Abstract
In this article we prove a Harnack inequality for non-negative weak solutions to doubly nonlinear parabolic equations of the form
where the vector field \({\mathbf {A}}\) fulfills p-ellipticity and growth conditions. We treat the slow diffusion case in its full range, i.e. all exponents \(m > 0\) and \(p>1\) with \(m(p-1) > 1\) are included in our considerations.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction and results
Let \(\Omega \subset {\mathbb {R}}^n\), \(n \ge 2\), be a bounded open domain and (0, T) with \(0<T<\infty \) a finite time interval. In the following, \(\Omega _T := \Omega \times (0,T)\) denotes the related space-time cylinder. The prototype of the doubly nonlinear equations we are concerned with is
for non-negative solutions \(u :\Omega _T \rightarrow {\mathbb {R}}_{\ge 0}\) with parameters \(m \in (0,\infty )\) and \(p \in (1,\infty )\). If \(m=1\), (1.1) reduces to the parabolic p-Laplace equation, whereas for \(p=2\) we retrieve the porous medium equation. Doubly nonlinear equations of type (1.1) are classified as doubly degenerate if \(m > 1\) and \(p >2\), singular-degenerate if \(m > 1\) and \(p \in (1,2)\), degenerate-singular if \(m \in (0,1)\) and \(p >2\) and doubly singular if \(m \in (0,1)\) and \(p \in (1,2)\). Furthermore, depending on the behavior of solutions, we distinguish between slow diffusion equations with \(m(p-1) > 1\) and fast diffusion equations with \(m(p-1) < 1\). The qualitative difference between both cases stems from the fact that in the former one solutions might have a compact support, while this is not possible in the latter one. In the present paper, we treat the complete slow diffusion range \(p(m-1) > 1\), which includes the doubly degenerate case and the singular-degenerate and degenerate-singular slow diffusion case.
In the literature, (1.1) often appears in equivalent forms; cf. [17,18,19,20, 28, 33]. More precisely, we note that formally (1.1) is a transformation of
with \(\widehat{m} := \frac{1}{m}\) and
where \(\ell := (m-1)(p-1)\). These representations of (1.1) can be shown to be equivalent. Let us also note that for \(m>1\) there are two different notions of weak solutions to the porous medium equation and doubly nonlinear equations in the literature. The first one assumes that \(u^{\frac{m+1}{2}}\) is weakly differentiable with respect to the space variable, whereas the second one claims this for \(u^m\) (in the case \(m<1\) only the latter one makes sense). For the prototype porous medium equation the equivalence of both notions of solutions has been shown in [6]. It is still an open problem if the same is true for doubly nonlinear equations and porous medium type equations with a general structure.
Harnack estimates play a crucial role in the regularity theory of partial differential equations. In the elliptic setting, essential contributions are due to Moser [25] for linear elliptic equations and Serrin [29] and Trudinger [31] for quasilinear elliptic equations. In the parabolic setting, the first results have been obtained by Hadamard [16] and Pini [27] for non-negative solutions of the heat equation. For the heat equation Harnack’s inequality takes the form
with waiting time \(\varrho ^2\). Moser [26] showed that this result is true for linear parabolic equations as well and demonstrated the necessity of the waiting time. Later, Trudinger [32] proved Harnack inequalities for quasilinear parabolic equations and the homogeneous doubly nonlinear equation
with \(p>1\). Using an approach based on mean value inequalities for suitable De Giorgi classes, Gianazza and Vespri [14] gave a proof that extends to more general operators A(x, t, u, Du) instead of \(|Du|^{p-2}Du\). Finally, simplifying an approach originally introduced by Moser, Kinnunen & Kuusi [22] obtained Harnack’s inequality for the homogeneous doubly nonlinear equation, where the Lebesgue measure is replaced by a more general Borel measure. In the case of non-homogeneous nonlinear equations, the situation is more involved. DiBenedetto [7] proved that non-negative weak solutions of the parabolic p-Laplace equation and the porous medium equation satisfy an intrinsic Harnack inequality of the form
with \(t_w = c u(x_o,t_o)^{2-p} \varrho ^p\) for the parabolic p-Laplace equation and \(t_w = c u(x_o,t_o)^{1-m} \varrho ^2\) for the porous medium equation. These Harnack inequalities are called intrinsic, because the waiting times depend on the solution itself. Loosely speaking, solutions of non-homogeneous equations behave like solutions of the heat equation in an intrinsic time scale. A counterexample [11] shows that a Harnack estimate with \(t_w\) independent of u is false. Since the proof in [7] relies on comparison with explicit solutions, it cannot be adapted for general quasilinear equations. Nearly 20 years later, this problem was overcome by DiBenedetto, Gianazza & Vespri [9], whose proof only uses measure theoretical tools. The main novelty is the so-called Expansion of Positivity. The same method was used by Kuusi [23] to obtain weak Harnack estimates for super-solutions of nonlinear degenerate parabolic equations. For an extensive overview regarding the parabolic p-Laplace equation and the porous medium equation with the definition of weak solution involving \(u^\frac{m+1}{2}\), we refer to the monograph [10] by DiBenedetto, Gianazza and Vespri and the survey [11] by Düzgün, Fornaro and Vespri. Harnack’s inequality for the prototype doubly nonlinear equation
has first been proved by Vespri [33] for the full range of parameters \(p >1\) and \(m+p > \max \{2,3-\frac{p}{n}\}\). The proof uses explicit constructions involving the Barenblatt solution and therefore cannot be applied to more general structures. For the doubly degenerate case Fornaro and Sosio [12] generalized the result to weak solutions of
where the operators \({\mathbf {A}}\) and \({\mathbf {B}}\) fulfill the conditions
with \(p \ge 2\), positive constants \(c_0, c_1, c_2, c\) and a function \(\Phi \) satisfying an \((m-1)\)-growth condition with \(m \ge 1\). They used a definition of weak solution involving \(u^\frac{m+1}{2}\). A weak Harnack inequality for super-solutions can be found in [24]. For the case of fast diffusion equations, we refer to the articles by Fornaro, Sosio and Vespri [13] and Vespri and Vestberg [34].
In this paper we prove Harnack’s inequality for the entire slow diffusion range and thereby close the gap for the by now missing singular-degenerate and degenerate-singular slow diffusion cases. Furthermore, we work with a definition of weak solution involving \(u^m\), which is new even for the doubly degenerate case and the slow diffusion porous medium equation.
1.1 Setting
We consider non-negative weak solutions to the doubly nonlinear equation
with \(m > 0\). For the vector field \({\mathbf {A}}:\Omega _T\times {\mathbb {R}}\times {\mathbb {R}}^{n} \rightarrow {\mathbb {R}}^{n}\) we assume that \({\mathbf {A}}\) is measurable with respect to \((x,t)\in \Omega _T\) for all \((u,\xi ) \in {\mathbb {R}}\times {\mathbb {R}}^{n}\) and continuous with respect to \((u,\xi )\) for a.e. \((x,t) \in \Omega _T\). Moreover, we assume that \({\mathbf {A}}\) satisfies the following growth and ellipticity conditions
for \(p>1\) and structure constants \(0<\nu \le L<\infty \). We demand that
for some \(\sigma >\frac{n+p}{p-1}\) and that the parameters m and p satisfy \(m(p-1) > 1\) which means that we are in the slow diffusion range. In the following we abbreviate
We now give the precise definition of weak solution to (1.3) that we use throughout the paper.
Definition 1.1
Assume that the vector field \({\mathbf {A}}\) satisfies (1.4). A non-negative measurable function \(u :\Omega _T \rightarrow {\mathbb {R}}_{\ge 0}\) in the class
is a non-negative weak sub(super)-solution to the doubly nonlinear equation (1.3) if and only if the identity
holds true for any testing function \(\varphi \in C_0^\infty (\Omega _T,{\mathbb {R}}_{\ge 0})\). If u is a weak sub- and super-solution it is called a weak solution.
We are now in the position to formulate the main result of our paper:
Theorem 1.2
Let \(m >0\), \(p>1\) with \(m(p-1) > 1\) and u be a continuous, non-negative, weak solution to (1.3) in the sense of Definition 1.1, where the vector field \({\mathbf {A}}\) satisfies (1.4) and F satisfies (1.5). Moreover, let \((x_o,t_o)\in \Omega _T\) such that \(u(x_o,t_o)>0\). Then, there exist constants \(c_o,\gamma >1\) depending only on \(n,m,p,L,\nu \) and \(\sigma \) such that for all cylinders \(B_{9\varrho }(x_o) \times (t_o - 4 \theta \varrho ^p,t_o + 4 \theta \varrho ^p) \Subset \Omega _T\), with
we either have
or
Note that the continuity assumption in Theorem 1.2 is not restrictive. The Harnack inequality continues to hold for a.e. point \((x_o,t_o)\in \Omega _T\) if we state it for an arbitrary non-negative, weak solution to (1.3). However, for the sake of a neater exposition of the result, we prefer to state it for continuous solutions.
1.2 Plan of the paper
In Sect. 2 we collect some auxiliary tools. Using \(u^m - a^m\) for some level a as test function in (a mollified version of) the definition of weak sub- and super-solutions, we derive certain Caccioppoli inequalities in Sect. 3. For convenience of the reader we state all intermediate results for weak sub- respectively super-solutions instead of weak solutions, so that it becomes clear what the minimal assumptions are. Next, in Sect. 4 we show that weak sub-solutions to (1.3) are locally bounded and give a quantitative estimate. In Sect. 5 we prove so-called De Giorgi type lemmas. Loosely speaking, the first lemma shows that if a super-solution u to (1.3) is smaller than some level M only on a small enough proportion of a suitable cylinder, then u is larger than \(\frac{M}{2}\) a.e. on a smaller cylinder contained in the first one. The second lemma gives an analogous statement for sub-solutions in the case that u is larger than a fixed level only on a small enough proportion of the bigger cylinder and consequently smaller than a fraction of the level on the smaller cylinder. The proofs of the statements rely in particular on the Caccioppoli estimates. In Sect. 6 we prove Expansion of Positivity of non-negative weak super-solutions. The conclusion of the section is that if
for a level \(M>0\), \(\alpha \in (0,1)\) and a suitable ball \(B_\varrho (x_o)\), then \(u \ge \kappa M\) a.e. in \(B_{2\varrho } (x_o) \times \big (t_o + \frac{1}{2} b (\kappa M)^{-d}\varrho ^p,t_o + b (\kappa M)^{-d}\varrho ^p \big ]\). Here, the constants \(b, \kappa \in (0,1)\) depend only on the data and \(\alpha \). In the proof, the Caccioppoli estimates and the first De Giorgi type lemma are used. Finally, in Sect. 7 we deduce the intrinsic Harnack inequality stated in Theorem 1.2. To show the forward inequality, i.e. the second inequality in (1.8), after a transformation we use the second De Giorgi type lemma and iteratively apply Expansion of positivity. Subsequently, we prove that the forward inequality implies the backward Harnack inequality, i.e. the first inequality in (1.8). Actually, a more general version of the backward Harnack inequality is shown in Sect. 7.
2 Preliminaries
2.1 Notation
First, we introduce some notation used throughout the paper. For functions defined on \(\Omega _T\), we denote the time slice at time \(t\in (0,T)\) by \(v(t) := v(\cdot ,t)\). For \(z_o=(x_o,t_o)\in {\mathbb {R}}^n \times {\mathbb {R}}\) we define space-time cylinders
with a radius \(\varrho >0\) and time length \(\theta > 0\) and let
As usual, we let
for \(u,a\in {\mathbb {R}}\). Furthermore, for \(u,a\ge 0\) we define the boundary term
2.2 Mollification in time
Since weak solutions do not possess a time derivative in general we have to use mollification. To this end, for \(v \in L^1(\Omega _T,{\mathbb {R}}^N)\) and \(h>0\) we define the following mollification in time
which formally satisfies the ordinary differential equation
Basic properties of \([\![ \cdot ]\!]_h\) are provided in the following lemma. For its proof and further information, we refer to [21, Lemma 2.2] and [5, Appendix B].
Lemma 2.1
Suppose that X is a separable Banach space. If \(v \in L^r(0,T;X)\) for some \(r \ge 1\), then the mollification \([\![ v ]\!]_h\) defined in (2.2) fulfills \([\![ v ]\!]_h \in L^r(0,T;X)\) and for any \(t_o \in (0,T]\) there holds
Moreover, in the case \(r<\infty \) we have \([\![ \cdot ]\!]_h \rightarrow v\) in \(L^r(0,T;X)\) as \(h \downarrow 0\).
Using the same technique as in [30, Lemma 3.6], we conclude that any sub(super)-solution to (1.3) in the sense of Definition 1.1 satisfies the mollified version of (1.6),
for any \(\varphi \in C_0^\infty (\Omega _T,{\mathbb {R}}_{\ge 0})\).
2.3 Transformation
The following Lemma is an easy consequence of a change of variables.
Lemma 2.2
Let \(T>0\), \(I \subset {\mathbb {R}}\) be an open interval and \(\Phi :I \rightarrow (0,T)\) an increasing \(C^\infty \)-diffeomorphism. Then, u is a weak sub(super)-solution to (1.3) associated to \({\mathbf {A}}\), F in \(B_\varrho \times (0,T)\) if and only if the function \(w(x,\tau ):=u(x,\Phi (\tau ))\) is a sub(super)-solution to (1.3) associated to the vector field
and right-hand side \({\widetilde{F}}(x,\tau ) := \Phi '(\tau ) F(x,\Phi (\tau ))\) in \(B_\varrho \times I\).
The next Lemma shows that the product of a non-negative weak super-solution u with a non-decreasing \(C^1\)-function \(\gamma \) is a super-solution to a modified equation. A similar argument has already been used in [9].
Lemma 2.3
Let \(\Omega \subset {\mathbb {R}}^n\) be bounded and open and \(I\subset {\mathbb {R}}\) an open interval. Assume that u is a non-negative weak super-solution to (1.3) in \(\Omega \times I\) associated to \({\mathbf {A}}\), F and \(\gamma \in C^1(I)\cap C^0({\overline{I}})\) is non-decreasing and satisfies \(\frac{1}{C} \le \gamma \le C\) on I for a constant \(C\ge 1\). Then, the function \({\tilde{u}} := \gamma u\) is a non-negative weak super-solution to (1.3) in \(\Omega \times I\) associated to the vector-field
and inhomogeneity \({\widetilde{F}}:=\gamma F\).
Proof
In the following we abbreviate \(\Omega _I:=\Omega \times I\). Let \(\varphi \in C_0^\infty (\Omega _I,{\mathbb {R}}_{\ge 0})\). Then \(\gamma \varphi \in C_0^1(\Omega _I,{\mathbb {R}}_{\ge 0})\). By assumption \(\gamma '\), u and therefore also \([\![ u ]\!]_h\) are non-negative. By an approximation argument we may use \(\gamma \varphi \) as testing function in the mollified weak formulation (2.4) on the interval I instead of (0, T). This leads to
Passing to the limit \(h \downarrow 0\) with the help of Lemma 2.1 and taking into account that \({{\,\mathrm{spt}\,}}(\gamma \varphi )\) is compact in the last term on the right-hand side, this leads to
for every \(\varphi \in C_0^\infty (\Omega _I,{\mathbb {R}}_{\ge 0})\), which is in view of the definition of \({\tilde{u}}\) and \({\widetilde{F}}\) equivalent to
Recalling the definition of \(\widetilde{{\mathbf {A}}}\), this yields the claim. \(\square \)
Combining the last two lemmata leads to the following statement, which is used in the proof of the expansion of positivity.
Corollary 2.4
Let \(T>0\) and u a non-negative weak super-solution to (1.3) in \(B_\varrho \times (0,T)\) associated to \({\mathbf {A}}\) and F. Further, assume that \(I \subset {\mathbb {R}}\) is an open interval, that \(\Phi :I \rightarrow (0,T)\) is an increasing \(C^\infty \)-diffeomorphism and that \(\gamma \in C^1(I)\cap C^0({\overline{I}})\) is non-decreasing and satisfies \(\frac{1}{C} \le \gamma \le C\) on I for some constant \(C\ge 1\). Then, the function \(v(x,\tau ) := \gamma (\tau ) \cdot u(x,\Phi (\tau ))\) is a non-negative weak super-solution to (1.3) in \(B_\varrho \times I\) associated to the vector-field
and inhomogeneity \(\widehat{F}(x,\tau ) := \gamma (\tau )\Phi '(\tau )F(x,\Phi (\tau ))\).
2.4 Auxiliary lemmata
For a function \(v \in W^{1,1}\) and \(k<\ell \) the next lemma gives a local estimate for the product of the measures of superlevel sets \(\{ v > \ell \}\) and sublevel sets \(\{ v < k\}\) in terms of the \(L^1\)-norm of Dv on the intersection of their complements, cf. [8, Chap. I.2, Lemma 2.2 and Remark 2.3].
Lemma 2.5
Let \(v\in W^{1,1}(B_\varrho (x_o))\) and \(k,\ell \in {\mathbb {R}}\) with \(k<\ell \). Then, there exists a constant c depending on n such that
The following lemma can be found in the literature; cf. [1, Lemma 2.2] for \(\alpha \in (0,1)\) and [15, inequality (2.4)] for \(\alpha >1\).
Lemma 2.6
For any \(\alpha >0\), there exists a constant \(c=c(\alpha )\) such that, for all \(a,b\ge 0\), the following inequality holds true:
The next lemma summarizes all properties we need concerning the boundary term \({\mathfrak {b}}\) defined in (2.1).
Lemma 2.7
Let \(m>0\). There exists a constant \(c=c(m)\) such that for every \(u,a \ge 0\) we have
-
(i)
\(\frac{1}{c} \big |u^\frac{m+1}{2} - a^\frac{m+1}{2}\big |^2 \le {\mathfrak {b}}[u^m,a^m] \le c \big |u^\frac{m+1}{2} - a^\frac{m+1}{2}\big |^2\).
-
(ii)
\(\frac{1}{c} |u^m-a^m|^2 \le (u+a)^{m-1}\, {\mathfrak {b}}[u^m,a^m] \le c |u^m-a^m|^2\).
Proof
The proof of (i) can be found in [4, Lemma 2.3] for \(m\ge 1\) and in [3, Lemma 3.4] for \(0<m<1\). The inequalities in (ii) are a consequence of (i) and Lemma 2.6. \(\square \)
The following iteration lemma is a well known result and can be found for instance in [8, Chap. I.4, Lemma 4.1].
Lemma 2.8
Let \((Y_i)_{i \in {\mathbb {N}}_0}\) be a sequence of non-negative numbers satisfying
with some positive constants \(\kappa , \gamma \) and \(b >1\). If
then \(Y_i \rightarrow 0\) as \(i \rightarrow \infty \).
Finally, we recall a parabolic version of the Gagliardo–Nirenberg inequality, see [8, Chapter I, Proposition 3.1] or [2, Lemma 3.1].
Lemma 2.9
Let \(Q^-_{\varrho ,\theta }(z_o) \subset {\mathbb {R}}^{n+1}\) be a parabolic cylinder and \(1<p,r<\infty \). For every
we have \(u \in L^q(Q^-_{\varrho ,\theta }(z_o) )\) for \(q=p(1+\frac{r}{n})\) with the estimate
where \(c=c(n,p,r)\).
3 Caccioppoli inequalities
In this section we derive energy estimates that are crucial in the course of the paper. We start with the energy estimates for weak super-solutions.
Lemma 3.1
Let \(m >0\), \(p>1\) with \(m(p-1) > 1\) and u be a non-negative weak super-solution to (1.3) in the sense of Definition 1.1, where the vector-field \({\mathbf {A}}\) fulfills the growth and ellipticity assumptions (1.4). Then, there exists a constant \(c=c(p,\nu ,L)\) such that on any cylinder \(Q^-_{\varrho ,\theta }(z_o) \Subset \Omega _T\) with \(\varrho ,\theta > 0\), and for any \(0<r<\varrho \), \(0<s<\theta \) and \(a\ge 0\) the following energy estimates
and
hold true, where \({\mathfrak {b}}[\cdot ,\cdot ]\) is defined in (2.1).
Proof
Throughout the proof we abbreviate \(Q^-_{\varrho ,\theta }\equiv Q^-_{\varrho ,\theta }(z_o)\) and \(B_\varrho \equiv B_\varrho (x_o)\). Since the claimed estimates are local in nature, we may assume without loss of generality that \(u \in C^0([0,T);L^{m+1}(\Omega ))\). An approximation argument shows that the mollified weak formulation (2.4) extends to non-negative testing functions \(\varphi \in L^p(0,T;W^{1,p}_0(\Omega ))\cap L^{\frac{m+1}{m}}(\Omega _T)\) with compact support, since \([\![ u ]\!]_h\in C^0([0,T);L^{m+1}(\Omega ))\), \([\![ {\mathbf {A}}(x,t,u,Du^m) ]\!]_h\), \([\![ F ]\!]_h \in L^\frac{p}{p-1}(\Omega _T)\) and \(u(0) \in L^{m+1}(\Omega )\) by the assumptions on u, growth condition (1.4) and Lemma 2.1. We therefore find that
holds true for any \(\varphi \in L^p(0,T;W_0^{1,p}(\Omega ,{\mathbb {R}}_{\ge 0}))\cap L^{\frac{m+1}{m}}(\Omega _T)\) with compact support. For \(\varepsilon >0\) and \(t_1 \in \Lambda _s(t_o)=(t_o-s,t_o)\) we define cutoff functions \(\eta \in W^{1,\infty }(B_\varrho (x_o),[0,1])\), \(\zeta \in W^{1,\infty }(\Lambda _{\theta }(t_o),[0,1])\) and \(\psi _\varepsilon \in W^{1,\infty }(\Lambda _{\theta }(t_o),[0,1])\) which satisfy
We choose
as testing function in the mollified version (3.3) of the differential equation. For the first term on the left hand side we have
where we used in turn (2.3), the fact that \((u - [\![ u]\!]_h)(u^m - [\![ u]\!]_h^m ) \ge 0\) by monotonicity of \(s \mapsto s^m\) and the definition of \({\mathfrak {b}}\). Since \([\![ u ]\!]_h \rightarrow u\) in \(L_{\text {loc}}^{m+1}(\Omega _T)\) in the limit \(h \downarrow 0\), we get
where the meaning of \(\mathrm {I}_\varepsilon \) and \(\mathrm {II}_\varepsilon \) is clear in this context. We let \(h\downarrow 0\) also in the diffusion term. For the resulting integral we use assumptions (1.4) and Young’s inequality to obtain
where \(c=c(p,\nu ,L)\). The second term on the right hand side of (3.3) vanishes in the limit \(h \downarrow 0\), since \(\varphi (0) \equiv 0\). In the first integral we pass to the limit \(h \downarrow 0\) and then apply Young’s inequality. This yields
Inserting the preceding estimates into (3.3), we conclude that
Now, we pass to the limit \(\varepsilon \downarrow 0\) in the preceding inequality. Since \(u \in C^0([0,T];L^{m+1}(\Omega ))\), for any \(t_1 \in \Lambda _s(t_o)\) we obtain
Further, we have
and
Altogether, we deduce the estimate
for any \(t_1 \in \Lambda _s(t_o)\). Finally, taking the supremum over \(t_1 \in \Lambda _s(t_o)\) in the first term and passing to the limit \(t_1 \uparrow t_o\) in the second term yields inequality (3.1).
In order to prove (3.2) we choose \(\varphi (x,t) = \eta ^p(x) \psi _\varepsilon (t) (u^m(x,t) - a^m)_- \) as testing function in (3.3), where \(\eta \) is defined as before and
for \(t_o-s \le t_1< t_2 < t_o\) and \(\varepsilon >0\) small enough. The term involving the time derivative of \([\![u]\!]_h\) is treated as in (3.4). Thus, we find that
for any \(t_o-s \le t_1< t_2 < t_o\). For the diffusion term and the right side the same arguments as in the proof of (3.1) are applicable. Therefore by passing to the limits \(h \downarrow 0\) and \(\varepsilon \downarrow 0\) we obtain
for any \(t_o-s \le t_1< t_2 < t_o\). Omitting the second term on the left side, choosing \(t_1 = t_o-s\) and taking the supremum over \(t_2 \in \Lambda _s(t_o)\) leads to (3.2). \(\square \)
Similarly, we obtain energy estimates for sub-solutions. However, in the course of the paper we only need the analogue of (3.1).
Lemma 3.2
Under the assumptions of Lemma 3.1 we obtain for any non-negative weak sub-solution to (1.3) the energy estimate
for a constant \(c=c(p,\nu ,L)\).
Proof
The proof is analogous to the one of the energy estimate (3.1). Here, we choose the testing function
with the positive part of \(u^m-a^m\) instead of the negative one. Similar arguments as in the proof of (3.1) then lead us to inequality (3.5). \(\square \)
4 Local boundedness of non-negative weak sub-solutions
In this section we establish that non-negative weak sub-solutions to (1.3) are locally bounded. We argue by a parabolic version of De Giorgi classes.
Theorem 4.1
Let \(m>0\) and \(p>1\) with \(m(p-1) > 1\). Assume that u is a non-negative weak sub-solution to (1.3) in the sense of Definition 1.1 and \(F \in L^\sigma (\Omega _T)\) with \(\sigma > \frac{n+p}{p-1}\). Then u is locally bounded in \(\Omega _T\) and for any cylinder \(Q_0:=Q^-_{\varrho ,\theta }(z_o) \Subset \Omega _T\) with \(0< \varrho ,\theta \le 1\) the quantitative estimate
holds true, where \(\frac{1}{2} Q_0:= Q^-_{\frac{\varrho }{2},\frac{\theta }{2}}(z_o)\) and c is a constant depending on \(n,m,p,\nu ,L\) and \(\sigma \).
Proof
Let \(m' := \frac{m+1}{m}\) denote the conjugate Hölder exponent of \(m+1\). For \(i \in {\mathbb {N}}_0\) we define radii \(\varrho _i\) and times \(\theta _i\) by
Throughout the proof, we use the short-hand notation
Furthermore, for a quantity \(k \ge 1\) to be chosen later on, we consider levels
and the sequence of integrals
Since \(u^m \in L^p(\Omega _T)\) by definition, \(Y_i\) is finite for any \(i \in {\mathbb {N}}_0\). The idea of proof is to show a recursive estimate for \(Y_i\). To this aim we first use Hölder’s inequality to obtain
where the definition of \(\mathrm {I}\) is clear in this context. First, by the Gagliardo–Nirenberg inequality from Lemma 2.9 we infer
for a constant \(c = c(n,m,p)\). We now consider the integrand in the first integral on the right-hand side. For \(u\ge k_{i+1}\) we have with the abbreviation
that
and
Therefore, in view of Lemma 2.7 (ii) we obtain
Using this inequality above and applying the Caccioppoli inequality (3.5) from Lemma 3.2, yields
for a constant \(c = c(n,m,p,\nu ,L)\). For \(u>{\tilde{k}}_i\) we now estimate the \({\mathfrak {b}}\)-term with the help of Lemma 2.7 (i), the assumption \(m+1\le mp\) and the fact that \(k_i<{\tilde{k}}_{i} < k\) with \(k \ge 1\). In this way we obtain
with \(c=c(p)\), so that
where \(c = c(n,m,p,\nu ,L)\). Further, we have that
which together with \(k \ge 1\) implies that
Finally, the preceding computations together with \(0<\varrho \le 1\) and \(Y_i \le \Vert u\Vert _{L^{mp}(Q_0)}^{mp}\) lead to
with a constant \(c=c(n,m,p,\nu ,L)\). Inserting this inequality into (4.1) and using (4.2), we conclude that
where we used the abbreviations
Since \(\sigma > \frac{n+p}{p-1}\), we have that \(\gamma >0\). Choosing \(k \ge 1\) large enough, such that
with a suitable constant \(c=c(n,m,p,\nu ,L,\sigma )\), we find that
Thus, the assumptions of Lemma 2.8 are satisfied. Consequently we find that \(Y_i \rightarrow 0\) as \(i \rightarrow \infty \), which implies \(u\le k\) a.e. in \(\frac{1}{2} Q_0\). The claim of the theorem now follows by an application of Young’s inequality. \(\square \)
5 De Giorgi type lemmas
In this section we will prove certain De Giorgi type lemmata for weak sub- and super-solutions. We start with the one for super-solutions.
Lemma 5.1
Let \(m >0\), \(p>1\) with \(m(p-1) > 1\) and u be a bounded non-negative weak super-solution to (1.3) in the sense of Definition 1.1, where the vector-field \({\mathbf {A}}\) satisfies (1.4) and \(F \in L^\sigma (\Omega _T)\) for some \(\sigma >\frac{n+p}{p-1}\). Moreover, consider \(z_o\in \Omega _T\) and \(\varrho ,\theta ,M>0\) such that
Then, there exists \(\nu _1\in (0,1)\) depending only on \(n,m,p,\nu ,L,\sigma \) and \(\theta \), such that: If
then
holds true.
Proof
For \( i \in {\mathbb {N}}_0\) define radii \(\varrho _i\) and times \(\tau _i\) by
as well as levels
To shorten notation, we introduce
At this stage, we use the Caccioppoli inequality (3.1). Since \(0 \le u < k_i\) on \(A_i\) and \(\frac{M}{2}\le k_i\le M\) and by Lemma 2.7 (ii), we estimate the term involving \({\mathfrak {b}}\) on the left-hand side by
while for the one on the right-hand side we obtain by Lemma 2.7 (i) that
Thus, we conclude that
where in the second last line we used assumption (5.1). Note that \(c=c(m,p,\nu ,L,\theta )\). Next, we use Hölder’s inequality with exponents \(\frac{n+2}{n}\) and \(\frac{n+2}{2}\), the Gagliardo–Nirenberg inequality from Lemma 2.9 with \(r=2\) and p and the preceding estimate. This leads to
with a constant \(c=c(n,m,p,\nu ,L)\). Moreover, due to Lemma 2.6 we have
so that
Combing the preceding estimates yields
with a constant \(c=c(n,m,p,\nu ,L,\theta )\). Dividing the above inequality by \(|Q_{i+1}|\), using the fact that \(\frac{|Q_i|}{|Q_{i+1}|} = c(n,p)\) shows that
where c depends only on \(n,m,p,\nu ,L,\theta \). This brings us into the position to apply Lemma 2.8 with \(\kappa = c\), \(b = 2^{p \left( m + \frac{n+p}{n+2}\right) }\) and \(\gamma = \frac{p}{n+2} - \frac{p(n+p)}{\sigma (n+2)(p-1)}>0\) (since \(\sigma > \frac{n+p}{p-1}\)), where \(\nu _1\in (0,1)\) can be chosen in dependence on the data. This shows \(Y_i \rightarrow 0\) as \(i\rightarrow \infty \), which yields the claim. \(\square \)
Now we turn our attention to the De Giorgi type lemma for sub-solutions.
Lemma 5.2
Let \(m >0\), \(p>1\) with \(m(p-1) > 1\) and u be a bounded non-negative weak sub-solution to (1.3) in the sense of Definition 1.1, where the vector-field \({\mathbf {A}}\) satisfies (1.4) and \(F \in L^\sigma (\Omega _T)\) for some \(\sigma >\frac{n+p}{p-1}\). Moreover, consider \(z_o\in \Omega _T\), \(\varrho , \theta ,M, \mu _+>0\) and \(a,\zeta \in (0,1)\), such that
and
Then, there exists \(\nu _2\in (0,1)\) depending only on \(n,m,p,\nu ,L,\sigma ,\theta ,a\) and \(\zeta \) such that: If
and
then
Proof
As before, we define for \(i\in {\mathbb {N}}_0\)
as well as levels
and sets
In the following we will apply the Caccioppoli inequality (3.5) from Lemma 3.2. Using the definition of \(A_i\) and Lemma 2.7 (i), (ii) and the fact that \(\frac{1}{c(\zeta )}M\le k_i\le u\le M\) on \(A_i\), we estimate the terms involving \({\mathfrak {b}}\) by
Thus, by the Caccioppoli inequality (3.5) and assumption (5.4), we obtain
for a constant \(c=c(m,p,\nu ,L,\theta ,\zeta )\). Similarly as before, we use Hölder’s inequality, the Gagliardo–Nirenberg inequality from Lemma 2.9 with \(r=2\) and p and the last estimate to conclude
for a constant \(c=c(n,m,p,\nu ,L,\theta ,\zeta )\). Notice that
Combining the preceding two estimates leads to
with a constant \(c=c(n,m,p,\nu ,L,\theta ,a,\zeta )\). By completely the same reasoning as in the proof of Lemma 5.1 we infer that \(Y_i \rightarrow 0\) as \(i\rightarrow \infty \), provided we choose \(\nu _2\in (0,1)\) small enough in dependence on the data. \(\square \)
6 Expansion of positivity
In this section, we prove the so called Expansion of Positivity of a non-negative weak super-solution u. The Expansion of Positivity is crucial in the proof of Harnack’s inequality. In a first step we show the following lemma, which ensures a certain propagation of positivity in measure.
Lemma 6.1
Let \(m >0\), \(p>1\) with \(m(p-1) > 1\) and u a non-negative weak super-solution to (1.3), and let \(\alpha \in (0,1]\) and \(M>0\). Then, there exist \(\varepsilon = \varepsilon (m,\alpha ) \in (0,1)\) and \(\delta = \delta (m,p,\nu ,L,\alpha ) \in (0,1)\) such that the following holds: Whenever \(z_o=(x_o,t_o)\in \Omega _T\) and \(\varrho >0\) such that \(Q^+_{\varrho ,\delta M^{-d}\varrho ^p}(z_o)\subset \Omega _T\) and
and
are satisfied, then
Proof
In the following we abbreviate \(Q_0:= Q^+_{\varrho ,\delta M^{-d}\varrho ^p}(z_o)\) with \(\delta \in (0,1)\) to be chosen later. The idea of the proof is to show that if (6.1) and (6.2) are valid, then
holds true for all \(t \in [ t_o,t_o+\delta M^{-d} \varrho ^p )\), which is equivalent to (6.3). Therefore in a first step we let \(s \in (0,1)\) and compute
To estimate the first term on the right hand side we use the Caccioppoli inequality (3.2) from Lemma 3.1. Taking \(r=(1-s)\varrho \) and \(a=M\) leads to
for any \(t \in [t_o,t_o+\delta M^{-d} (1-s)^p\varrho ^p)\) with a constant \(c=c(p,\nu ,L)\). Recalling the definition of the boundary term \({\mathfrak {b}}\) from (2.1) we estimate the left hand side by
for \(\varepsilon \in (0,\frac{m}{m+1})\) to be chosen later. For the first term on the right-hand side, we use again the definition of \({\mathfrak {b}}\) and assumption (6.1) to obtain
Further, we have that
and in view of assumption that (6.2) we obtain
Altogether this leads to
which is the same as
Combining the last estimate with (6.4), and taking into account that \(0<1-\varepsilon \tfrac{m+1}{m} < 1\), we get
for any \(t \in [ t_o,t_o+\delta M^{-d} \varrho ^p )\) with \(c=c(p,\nu ,L)\). Now we choose \(s=\frac{\alpha }{8n} \in (0,1)\) and thereafter \(\delta =\delta (m,p,\nu ,L,\alpha )\) small enough to ensure \(c \delta \frac{m+1}{m} (s^{-p} + 1) \le \frac{\alpha }{8}\). This leads to
for all \(t \in [ t_o,t_o+\delta M^{-d} \varrho ^p )\). Choosing
we conclude the proof. \(\square \)
Remark 6.2
From the proof of Lemma 6.1 we observe that \(\varepsilon \) and \(\delta \) are monotonically increasing with respect to \(\alpha \).
The preceding lemma at hand, we are now able to prove the Expansion of Positivity for non-negative weak super-solutions to the doubly degenerate equation (1.3).
Proposition 6.3
(Expansion of Positivity) Let \(m >0\), \(p>1\) with \(m(p-1) > 1\) and u be a non-negative weak super-solution to (1.3). For fixed \(\alpha \in (0,1]\) there exist constants \(b, \kappa \in (0,1)\) and \(c\ge 1\) depending only on n, m, p, \(\nu \), L, \(\sigma \) and \(\alpha \) such that the following holds true: We consider \(z_o=(x_o,t_o)\in \Omega _T\), \(M>0\) and
Supposed that
and
are satisfied, then we have
Proof
The proof of Proposition 6.3 is divided into several steps. Throughout the proof we denote by \(\varepsilon = \varepsilon (m,\alpha ) \in (0,1)\) and \(\delta = \delta (m,p,\nu ,L,\alpha ) \in (0,1)\) the constants from Lemma 6.1.
6.1 Application of lemma 6.1
For \(j_\star \in {\mathbb {N}}\) to be chosen later in dependence on n, m, p, \(\nu \), L, \(\sigma \) and \(\alpha \) we define
and
Note that
Now we fix \(\varrho \in (0,\varrho _0]\) and assume that (6.6) is satisfied and that
Then, the assumptions of Lemma 6.1 are fulfilled with M replaced by sM for any \(s \in [s_0,1]\). Thus, we find that
for all \(s\in [s_0,1]\) and all \(t \in [t_o,t_o+\delta (sM)^{-d} \varrho ^p)\).
6.2 Transforming to another problem
For \(\tau \ge 0\) we let \(s(\tau ) := e^{-\frac{\tau }{d}}\). Then, we have \(s(\tau ) \in [s_0,1]\) for \(\tau \in [0,\tau _0]\), where
Next, we define
From Step 6.1 we deduce that for any \(\tau \in [0,\tau _0]\) there holds
and
In particular, letting
we have that
Finally, we let \(\gamma (\tau ) := {\widetilde{\Phi }}(\tau )^\frac{1}{d}\). Then, Corollary 2.4 ensures that
is a non-negative weak super-solution to
with
and
Using the growth assumptions (1.4) of \({\mathbf {A}}\) together with the definition of the functions \(\Phi \) and \(\gamma \) we compute that \(\widehat{{\mathbf {A}}}\) satisfies the growth and ellipticity conditions
Defining
we observe that \(k_0=\varepsilon \gamma (\tau ) s(\tau ) M\) for any \(\tau \in [0,\tau _0]\) and therefore inequality (6.9) can be rewritten as
6.3 Gradient estimates on intrinsic sublevel sets
Next, we define
and consider cylinders
for \(0<r \le 8\varrho \). Moreover, for \(j=1,\ldots ,j_\star \), we let
and observe that
Further, a simple computation shows that
By definition of \(\widehat{F}\) and Hölder’s inequality, we obtain for \(q\in [1,\sigma ]\) that
where in the second last line we used the area formula and the fact that \(\vartheta (4\varrho )^p=\frac{4^p}{\delta }(\frac{2^{j_\star }}{\varepsilon })^{d}\), so that \(c=c(n,m,p,\nu ,L,\sigma ,\alpha ,j_\star )\). Assuming that
for some constant \(c_\star \ge 1\) to be chosen later, we further estimate
again with a constant \(c=c(n,m,p,\nu ,L,\sigma ,\alpha ,j_\star )\). Therefore, we may choose \(c_\star \) in dependence on \(n,m,p,\nu ,L,\sigma ,\alpha \) and \(j_\star \) in such a way that
holds true for any \(q\in [1,\sigma ]\). Note that we replaced \(\varrho \) by \(2\varrho \) in the denominator and by for later purpose. We observe that
Thus, the Caccioppoli inequality (3.1) from Lemma 3.1 together with Lemma 2.7 (i), estimate (6.12) with \(q=\frac{p}{p-1}\) and the fact that implies
with \(c=c(n,m,p,\nu ,L)\).
6.4 Measure estimates for intrinsic sublevel sets
Now, we exploit the estimate
with \(c(m) \in (0,1)\) together with Lemma 2.5 and inequality (6.10) to obtain
for all \(j=0,\ldots ,j_\star \) and all \(\tau \in (0, \vartheta (4\varrho )^p)\). We integrate this inequality with respect to \(\tau \) over \((\vartheta (2\varrho )^p, \vartheta (4\varrho )^p)\), apply Hölder’s inequality on the right-hand side and use the gradient bound (6.13) to get
Dividing both sides by \(k_j^m > 0\) and summing over \(j=0,\dots ,j_\star -1\), we find that
so that
for a constant c depending only on n, m, p, \(\nu \) and L.
6.5 Application of De Giorgi type lemma 5.1
At this stage, we exploit Lemma 5.1. Observe that the cylinder \(\widehat{Q}_{4\varrho }^\vartheta = Q_{4\varrho ,\vartheta (4^p-2^p)\varrho ^p}^-(x_o,\vartheta (4\varrho )^p)\) satisfies the requirements of the Lemma with \(\varrho \), \(\theta \) and M replaced by \(2\varrho \), \(\frac{4^p-2^p}{4^p}\) and \(k_{j_\star }\). Then, the constant \(\nu _1\) from Lemma 5.1 depends only on n, m, p, \(\nu \), L and \(\sigma \), but is independent of \(j_\star \). Note that (5.1) is implied by (6.12) applied with \(q=\sigma \). Thus, choosing \(j_\star \) large enough, so that
all assumptions of Lemma 5.1 are satisfied and we conclude that
Note that \(j_\star \) depends on \(n,m,p,\nu ,L,\sigma \) and \(\alpha \). This also fixes \(c_\star \) in (6.11) in dependence on \(n,m,p,\nu ,L,\sigma \) and \(\alpha \). In turn, we choose \(c \ge 1\) in dependence on \(n,m,p,\nu ,L,\sigma \) and \(\alpha \) in such a way that condition (6.7) implies the validity of (6.11) and (6.8).
6.6 Returning to the original problem and conclusion
Finally we use the definition of v and \(k_0\) to rewrite (6.15) as
for a.e. \((x,\tau )\in B_{2\varrho } \times \left( (4^p-2^p+1) \vartheta \varrho ^p, \vartheta (4\varrho )^p \right] \), where
Returning to the original time variable, we obtain
with \(b := \delta \varepsilon ^d2^{-(j_\star +1)d} \in (0,1)\) and \(\beta := e^{-(2^p-1) \vartheta \varrho ^p}\) depending only on the data. Note that by the definitions of b and \(\kappa \) we have \(\frac{1}{\delta }\exp \big (-\frac{2^{2p+j_\star d}}{\delta \varepsilon ^{d}}\big )=\frac{\kappa ^d}{b}\), so that \(\varrho _0\) can be re-written exactly as in (6.5). Since \(\beta \le \frac{1}{2}\) this completes the proof of Proposition 6.3. \(\square \)
Remark 6.4
From the proof of Proposition 6.3 we observe that
Moreover, the parameter b in Proposition 6.3 is monotonically increasing with respect to \(\alpha \). This can be seen from the definition \(b= \delta \varepsilon ^d2^{-(j_\star +1)d}\), where \(j_\star \) is decreasing and \(\varepsilon \) and \(\delta \) are increasing with respect to \(\alpha \); see Remark 6.2.
7 Harnack’s inequality
We are now ready to prove our main result, Theorem 1.2. In the following section, the second (forward in time) inequality of (1.8) is shown. In a subsequent step, we ensure the validity of the first (backward in time) inequality of (1.8).
7.1 Forward inequality
Let \(c_o \ge 1\) to be fixed later, consider \((x_o,t_o) \in \Omega _T\) with \(u(x_o,t_o)>0\) and define
Moreover, assume that \(\varrho > 0\) is small enough so that \(B_{9\varrho }(x_o) \times (t_o - 2\theta \varrho ^p , t_o + 2\theta \varrho ^p) \Subset \Omega _T\). Note that the stronger assumption \(B_{9\varrho }(x_o) \times (t_o - 4\theta \varrho ^p , t_o + 4\theta \varrho ^p) \Subset \Omega _T\) will only be needed in the proof of the backward Harnack inequality. Finally, we define the rescaled function
where \((\tilde{x},\tilde{t}) :\widehat{\Omega _T} \rightarrow \Omega _T\) with \(\widehat{\Omega _T} := \{(x,t) \in {\mathbb {R}}^{n+1}: (\tilde{x},\tilde{t}) \in \Omega _T\}\) is defined by
A straightforward computation shows that v is a bounded, continuous, non-negative weak super-solution of
in \(B_9(0) \times (-2 c_o^{d}, 2 c_o^{d})\) in the sense of Definition 1.1 with
and
The main step towards Theorem 1.2 is the following lemma. After returning to the original variables this proves the intrinsic forward Harnack inequality, i.e. the second inequality of (1.8). Indeed, if Lemma 7.1 is valid, we obtain that
or
which shows the second inequality of (1.8) for \(\gamma =\max \big \{\frac{1}{\gamma _0|B_1(0)|^{1/\sigma }}, \frac{1}{\gamma _1}\big \}\).
Lemma 7.1
For v, \(\tilde{{\mathbf {A}}}\) and \(\tilde{F}\) as above, there exist constants \(\gamma _0, \gamma _1 \in (0,1)\) and \(c_o > 1\) depending only on the data, but independent of \(u(x_o,t_o)\) such that either
or
Proof
In the following we abbreviate \(Q^-_r:=Q^-_{r,r^p}(0)=B_r(0)\times (-r^p,0]\) for \(r>0\). For \(\tau \in [0,1)\) we consider the family of cylinders \(\{ Q_\tau ^- \}\) and the functions \(M,N :[0,1) \rightarrow [0,\infty )\) defined by
with \(\delta >1\) to be chosen later on. Note that the functions M and N are both monotonically increasing and \(M_0 = 1 = N_0\), since \(v(0,0)=1\). Moreover, as \(\tau \uparrow 1\), \(N(\tau ) \rightarrow \infty \) while \(M(\tau )\) remains bounded, since v is bounded in \(Q_1^-\). Together with the continuity of v this ensures that there exist
and \((x_\star ,t_\star ) \in Q_{\tau _\star }^-\) such that
Let \(\tilde{n} \in {\mathbb {N}}_{\ge 2}\) such that \(2^{1-\tilde{n}} < 1- \tau _\star \le 2^{2-\tilde{n}}\) and define \(r:=2^{-\tilde{n}}\). Then \(\tau _\star + r < \tau _\star +\frac{1}{2}(1-\tau _\star ) =\frac{1+\tau _\star }{2}\), which implies
Moreover, by definition of M, N and \(\tau _\star \) we have
Observe that \(M_\star >1\). Next, on the cylinder \(Q^-_{r,M_\star ^{-d} r^p}(x_\star , t_\star ) \subset (x_\star ,t_\star )+Q_r^-\) we apply the De Giorgi type Lemma 5.2 to v with
and \((\mu _+,M,\theta ,\varrho )\) replaced by \((M_\star ,M_\star ,1,\tfrac{r}{2})\). Indeed, hypothesis (5.2) is satisfied, since
By \({\tilde{\nu }}\) we denote the constant \(\nu _2\) from Lemma 5.2 depending on \(n,m,p,\nu ,L,\theta ,a,\zeta \); hence \({\tilde{\nu }}={\tilde{\nu }}(n,m,p,\nu ,L,\delta )\). Moreover, observe that
This shows that conclusion (5.5) of Lemma 5.2 is false. Hence, either (5.3) or (5.4) is violated. This means, we either have
or
If (7.4) is satisfied, by Fubini’s theorem there exists \({\bar{t}}_\star \in (t_\star -M_\star ^{-d}r^p,t_\star ]\) with
By \(\tilde{b}, {\tilde{\kappa }}\in (0,1)\) and \(\tilde{c}\ge 1\) we denote the constants \(b,\kappa ,c\) from the Expansion of Positivity in Proposition 6.3 applied with \(\alpha ={{\tilde{\nu }}}\). Note that \(\tilde{b}, {\tilde{\kappa }}\) and \(\tilde{c}\) depend on \(n, m, p, \nu , L,\sigma \) and \(\delta \). Supposed that
we are allowed to apply Proposition 6.3 with \((F,\alpha ,M,\varrho )\) replaced by \((\tilde{F},{\tilde{\nu }},2^{-4\delta } M_\star ,r)\) and conclude that either
or
holds true. In the second case we find that
where \(\tilde{t}_o := {\bar{t}}_\star + \tilde{b} \big ( 2^{-4\delta } {{\tilde{\kappa }}} M_\star \big )^{-d} r^p\). This allows to apply the Expansion of Positivity in the next step with \(\alpha =1\). Therefore, by \(b, \kappa \in (0,1)\) and \(c\ge 1\) we denote the constants \(b,\kappa ,c\) from Proposition 6.3 applied with \(\alpha =1\). Then, \(b, \kappa \) and c depend on \(n, m, p, \nu , L\) and \(\sigma \), but not on \(\delta \). Supposed that
we may apply Proposition 6.3 with \((F,\alpha ,M,\varrho )\) replaced by \((\tilde{F},1,2^{-4\delta } {\tilde{\kappa }}M_\star ,2r)\) and conclude that either
or
holds true, where \(t_1 := {\tilde{t}}_o + b \big (2^{-4\delta } {\tilde{\kappa }}\kappa M_\star \big )^{-d} (2r)^p\). In the second case, we have
We recursively define \(t_2,\ldots ,t_{\tilde{n}}\) by
for \(j \in \{2,\ldots ,\tilde{n}\}\). Iterating the procedure of Expansion of Positivity we arrive at the following assertion. Supposed that
for every \(j=2,\ldots ,\tilde{n}\), we find that either
is satisfied for some \(j\in \{2,\ldots ,\tilde{n}\}\) or
We first ensure that (7.10) is satisfied for \({\tilde{n}}\). We note that \(2^{\tilde{n}}r=1\). Since \(x_\star \in B_1(0)\), we immediately observe that \(2^{\tilde{n}}r=1\le \tfrac{1}{8}{{\,\mathrm{dist}\,}}(x_\star ,\partial B_9(0))\). Next, we choose \(\delta > 1\) in dependence on \(n,m,p,\nu ,L\) and \(\sigma \) such that \(2^\delta \kappa =1\), which is possible, since \(\kappa \) is independent of \(\delta \). In view of the definition of \(M_\star \) we find that
Note that \(\gamma _1\) depends on \(n,m,p,\nu ,L\) and \(\sigma \). The second condition in (7.10) is equivalent to \(t_{{\tilde{n}}}\le 2 c_o^d\). Therefore, we compute
We note that due to Remark 6.4 we have \({\tilde{b}}\le b\) and therefore the expression \(\frac{2^p b}{2^p-\kappa ^d} - \tilde{b}\) is positive. Hence, choosing \(c_o\) such that
and taking into account that \({\bar{t}}_\star \le 0\) we find that
Provided that (7.13) holds true, (7.10) is satisfied for \({\tilde{n}}\) and in turn implies that (7.10) is satisfied for any \(j=2,\ldots ,\tilde{n}\) and in particular also (7.5) and (7.8) are satisfied.
To summarize, we have now shown that either (7.12) is satisfied or one of the alternatives (7.6), (7.9) or (7.11) if \(c_o\) is chosen large enough. We start with the former case where (7.12) is satisfied. Since \(2^{\tilde{n}}r=1\) we have \(B_1(0) \subset B_2(x_\star ) = B_{2^{\tilde{n}+1}r}(x_\star )\), so that
Unfortunately, the interval depends on \(\tilde{n}\) and hence on v. Therefore, we need to find a subinterval which is independent of \(\tilde{n}\). In view of Remark 6.4 we have \(\frac{b}{{{\tilde{\kappa }}}^d}\ge \frac{{\tilde{b}}}{{{\tilde{\kappa }}}^d}>4^{p}\) and hence \(\gamma _1^d=2^{-4 \delta d} {\tilde{\kappa }}^d<2^{-2p-4 \delta d} b<2^{-2p} b\). Therefore, we observe from the preceding computation of \(t_{{\tilde{n}}}\) that
In the second term on the right-hand side we use \(\kappa< 4^{-p}b<4^{-p}\), which once again is a consequence of Remark 6.4. This leads us to the lower bound
For the left interval limit in (7.14) we obtain
The preceding computations show that with the choice
we have
Note that \(c_o\) depends on \(n,m,p,\nu ,L\) and \(\sigma \) and a straightforward calculation shows that (7.13) is satisfied. From (7.14) we now conclude that
which concludes the proof of the lemma in the case that (7.12) is satisfied.
Finally, we are left with the case where one of the alternatives (7.6), (7.9) or (7.11) is satisfied. In any of these cases we conclude that
is valid. We note that \(\gamma _0\in (0,1)\), since \({{\tilde{\kappa }}}\le \kappa \) by Remark 6.4 and that \(\gamma _0\) depends on \(n,m,p,\nu ,L\) and \(\sigma \). This concludes the proof of the lemma. \(\square \)
7.2 Backward inequality
With the intrinsic forward Harnack inequality on hand, we are able to show the intrinsic backward Harnack inequality, i.e. the first inequality of (1.8). Actually, in the following we will prove the more general version
with positive constants \(c_1,c_2\) such that \(c_1\gamma ^{c_2-2} > 1\), which implies the first inequality of (1.8) by choosing \(c_1 = c_2 = 2\). We have already fixed \((x_o,t_o) \in \Omega _T\) with \(u(x_o,t_o)>0\). Now we assume that \(B_{9\varrho }(x_o) \times (t_o - 4\theta \varrho ^p, t_o + 4 \theta \varrho ^p) \Subset \Omega _T\). Moreover, let \(c_1,c_2>0\) be positive constants such that \(c_1\gamma ^{c_2-2}>1\) and suppose that alternative (1.7) is not valid, i.e.
In order to prove the backward Harnack inequality, we consider two alternatives. First, we assume that
with \(\gamma \) as in the right-hand side of (1.8). Our aim is to prove that (7.18) implies
Indeed, assume that (7.19) was not satisfied. Then there exists \(x_\star \in B_\varrho (x_o)\) such that \(u(x_\star ,t_1) = c_1\gamma ^{c_2} u(x_o,t_o)\), where we abbreviated \(t_1 := t_o-(c_1\gamma ^{c_2-1})^{-d}\theta \varrho ^p\), since u is continuous and (7.18) is in force. Let \(\theta _\star := c_o^{d}u(x_\star ,t_1)^{-d}\). A simple calculation shows that
Since \(d>0\), \(\gamma >1\) and \(c_1\gamma ^{c_2-1}>1\), this implies \((t_1 - 2 \theta _\star \varrho ^p, t_1 + 2 \theta _\star \varrho ^p) \subset (t_o - 4 \theta \varrho ^p, t_o + 4 \theta \varrho ^p) \Subset \Omega _T\). Thus, we are able to apply the forward Harnack inequality with \((x_o,t_o)\) replaced by \((x_\star , t_1)\). This leads to
which contradicts (7.17), or
In view of (7.18) and the facts that \(x_o \in B_\varrho (x_\star )\) and \(t_1 + \theta _\star \varrho ^p < t_o\), this yields the contradiction
Therefore (7.18) implies (7.19).
It remains to treat the case where (7.18) is violated. This means that there exists \(t \in (t_o - 2\theta \varrho ^p,t_o)\) such that \(u(x_o,t) = c_1\gamma ^{c_2-1} u(x_o,t_o)\). We define \(\tau \) as the largest value with this property (note that u is continuous) and let
We claim that
Indeed, if (7.20) was not valid, there existed \(0 < {\tilde{\varrho }} \le \varrho \) such that
Computing that \((\tau - 2 \theta _\tau {\tilde{\varrho }}^p, \tau + 2 \theta _\tau {\tilde{\varrho }}^p) \subset (t_o - 4 \theta \varrho ^p, t_o + 4 \theta \varrho ^p)\), we are allowed to apply the forward Harnack inequality with \((x_o,\tau )\) instead of \((x_o,t_o)\). This gives that either
or
holds true. The former one contradicts (7.17), while the latter one contradicts \(c_1\gamma ^{c_2-2}>1\). Therefore, (7.20) is valid. Next, we define
By definition of \(\tau \) and (7.20), we find that
In the following we show by contradiction that
Indeed, otherwise by the continuity of u there existed \(y \in B_\varrho (x_o)\) with \(u(y,s) = c_1 \gamma ^{c_2-1} u(x_o,t_o)\). For \(\theta _s := c_o^{d}u(y,s)^{-d}\) we have that \((s - 2 \theta _s \varrho ^p, s + 2 \theta _s \varrho ^p) \subset (t_o - 4 \theta \varrho ^p, t_o + 4 \theta \varrho ^p)\). Thus, applying the forward Harnack inequality with (y, s) instead of \((x_o,t_o)\) leads to
which contradicts (7.17), or
Since \(s+\theta _s \varrho ^p = t_o\) and \(y \in B_\varrho (x_o)\), we obtain the contradiction
Therefore (7.21) is valid. Recalling the definition of s, we conclude that the desired backwards Harnack inequality is in force also in this case. This finishes the proof of inequality (7.16) and thus the proof of Theorem 1.2.
References
Acerbi, E., Fusco, N.: Regularity for minimizers of nonquadratic functionals: the case \(1<p<2\). J. Math. Anal. Appl. 140(1), 115–135 (1989)
Bögelein, V., Lukkari, T., Scheven, C.: Hölder regularity for degenerate parabolic obstacle problems. Ark. Mat. 55(1), 1–39 (2017)
Bögelein, V., Duzaar, F., Kinnunen, J., Scheven, C.: Higher integrability for doubly nonlinear parabolic systems. J. Math. Pures Appl. 143, 31–72 (2020)
Bögelein, V., Duzaar, F., Korte, R., Scheven, C.: The higher integrability of weak solutions of porous medium systems. Adv. Nonlinear Anal. 8(1), 1004–1034 (2019)
Bögelein, V., Duzaar, F., Marcellini, P.: Parabolic systems with p; q-growth: a variational approach. Arch. Ration. Mech. Anal. 210(1), 219–267 (2013)
Bögelein, V., Lehtelä, P., Sturm, S.: Regularity of weak solutions and supersolutions to the porous medium equation. Nonlinear Anal. 185, 49–67 (2019)
DiBenedetto, E.: Intrinsic Harnack type inequalities for solutions of certain degenerate parabolic equations. Arch. Ration. Mech. Anal. 100(2), 129–147 (1988)
DiBenedetto, E.: Degenerate parabolic equations, Springer Science & Business Media (1993)
DiBenedetto, E., Gianazza, U., Vespri, V.: Harnack estimates for quasi-linear degenerate parabolic differential equations. Acta Math. 200(2), 181–209 (2008)
DiBenedetto, E., Gianazza, U., Vespri, V.: Harnack’s Inequality for Degenerate and Singular Parabolic Equations. Springer Monographs in Mathematics. Springer, New York (2012)
Düzgün, F.G., Fornaro, S., Vespri, V.: Interior Harnack estimates: the state-of-the-art for quasilinear singular parabolic equations. Milan J. Math. 83(2), 371–395 (2015)
Fornario, S., Sosio, M.: Intrinsic Harnack estimates for some doubly nonlinear degenerate parabolic equations. Adv. Diff. Equ. 13(1–2), 139–168 (2008)
Fornario, S., Sosio, M., Vespri, V.: Harnack type inequalities for some doubly nonlinear singular parabolic equations. Discr. Cont. Dyn. Syst. A 35(12), 5909–5926 (2015)
Gianazza, U., Vespri, V.: A Harnack inequality for solutions of doubly nonlinear parabolic equations. J. Appl. Funct. Anal. 1(3), 271–284 (2006)
Giaquinta, M., Modica, G.: Remarks on the regularity of the minimizers of certain degenerate functionals. Manuscripta Math. 57(1), 55–99 (1986)
Hadamard, J.: Extension à l’èquation de la chaleur d’un théorème de A. Harnack. Rend. Circ. Mat. Palermo 2(3), 337–346 (1954)
Ivanov, A.V.: Hölder estimates for a natural class of equations of fast diffusion type, (Russian) Zap. Nauchn. Sem. S.-Peterburg. Otdel. Mat. Inst. Steklov. (POMI) 229, : Chisl. Metody i Voprosy Organ. Vychisl. 11(29–62), 322 (1995)
Ivanov, A.V.: Hölder estimates for a natural class of equations of fast diffusion type, translation in. J. Math. Sci. (New York) 89(6), 1607–1630 (1998)
Ivanov, A.V.: Hölder estimates for equations of slow and normal diffusion type, (Russian) Zap. Nauchn. Sem. S.-Peterburg. Otdel. Mat. Inst. Steklov. (POMI) 215, : Differentsial’naya Geom. Gruppy Li i Mekh. 14(130–136), 311 (1994)
Ivanov, A.V.: Hölder estimates for equations of slow and normal diffusion type, translation in. J. Math. Sci. (New York) 85(1), 1640–1644 (1997)
Kinnunen, J., Lindqvist, P.: Pointwise behaviour of semicontinuous supersolutions to a quasilinear parabolic equation. Ann. Mat. Pura Appl. 185(3), 411–435 (2006)
Kinnunen, J., Kuusi, T.: Local behaviour of solutions to doubly nonlinear parabolic equations Math. Annals 337(3), 705–728 (2007)
Kuusi, T.: Harnack estimates for weak supersolutions to nonlinear degenerate parabolic equations. Ann. Sc. Norm. Super. Pisa Cl. Sci. (5) (4), 673–716 (2008)
Li, Q.: Weak Harnack estimates for supersolutions to doubly degenerate parabolic equations. Nonlinear Anal. 170, 88–122 (2018)
Moser, J.: On Harnack’s theorem for elliptic differential equations. Comm. Pure Appl. Math. 14, 577–591 (1961)
Moser, J.: A Harnack inequality for parabolic differential equations. Comm. Pure Appl. Math. 17, 101–134 (1964)
Pini, B.: Sulla soluzione generalizzata di Wiener per il primo problema di valori al contorno nel caso parabolico. Rend. Sem. Mat. Univ. Padova 23, 422–434 (1954)
Porzio, M.M., Vespri, V.: Hölder estimates for local solutions of some doubly nonlinear degenerate parabolic equations. J. Diff. Equ. 103(1), 146–178 (1993)
Serrin, J.: Local behavior of solutions of quasi-linear equations. Acta Math. 111, 247–302 (1964)
Singer, T., Vestberg, M.: Local boundedness of weak solutions to the diffusive wave approximation of the shallow water equations. J. Diff. Equ. 266(6), 3014–3033 (2019)
Trudinger, N.S.: On Harnack type inequalities and their application to quasilinear elliptic equations. Comm. Pure Appl. Math. 20, 721–747 (1967)
Trudinger, N.S.: Pointwise estimates and quasilinear parabolic equations. Comm. Pure Appl. Math. 21(3), 205–226 (1968)
Vespri, V.: Harnack type inequalities for solutions of certain doubly nonlinear parabolic equations. J. Math. Anal. Appl. 181(1), 104–131 (1994)
Vespri, V., Vestberg, M.: An extensive study of the regularity of solutions to doubly singular equations. Adv. Calc. Var. (2020). https://doi.org/10.1515/acv-2019-0102
Acknowledgements
V. Bögelein has been supported by the FWF-Project P31956-N32 “Doubly nonlinear evolution equations". A. Herán has been supported by the DFG-Project HA 7610/1-1 “Existenz- und Regularitätsaussagen für parabolische Quasiminimierer auf metrischen Maßräumen". L. Schätzler has been supported by Studienstiftung des deutschen Volkes.
Funding
Open access funding provided by Austrian Science Fund (FWF).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by N. Trudinger.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
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/.
About this article
Cite this article
Bögelein, V., Heran, A., Schätzler, L. et al. Harnack’s inequality for doubly nonlinear equations of slow diffusion type. Calc. Var. 60, 215 (2021). https://doi.org/10.1007/s00526-021-02044-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00526-021-02044-z