
To proceed, we restrict c so that (5.4) does not happen. Indeed, since, according to the definition of \(\varvec{\omega }\), the tail can be easily estimated by (cf. (4.4))
we impose
Note that this is not the final choice of c yet and it is subject to further smallness requirements in the course of the proof.
The pointwise estimate in (5.5) at \(t_*=\bar{t}- \theta (\tfrac{1}{2} cR)^{sp}\) allows us to apply Lemma 3.2 with \(\varrho =\tfrac{1}{2}cR\) and obtain that for some free parameter \(\xi _o\in (0,\tfrac{1}{8})\), either
or
The number \(\xi _o\) is chosen to fulfill
In this way, the estimate (5.8) can be claimed up to \(t=0\) and yields the reduction of oscillation
Note that in the dependence of \(\xi _o\), the constants a and c are still to be determined. For the moment, let us suppose c has been fixed. We then select a to ensure (5.7) does not occur. In fact, since the tail is bounded by \(\varvec{\gamma }\varvec{\omega }\) as before, we impose
where we have employed the selection of \(\xi _o\) in (5.9). Consequently, the above display yields the relation of a and c, that is,
Hence, by this choice, the estimate (5.7) does not occur and the reduction of oscillation (5.10) actually holds.
Moreover, the assumption \(a>2c^{sp}\) made at the beginning is verified, if we use (5.11) and further restrict c by
The number a will be eventually fixed via (5.11), once we determine c in the end.
5.2 The second alternative
In this section, we work with u as a sub-solution near its supremum. Suppose (5.3) does not hold for any \(\bar{t}\in \big (-a\theta R^{sp}+\theta (cR)^{sp},0\big ]\). Due to (5.2), this can be rephrased as
Based on this, it is not hard to find some \(t_*\in \big [\bar{t}-\theta (cR)^{sp}, \bar{t}-\tfrac{1}{2}\nu \theta (cR)^{sp}\big ]\), such that
Indeed, if the above inequality were not to hold for any s in the given interval, then
which would yield a contradiction.
Starting from this measure theoretical information, we may apply Lemma 3.3 (with \(\alpha =\tfrac{1}{2}\nu \) and \(\varrho =cR\)) to obtain \(\delta \) and \(\varepsilon \) depending on the data and \(\nu \), such that, for some free parameter \(\xi _1\in (0,\frac{1}{4})\), either
or
The number \(\xi _1\) is chosen to satisfy
In this way, the measure theoretical information (5.14) can be claimed up to the time level \(\bar{t}\). Whereas the constant c is again chosen so small that (5.13) does not happen. Indeed, according to the definition of \(\varvec{\omega }\), just like in (4.4) the tail can be easily estimated by
A simple calculation as before then gives
Consequently, the measure theoretical information (5.14) yields
thanks to the arbitrariness of \(\bar{t}\).
Given (5.16), we want to apply Lemma 3.4 with \(\delta =1\), \(\xi =\varepsilon \xi _1\) and \(\varrho =cR\) next. To this end, we first let \(\nu \) be fixed in Lemma 3.1 (with \(\delta =1\)) and choose \(\sigma \in (0,\tfrac{1}{2})\) to satisfy
This choice is possible because \(\varvec{\gamma }\) of Lemma 3.4 is independent of \(\sigma \). Then we use (5.11) and restrict c further to satisfy
In this way, the measure theoretical information (5.16) gives that
which allows us to implement Lemma 3.4. Namely, we have either
or
By Lemma 3.1 (with \(\delta =1\)), the last display yields
which in turn gives the reduction of oscillation
Once again, we may restrict the choice of c to ensure that (5.18) does not happen.Indeed, according to the definition of \(\varvec{\omega }\), just like in (4.4) the tail can be easily estimated by
By a similar calculation as before, this amounts to requiring
Combining (5.10) and (5.19), we arrive at
where
Now, set \(\theta _1=(\tfrac{1}{4}\varvec{\omega }_1)^{2-p}\) and \(R_1=\tfrac{1}{4} c R\). To prepare the induction, we need to verify the set inclusion
This, by the choice of a in (5.11), amounts to further requiring the smallness of c.
As a result of this inclusion and (5.21), we obtain
which takes the place of (5.1) in the next stage. Note that this oscillation estimate also takes into account the reverse case of (5.2).
5.3 The induction
Now we may proceed by induction.
Suppose up to \(i=1,\cdots , j\), we have built
The induction argument will show that the above oscillation estimate continues to hold for the \((j+1)\)-th step.
Like in the proof for \(1<p\le 2\), we can repeat all the previous arguments, which now are adapted with \(\varvec{\mu }^{\pm }_j\), \(\varvec{\omega }_j\), \(R_j\), \(\theta _j\), \(Q_j\), etc. In the end, we have a reduction of oscillation parallel with (5.21), that is,
Now, setting \(\theta _{j+1}=(\tfrac{1}{4}\varvec{\omega }_{j+1})^{2-p}\) and \(R_{j+1}=\tfrac{1}{4} c R_j\), it is straightforward to verify the set inclusion
which, by (5.22), implies
A key step lies in determining c as done in (5.6), (5.15), (5.12), (5.17) and (5.20), such that the alternative involving the tail, along the course of the arguments, does not really occur and hence (5.22) can be reached. This hinges upon the following estimate of the tail:
The computations leading to the above tail estimate can be performed as those leading to (4.9). We omit the details to avoid repetition. After the number c is determined independent of j, the number a is finally chosen via the relation (5.11). Hence the induction is completed and the derivation of a Hölder modulus of continuity follows.
Data availability
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
Notes
After completion of the paper, we noticed a similar result in the recent preprint [1], via exponential change of variables.
The reader is advised to compare the either-or statement with Lemma 3.1 of [11, Chapter 3]. There, the constant C in (3.2) information of lower order terms; Here, we treat the tail as a kind of lower order term.
References
Adimurthi, K., Prasad, H., Tewary, V.: Local Hölder regularity for nonlocal parabolic \(p\)-Laplace equations, arXiv:2205.09695
Banerjee, A., Garain, P., Kinnunen, J.: Some local properties of subsolutons and supersolutions for a doubly nonlinear nonlocal parabolic p-Laplace equation. Ann. Mat. Pura Appl. 201(4), 1717–1751 (2022)
Banerjee, A., Garain, P., Kinnunen, J.: Lower semicontinuity and pointwise behavior of supersolutions for some doubly nonlinear nonlocal parabolic \(p\)-Laplace equations. Commun. Contemp. Math. 25(8), 2250032 (2023)
Brasco, L., Lindgren, E., Strömqvist, M.: Continuity of solutions to a nonlinear fractional diffusion equation. J. Evol. Equ. 21(4), 4319–4381 (2021)
Caffarelli, L.A.: Non-local diffusions, drifts and games. Nonlinear partial differential equations, In: The Abel Symposium, Springer, Heidelberg 7, pp. 37–52 (2012)
Caffarelli, L.A., Chan, C.H., Vasseur, A.: Regularity theory for parabolic nonlinear integral operators. J. Am. Math. Soc. 24(3), 849–869 (2011)
Chen, Y.-Z., DiBenedetto, E.: Hölder estimates of solutions of singular parabolic equations with measurable coefficients. Arch. Ration. Mech. Anal. 118(3), 257–271 (1992)
Cozzi, M.: Regularity results and Harnack inequalities for minimizers and solutions of nonlocal problems: a unified approach via fractional De Giorgi classes. J. Funct. Anal. 272, 4762–4837 (2017)
DiBenedetto, E.: Degenerate Parabolic Equations. Springer-Verlag, New York (1993)
DiBenedetto, E.: On the local behaviour of solutions of degenerate parabolic equations with measurable coefficients. Ann. Scuola Norm. Sup. Pisa Cl. Sci. 13(3), 487–535 (1986)
DiBenedetto, E., Gianazza, U., Vespri, V.: Harnack’s Inequality for Degenerate and Singular Parabolic Equations. Springer Monographs in Mathematics, Springer-Verlag, New York (2012)
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 Castro, A., Kuusi, T., Palatucci, G.: Nonlocal Harnack inequalities. J. Funct. Anal. 267, 1807–1836 (2014)
Di Nezza, E., Palatucci, G., Valdinoci, E.: Hitchhiker’s guide to the fractional Sobolev spaces. Bull. Sci. Math. 136(5), 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(1), 38–45 (2021)
Felsinger, M., Kassmann, M.: Local regularity for parabolic nonlocal operators. Comm. Partial Differ. Equ. 38(9), 1539–1573 (2013)
Kassmann, M.: A priori estimates for integro-differential operators with measurable kernels. Calc. Var. Partial Differ. Equ. 34, 1–21 (2009)
Kim, Y.-C.: Nonlocal Harnack inequalities for nonlocal heat equations. J. Differ. Equ. 267(11), 6691–6757 (2019)
Kim, Y.-C.: Local properties for weak solutions of nonlocal heat equations. Nonlinear Anal. 192(30), 111689, 30 (2020)
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)
Silvestre, L.: Hölder estimates for solutions of integro-differential equations like the fractional Laplace. Indiana Univ. Math. J. 55(3), 1155–1174 (2006)
Strömqvist, M.: Local boundedness of solutions to nonlocal parabolic equations modeled on the fractional \(p\)-Laplacian. J. Differ. Equ. 266, 7948–7979 (2019)
Strömqvist, M.: Harnack’s inequality for parabolic nonlocal equations. Ann. Inst. H. Poincaré Anal. Non Linéaire 36, 1709–1745 (2019)
Urbano, J.M.: The method of intrinsic scaling. A systematic approach to regularity for degenerate and singular PDEs, Lecture Notes in Mathematics, 1930. Springer-Verlag, Berlin (2008)
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)
Acknowledgements
This work was supported by the FWF-Project P36272-N “On the Stefan type problems" and the FWF-Project P31956-N32 "Doubly nonlinear evolution equations”.
Funding
Open access funding provided by Paris Lodron University of Salzburg.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Jan Kristensen.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A. Fractional Sobolev inequalities
Note that the definition of the space \(W^{s,p}\) in Sect. 1.2 is also valid for \(p=1\), though we will not use it. An elementary proof of the following result can be retrieved from [14, Theorem 6.5].
Proposition A.1
Let \(s\in (0,1)\) and \(p\ge 1\) satisfying \(sp<N\). For any measurable and compactly supported function \(u:\mathbb {R}^N\rightarrow \mathbb {R}\), there holds
for some positive constant \(C=C(s,p,N)\).
It is our intention to circumvent a more advanced theory of function spaces, extension domains, etc. and keep this section as elementary as possible.
The following local version is a direct consequence of Proposition A.1. Here and in the sequel, we omit the reference to the center of a ball \(K_R\).
Proposition A.2
Let \(s\in (0,1)\) and \(p\ge 1\). For any function \(u\in W^{s,p}(K_R)\) that is compactly supported in \(K_{(1-d)R}\) with some \(d\in (0,1)\), there holds
for some positive constant \(C=C(\kappa , s, p, N)\), where
Proof
It suffices to show the inequality for \(R=1\), thanks to its scaling invariance in R. Since u is compactly supported in \(K_1\), we may view it as a function in \(W^{s,p}(\mathbb {R}^N)\) upon zero extension. This fact is verified by a simple calculation (similar to what follows). First consider the case \(sp<N\). According to Proposition A.1, we estimate
In the last line, we used the fact that \(y\in K_{1-d}\) (recalling \({\text {supp}}u\)) and \(x\in \mathbb {R}^N\setminus K_1\). Consequently, the desired inequality for any \(\kappa \) in the interval follows from an application of Hölder’s inequality. Note also the constant C is actually independent of \(\kappa \) in this case.
Next, we consider the case \(sp\ge N\). A first observation is that \(W^{\bar{s},p}(K_1)\subset W^{s,p}(K_1)\) for any \(0<\bar{s}\le s<1\). Quantitatively, we estimate
Now we select \(\bar{s}=\frac{(\kappa -1)}{\kappa p}N\), which verifies \(\bar{s}p<N\), and apply (A.1) in \(W^{\bar{s},p}(K_1)\), together with the above observation, to conclude. \(\square \)
Based on Proposition A.2, the following parabolic imbedding is in order. It plays an essential role in proving DeGiorgi type lemmas.
Proposition A.3
Let \(s\in (0,1)\), \(p\ge 1\) and
For any function
which is compactly supported in \(K_{(1-d)R}\) for some \(d\in (0,1)\) and for a.e. \(t\in (t_1,t_2)\), there holds
for some positive constant \(C=C(s,p,N)\), where
Proof
The inequality is scaling invariant in R. Hence, it suffices to show it for \(R=1\). An application of Hölder’s inequality, followed by Proposition A.2 and the choice of \(\kappa \), yields that
This finishes the proof. \(\square \)
Appendix B: Time mollification
The time derivative of a weak solution in general does not exist in the Sobolev sense. On the other hand, it is desirable to use the solution in a testing function when we derive the energy estimate in Proposition 2.1. To overcome this well-organized difficulty, we find the following mollifier quite convenient. Namely, we introduce for any \(v\in L^1(E_T)\),
Properties of this mollification can for instance be found in [20, Lemma 2.2]. In particular, by direct differentiation, we obtain the following identities:
They are paired by the identity
Another fact, we will rely on is that, if \(v\in L^q(E_T)\) for some \(q\ge 1\), then \(\llbracket v \rrbracket _h,\, \llbracket v \rrbracket _{\bar{h}}\in L^q(E_T)\), and moreover, as \(h\rightarrow 0\) we have
Now we take on a rigorous justification of formal calculations in Proposition 2.1. By working with a closed interval in (0, T), we may assume \(u\in C\big ([0,T];L^2_{{\text {loc}}}(E)\big )\) with no loss of generality. For ease of notation, we denote
Let the cutoff function \(\zeta \) be as in Proposition 2.1 and introduce \(\psi _\varepsilon (t),\,\varepsilon >0\) to be a Lipschitz approximation of \(\chi _{\{t_o-S<t<t_o\}}\). Namely, \(\psi _\varepsilon =1\) in \((t_o-S+\varepsilon , t_o-\varepsilon )\), \(\psi _\varepsilon =0\) outside \((t_o-S,t_o)\), while linearly interpolated otherwise. In the weak formulation (1.7), we take the testing function
we obtain
The first term in (B.3) can be written as, recalling (B.1),
On the right-hand side of the last display, the second term is discarded due to its non-negative contribution, whereas the third term converges to 0 as \(h\rightarrow 0\) owing to (B.2). For the first term, we integrate by parts in time, let \(h\rightarrow 0\) and then let \(\varepsilon \rightarrow 0\), to obtain
These are the last terms on the left/right-hand side of the energy estimate in Proposition 2.1.
The next goal is to show that the second integral in (B.3) converges to
if we send \(h\rightarrow 0\), where
Once this is shown, we will justify the formal calculations in the proof of Proposition 2.1.
For this purpose, we take the difference between the two integrals and split the obtained integral into two parts, that is,
Note that we used \(\varphi _h(y,t)=\varphi (y,t)=0\) for \(y\in \mathbb {R}^N{\setminus } K_{2R}\) in defining \(I_2\). We then show that \(I_1\) and \(I_2\) converge to 0 after sending \(h\rightarrow 0\).
For \(I_1\), we first claim that
for some constant \(C=C(s,p,N, R, \Vert D\zeta \Vert _{\infty } )\). This uniform bound allows us to extract a weakly convergent subsequence (still labeled by h). Since, up to a subsequence, \(\varphi _h\rightarrow \varphi \) a.e. in \(E_T\), we obtain that, up to a subsequence,
Consequently, we have \(I_1\rightarrow 0\) in view of this weak convergence and the fact that
To prove (B.4), the numerator of the integrand needs to be controlled. Using the triangle inequality and the fact that \(\zeta \) and \(\psi _\varepsilon \) are bounded by 1, we compute
for some proper \(c=c(p)\). Next, we estimate the first term on the right-hand side of the last display by Hölder’s inequality:
A time integration of the above estimate, followed by Fubini’s theorem, gives that
Thus we deduce that
Analogously, for the second term, we may first estimate
and then,
Combining (B.5) and (B.6), we have shown (B.4).
To deal with \(I_2\), we first notice that \(|x-y|\ge \tfrac{1}{2} |y|\) for \(y\in \mathbb {R}^N{\setminus } K_{2R}\) and \(x\in K_{R}\). Hence, we estimate
Now we easily see that \(I_2\rightarrow 0\) as \(h\rightarrow 0\) due to (B.2).
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.