Abstract
We prove a local higher integrability result for the gradient of a weak solution to degenerate parabolic double-phase systems of p-Laplace type. This result comes with reverse Hölder type estimates. The proof is based on a careful phase analysis, estimates in the intrinsic geometries and stopping time arguments.
Similar content being viewed by others
1 Introduction
This paper discusses the local higher integrability of the spatial gradient of a weak solution to a double-phase parabolic system
where \(z=(x,t)\), \(\Omega _T=\Omega \times (0,T)\) is a space-time cylinder with a bounded open set \(\Omega \subset \mathbb {R}^n\) for \(n\geqq 2\) and \(2\leqq p<q<\infty \). Here \(\mathcal {A}(z,\nabla u):\Omega _T\times \mathbb {R}^{Nn}\longrightarrow \mathbb {R}^{Nn}\) with \(N\geqq 1\) is a Carathéodory vector field satisfying that there exist constants \(0<\nu \leqq L<\infty \) such that
for almost every \(z\in \Omega _T\) and every \(\xi \in \mathbb {R}^{Nn}\). It is further assumed that the source term \(F:\Omega _T\longrightarrow \mathbb {R}^{Nn}\) satisfies
Here we denote \(H(z,s):\Omega _T\times \mathbb {R}^{+ }\longrightarrow \mathbb {R}^+\),
We assume that the non-negative coefficient function \(a:\Omega _T\longrightarrow \mathbb {R}^+\) satisfies
Here \(a\in C^{\alpha ,\frac{\alpha }{2}}(\Omega _T)\) means that \(a\in L^{\infty }(\Omega _T)\) and that there exists a constant \([a]_{\alpha ,\frac{\alpha }{2};\Omega _T}<\infty \) such that
for every \((x,y)\in \Omega \) and \((t,s)\in (0,T)\). For short we denote \([a]_{\alpha }=[a]_{\alpha ,\frac{\alpha }{2};\Omega _T}\).
We summarize the existing related results in the elliptic and parabolic cases. The elliptic double-phase system
in \(\Omega \), where
models a class (p, q)-growth problems related to strongly anisotropic materials in the contexts of homogenization and nonlinear elasticity, see [26,27,28]. The proper function space for weak solutions is \(u\in W^{1,1}(\Omega ,\mathbb {R}^N)\) with
Under (1.4) it has been proved that \(|\nabla u|\in L_{{{\,\textrm{loc}\,}}}^q(\Omega )\) in [13] (see also [21, 22] for the (p, q)-growth problems). Harnack’s inequality, Hölder continuity, gradient Hölder continuity, gradient higher integrability and Calderón–Zygmund type estimates have been discussed in [2, 8, 9, 11] (see also [17, 18]). For applications and more information, we refer to [23, 24]. A standard approach in the elliptic double-phase systems is to consider two cases: for each ball \(B_r(x_0)\subset \Omega \), either
The first condition in (1.5) is called the p-phase and in this case the behavior is similar to the p-Laplace systems in \(B_r(x_0)\). The second condition in (1.5) implies that
and this leads to the behavior similar to the (p, q)-Laplace systems in \(B_r(x_0)\). For this reason the second condition in (1.5) is called the (p, q)-phase.
Parabolic double-phase problems have not been investigated until very recently. The existence of weak solutions to (1.1) has been considered in [7, 25]. These results seem to cover different ranges of exponents already in the stationary case, see [6]. It has been proved in [25] by using the difference quotient method that \(|\nabla u|\in L^q_{{{\,\textrm{loc}\,}}}(\Omega _T)\) under appropriate structural assumptions (see also [1, 4, 5, 10, 14] for the (p, q)-growth problems).
The main result of this paper is an a priori estimate for the gradient of a weak solution to (1.1). We denote \(Q_{r}(z_0)=B_{r}(x_0)\times (t_0-r^2,t_0+r^2)\) and
Theorem 1.1
Assume that (1.2) and (1.3) hold true and let u be a weak solution to (1.1). Then there exist constants \(0<\varepsilon _0=\varepsilon _0( data )\) and \(c=c( data ,\Vert a\Vert _{L^\infty (\Omega _T)})\geqq 1\) such that
for every \(Q_{2r}(z_0)\subset \Omega _T\) and \(\varepsilon \in (0,\varepsilon _0)\).
As far as we are aware this is the first regularity result for parabolic double-phase problems under the general structural conditions (1.2) and (1.3). We consider weak solutions that satisfy a technical assumption \(|\nabla u|\in L^q(\Omega _T)\), see Definition 2.1. It is also possible to obtain the main result under the assumption
by applying a parabolic Lipschitz truncation, see Remark 2.2. This technique is out of the scope of this paper and it is discussed in [19]. This extends the corresponding results for the p-Laplace systems \((a(z)\equiv 0)\) in [20], where a reverse Hölder inequality for the gradient has been proved in p-intrinsic cylinders
by using parabolic Caccioppoli and Poincaré inequalities and a stopping time argument. Appropriate intrinsic cylinders have to be considered for other parabolic systems. The parabolic (p, q)-Laplace system (\(a(z)\equiv a_0\) for some constant \(a_0>0\)) was considered in [16] and the gradient reverse Hölder inequality was proved in (p, q)-intrinsic cylinders
In [3], the gradient higher integrability result has been discussed for the parabolic \(p(\cdot )\)-Laplace type system
in \(\Omega _T\), where \(p(\cdot ):\Omega _T \longrightarrow \mathbb {R}^+\) is a continuous function with
and \(p(\cdot )\) satisfies a logarithmic modulus of continuity condition. In this case the intrinsic cylinders are of the form
Several new features appear in the parabolic double-phase problem (1.1) compared to the p-Laplace systems in [20] and to the (p, q)-case in [16]. The first novelty in our argument is that we provide a new criterion replacing (1.5) in order to be able to adopt the stopping time argument with intrinsic cylinders in [20]. For each point
we consider \(\lambda =\lambda (z_0)>0\) such that \(\Lambda =\lambda ^p+a(z_0)\lambda ^q\). Employing the fact that \(s \rightarrow s^p + a(z_0)s^q\) is strictly increasing and noting that \(z_0\in \{ z\in \Omega _T:|\nabla u(z)|^p>\lambda ^p\}\), we may apply a stopping time argument with the p-intrinsic cylinders. The second novelty is to consider two alternatives: for \(K>1\), either
These are called p-intrinsic and (p, q)-intrinsic cases, respectively. The p-intrinsic case is related to the p-Laplace systems and the (p, q)-intrinsic case is related to the (p, q)-problems. For the double-phase problems we have to consider both of them. We are convinced that this technique will be useful in other regularity results for parabolic doubly-nonlinear problems. In the p-intrinsic case, it is possible to obtain the reverse Hölder inequality in the p-intrinsic cylinders as in [20]. Roughly speaking, we have
in the stopping time argument with a p-intrinsic cylinder. On the other hand, the (p, q)-intrinsic case implies the second condition in (1.5) for a sufficiently large \(K>1\). This leads (1.6) and we have
in the stopping time argument with p-intrinsic cylinder. Consequently, we may apply the (p, q)-intrinsic cylinders to obtain the reverse Hölder inequality. Note that
and \(G_\rho ^\lambda (z_0)\subset Q_\rho ^\lambda (z_0)\) with \(a_0=a(z_0)\). Thus, it is possible to obtain a stopping time argument in the (p, q)-intrinsic cylinders from the stopping time argument in the p-intrinsic cylinders. Finally, the continuity of \(a(\cdot )\) implies the continuity of \(\lambda (\cdot )\) and this enables us to prove a Vitali type covering lemma. The desired estimate follows by using Fubini’s theorem.
2 Energy estimates
We apply the following definition of weak solution:
Definition 2.1
A function \(u:\Omega _T\longrightarrow \mathbb {R}^N\) with
is a weak solution to (1.1), if
for every \(\varphi \in C_0^\infty (\Omega _T,\mathbb {R}^N)\).
Remark 2.2
A more standard assumption on the function space would be
with
However, this assumption does not seem to be enough in the proof of the energy estimate using the Steklov averages, see Lemma 2.3 below. This unexpected challenge does not occur in the elliptic case, since the mollification in time is not needed. It is possible to derive Lemma 2.3 under the natural function space assumption above by a parabolic Lipschitz truncation technique, see [19]. We emphasize that the assumption \(|\nabla u|\in L^q(\Omega _T)\) is only applied in the proof of Lemma 2.3 and it is not needed in the rest of the paper. With this observation Theorem 1.1 holds true also under the natural function space assumption above.
In the rest of this section, we provide three energy estimates. The first lemma is a parabolic double-phase Caccioppoli inequality. In general, the time derivative of a weak solution does not belong to \(L^2\) and does not even exist a priori. To be able to derive a suitable energy estimate, we use the following mollification in time. We define the Steklov average \(f_h\), with \(0<h<T\), of \(f\in L^1(\Omega _T)\) by
For the properties of Steklov averages, we refer to [12].
We apply the following notation. A space-time cylinder in \(\mathbb {R}^{n+1}\) is denoted by
and the integral average of u over \(Q_{R,\ell }(z_0)\) is denoted by
Lemma 2.3
Let u be a weak solution to (1.1). Then there exists a constant \(c=c(n,p,q,\nu ,L)\) such that
for every \(Q_{R,\ell }(z_0)\subset \Omega _T\), with \(R,\ell >0\), \(r\in [R/2,R)\) and \(\tau \in [\ell /2^2,\ell )\).
Proof
Let \(\eta \in C_0^\infty (B_{R}(x_0))\) be a cut-off function with
For \(\tau \in [\ell /2^2,\ell )\), let \(h_0>0\) be sufficiently small so that there exists a cut-off function \(\zeta \in C_0^\infty (I_{\ell -h_0}(t_0))\) with
Let \(t_*\in I_{\tau }(t_0)\) and \(\delta \in (0,h_0)\). We define \(\zeta _\delta \) as
For \(h\in (0,h_0)\), we consider (1.1) in terms of Steklov averages and obtain
in \(B_{R}(x_0)\times I_{\ell -h}(t_0)\). Then we observe that
By applying \(\varphi =[u-u_{Q_{R,\ell }(z_0)}]_h\eta ^q\zeta ^2\zeta _{\delta }\) as a test function in (2.4), we have
We show that \(\textrm{II}\) and \(\textrm{III}\) are finite under the assumption \(|\nabla u|\in L^q(\Omega _T)\). The structural assumptions and the properties of the Steklov average lead to
The first term on the right-hand side is finite as in the case of the parabolic p-Laplace systems. The second term on the right-hand side can be written as
By Hölder’s inequality and the properties of the Steklov average, there exists a constant \(c=c(n)\) such that
This shows that \(\textrm{II}\) is finite if \(|\nabla u|\in L^q(\Omega _T)\). A similar argument applies for \(\textrm{III}\).
Estimate of \(\textrm{I}\): Integration by parts gives
We estimate the first term on the right-hand side of (2.6) by (2.2) and obtain
For the second term on the right-hand side of (2.6), by (2.3) we have
Thus, we get
Estimate of \(\textrm{II}\): It holds that
To estimate the first term in (2.7), we apply (1.2) to get
To estimate the second term in (2.7), we use (1.2) and (2.1) to conclude that
By Young’s inequality, there exists a constant \(c=c(p,q,\nu ,L)\) such that
It follows that
Estimate of \(\textrm{III}\): We apply Young’s inequality as above and obtain
By applying the estimates above in (2.5), we obtain
Since \(t_*\in I_{\ell }(t_0)\) is arbitrary, \(|B_{R}|\approx c(n)|B_{r}|\) and \(|I_{\ell }|\approx |I_{\tau }|\), we get
\(\square \)
The second lemma is a gluing lemma, which enables us to estimate integral averages over time-slices. The spatial integral average of u over \(B_R(x_0)\) is denoted by
Lemma 2.4
Let u be a weak solution to (1.1) and let \(\eta \in C_0^\infty (B_{R}(x_0))\) be a function such that
where \(c=c(n)\). Then there exists a constant \(c=c(n,L)\) such that
for every \(Q_{R,\ell }(z_0)=B_{R}(x_0)\times (t_0-\ell ,t_0+\ell )\subset \Omega _T\) with \(R,\ell >0\).
Proof
Let \(t_1,t_2\in (t_0-\ell ,t_0+\ell )\) with \(t_1<t_2\). For \(\delta \in (0,1)\) small enough, we define \(\zeta _\delta \in W_0^{1,\infty }(t_0-\ell ,t_0+\ell )\) by
By applying \(\eta \zeta _\delta \in W_0^{1,\infty }(Q_{R,\ell }(z_0))\) as a test function in (1.1), we obtain
Letting \(\delta \longrightarrow 0^+\) and using the third condition in (2.8), we obtain
where \(c=c(n,L)\). This completes the proof. \(\quad \square \)
Then we consider a parabolic Poincaré inequality.
Lemma 2.5
Let u be a weak solution to (1.1). Then there exists a constant \(c=c(n,N,m,L)\) such that
for every \(Q_{R,\ell }(z_0)=B_{R}(x_0)\times (t_0-\ell ,t_0+\ell )\subset \Omega _T\) with \(R,\ell >0\), \(m\in (1,q]\) and \(\theta \in (1/m,1]\),
Proof
The triangle inequality gives
where \(c=c(m)\). By applying the Poincaré inequality in the spatial direction, we have
where \(c=c(n,N,m)\).
To complete the proof, we estimate the second term on the right-hand side in the estimate above. By Hölder’s inequality, we have
For \(\eta \in C_0^\infty (B_R(x_0))\) satisfying (2.8), it holds that
where the second term on the right-hand side can be estimated by Lemma 2.4. For the first term on the right-hand side we may apply (2.8) and obtain
Therefore, using the Poincaré inequality in the spatial direction and Hölder’s inequality, we have
This completes the proof. \(\quad \square \)
3 Parabolic Sobolev-Poincaré inequalities
This section provides a parabolic Sobolev-Poincaré inequality by adapting techniques in [20] to the double-phase case. Throughout this section, let \(z_0=(x_0,t_0)\in \Omega _T\), with \(x_0\in \Omega \) and \(t_0\in (0,T)\), be a Lebesgue point of \(|\nabla u(z)|^p+a(z)|\nabla u(z)|^q\) satisfying
for some \(\Lambda >1+\Vert a\Vert _{L^\infty (\Omega _T)}\). Recall that \(H(z,s):\Omega _T\times \mathbb {R}^{+ }\longrightarrow \mathbb {R}^+\), \(H(z,s)=s^p+a(z)s^q\). For a fixed point \(z_0\), we denote
Note that \(H_{z_0}(s)\) is strictly increasing and continuous with
By the intermediate value theorem for continuous functions, there exists \(\lambda =\lambda (z_0)>1\) such that
Let
The parameter \(K=K(n,\alpha ,[a]_{\alpha },M_1)>1\) will be determined later in (5.2).
The p-intrinsic cylinders and the (p, q)-intrinsic cylinders are considered separately in the argument. In the p-intrinsic case we assume that
where
is a p-intrinsic cylinder. In the (p, q)-intrinsic case we assume that
where
is a (p, q)-intrinsic cylinder.
3.1 The p-intrinsic case
In this case we consider estimates in p-intrinsic cylinders as in (3.5) and assume that (3.4) holds. We begin by estimating the last term in Lemma 2.5.
Lemma 3.1
Let u be a weak solution to (1.1). Then, for \(s\in [2\rho ,4\rho ]\) and \(\theta \in ((q-1)/p,1]\), there exists a constant \(c=c(n,p,q,\alpha ,L,[a]_{\alpha },M_1)\) such that
whenever \(Q_{4\rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (3.4).
Proof
It follows from (1.3) that \(q-1<p\). By (1.3) there exists a constant \(c=c([a]_{\alpha })\) such that
We apply the first condition in (3.4) to estimate the second term on the right-hand side of (3.8) and obtain
In order to estimate the last term on the right-hand side of (3.8), we recall that \(|Q_{s}^\lambda (z_0)|=c(n)s^{n+2}\lambda ^{2-p}\). Hölder’s inequality gives
where \(c=c(n)\), \(\gamma =\alpha p/(n+2)\) and \(\theta \in ((q-1)/p,1]\). We have \(\gamma \in (0,p-1)\), since
It follows from the second condition in (3.4), \(\lambda \geqq 1\) and the first condition in (1.3) that
where \(c=c(n,p,q,\alpha )\). Therefore, we obtain
where \(c=c(n,p,q,\alpha ,M_1)\). Similarly, replacing \(|\nabla u|\) by |F| in the above argument, we have
This completes the proof. \(\quad \square \)
Next we provide a p-intrinsic parabolic Poincaré inequality.
Lemma 3.2
Let u be a weak solution to (1.1). Then, for \(s\in [2\rho ,4\rho ]\) and \(\theta \in ((q-1)/p,1]\), there exists a constant \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\) such that
whenever \(Q_{4\rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (3.4).
Proof
By Lemmas 2.5 and 3.1, there exists a constant \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\) such that
To estimate the second term on the right-hand side of (3.9), we use the second condition in (3.4) and obtain
where \(c=c(n,p)\). Similarly, the third term on the right-hand side of (3.9) is estimated as
where \(c=c(n,p,q)\). The conclusion follows from Hölder’s inequality. \(\quad \square \)
Lemma 3.3
Let u be a weak solution to (1.1). Then for \(Q_{4\rho }^{\lambda }(z_0)\subset \Omega _T\) satisfying (3.4), \(s\in [2\rho ,4\rho ]\) and \(\theta \in ((q-1)/p,1]\), there exists a constant \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\) such that
Proof
By Lemmas 2.5 and 3.1, there exists a constant \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\) such that
By (3.4) for the second term on the right-hand side of (3.10), we obtain
Similarly, the third and the fourth terms on the right-hand side of (3.10) can be estimated as
and
The conclusion follows from Hölder’s inequality. \(\quad \square \)
3.2 The (p, q)-intrinsic case
In this case we consider estimates in (p, q)-intrinsic cylinders as in (3.7) and assume that (3.6) holds. The second and third conditions in (3.6) imply
It follows that
Next we discuss a (p, q)-intrinsic parabolic Poincaré inequality.
Lemma 3.4
Let u be a weak solution to (1.1). Then, for \(\theta \in ((q-1)/p,1]\) and \(s\in [2\rho ,4\rho ]\), there exists a constant \(c=c(n,N,p,q,L)\) such that
whenever \(G_{4\rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (3.6).
Proof
Note that
By Lemma 2.5, there exists a constant \(c=c(n,N,p,q,L)\) such that
Since
for every \(s>0\), we have
for every \(s>0\). We estimate the last term on the right-hand side of (3.12) by (3.13) and obtain
By the same argument as above for \(H_{z_0}'(|\nabla u|+|F|)\), it follows from (3.11) that
where \(c=c(n,p,q)\). Therefore, we obtain
In order to estimate the first term on the right-hand side of (3.14), we apply (3.11). Keeping in mind that \(q-1<p\) and \(p\geqq 2\), we have
for any \(\theta \in ((q-1)/p,1]\) with \(c=c(n,p)\). We estimate the last term on the right-hand side of (3.14) in a similar way. Then for any \(\theta \in ((q-1)/q,1]\), we have
where \(c=c(n,p)\). Hence, we conclude that
which completes the proof. \(\quad \square \)
Note that by replacing \(H_{z_0}^{\theta }(s)\) with \(s^{\theta p}\) in the proof of Lemma 3.4, we will also have the following result. All necessary calculations are already contained in the proof of the previous lemma.
Lemma 3.5
Let u be a weak solution to (1.1). Then, for \(\theta \in ((q-1)/p,1]\) and \(s\in [2\rho ,4\rho ]\), there exists a constant \(c=c(n,N,p,q,L)\) such that
whenever \(G_{4\rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (3.6).
4 Reverse Hölder inequalities
In this section, we assume that \(z_0\), \(\Lambda \) and \(\lambda =\lambda (z_0)\geqq 1\) satisfy (3.1)–(3.3). Let
The distribution sets are denoted as
and
We consider the p-intrinsic and (p, q)-intrinsic cases separately. In the first case we assume that
where the p-intrinsic cylinder is defined in (3.5). In the second case we assume that
where the (p, q)-intrinsic cylinder is defined in (3.7). We discuss reverse Hölder inequalities in both cases separately.
The following auxiliary lemmas will be employed in the argument the first lemma is a Gagliardo–Nirenberg inequality and the second one is a standard iteration lemma, see [15, Lemma 8.3]:
Lemma 4.1
Let \(B_{\rho }(x_0)\subset \mathbb {R}^n\), \(\sigma ,s,r\in [1,\infty )\) and \(\vartheta \in (0,1)\) such that
Then there exists a constant \(c=c(n,\sigma )\) such that
for every \(v\in W^{1,s}(B_{\rho }(x_0))\).
Lemma 4.2
Let \(0<r<R<\infty \) and \(h:[r,R]\longrightarrow \mathbb {R}\) be a non-negative and bounded function. Suppose there exist \(\vartheta \in (0,1)\), \(A,B\geqq 0\) and \(\gamma >0\) such that
Then there exists a constant \(c=c(\vartheta ,\gamma )\) such that
4.1 The p-intrinsic case
In this case we consider estimates in p-intrinsic cylinders as in (3.5) and assume that (4.3) holds. We denote
and \(M_2=\Vert u\Vert _{L^\infty (0,T;L^2(\Omega ))}\).
Lemma 4.3
Let u be a weak solution to (1.1). Then there exists a constant \(c=c( data )\) such that
whenever \(Q_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.3).
Proof
Let \(2\rho \leqq \rho _1<\rho _2\leqq 4\rho \). By Lemma 2.3 there exists a constant \(c=c(n,p,q,\nu ,L)\) such that
We estimate the first term on the right-hand side of (4.5). By Lemma 3.2 with \(\theta =1\) and the second condition in (4.3), we obtain
where \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\). On the other hand, we observe that
By Lemma 3.3 with \(\theta =1\) and (4.3), we obtain
where \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\). On the other hand, by Lemma 4.1 with \(\sigma =q\), \(s=p\), \(r=2\) and \(\vartheta =\tfrac{p}{q}\), we obtain
where \(c=c(n,q)\). We observe that
where \(c=c(n,N,p,q,\alpha ,{{\,\textrm{diam}\,}}(\Omega ),M_2)\). Furthermore, by (4.6) we have
where \(c=c(n,N,p,q,\alpha ,L,[a]_\alpha ,{{\,\textrm{diam}\,}}(\Omega ),M_1,M_2)\).
For the second term on the right-hand side of (4.5), the Poincaré inequality implies that
where \(c=c(n,N,p)\). By Hölder’s inequality and (4.6), we have
where \(c=c(n,N,p,q,\alpha ,L,[a]_{\alpha },M_1)\).
For the last term on the right-hand side of (4.5), by (4.3) we obtain
By combining all estimates above, we conclude from (4.5) that
Finally, we apply Young’s inequality to obtain
The proof is concluded by an application of Lemma 4.2. \(\quad \square \)
Next we prove an estimate for the first term on the right-hand side of the energy estimate in Lemma 2.3 by using Lemma 4.1.
Lemma 4.4
Let u be a weak solution to (1.1). Then there exist constants \(c=c( data )\) and \(\theta _0=\theta _0(n,p,q)\in (0,1)\) such that for any \(\theta \in (\theta _0,1)\),
whenever \(Q_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.3).
Proof
By (1.4) we obtain
We begin with the first term on the right-hand side of (4.7). By choosing \(\sigma =p\), \(s=\theta p\) and \(r=2\), we see that any \(\theta \in (n/(n+2),1)\) satisfies the condition in Lemma 4.1 as
Thus, we obtain
where \(c=c(n,p)\).
For the second term on the right-hand side of (4.7), we apply Lemma 4.1 with \(\sigma =q\), \(s=\theta q\) and \(r=2\). For any \(\theta \in (n/(n+2),1)\), we have
where \(c=c(n,q)\). By using the first condition in (4.3), we have
Then we consider the last term on the right-hand side of (4.7). We observe that
Thus, by letting \(\sigma =q\), \(s=\theta p\), \(r = 2\) and \(\vartheta =\theta p/q\), the assumptions in Lemma 4.1 are satisfied for any \(\theta \in (nq/((n+2)p),1)\), since
Therefore, we have
where \(c=c(n,q)\). Note that
Thus we obtain
where \(c=c(n,p,q,\alpha ,{{\,\textrm{diam}\,}}(\Omega ),M_2)\). The claim follows by combining the estimates above. \(\quad \square \)
At this stage, we have all the required tools to prove the reverse Hölder inequality when (4.3) holds true.
Lemma 4.5
Let u be a weak solution to (1.1). Then there exist constants \(c=c( data )\) and \(\theta _0=\theta _0(n,p,q)\in (0,1)\) such that for any \(\theta \in (\theta _0,1)\),
whenever \(Q_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.3).
Proof
Lemma 2.3 implies that
where \(c=c(n,p,q,\nu ,L)\). To estimate the first term on the right-hand side in (4.8), we apply Lemmas 4.3 and 4.4 to conclude that there exist \(\theta _0=\theta _0(n,p,q)\in (0,1)\) and \(c=c( data )\) such that for any \(\theta \in (\theta _0,1)\),
By Lemmas 3.2 and 3.3 we obtain
By recalling that \(\tfrac{\alpha p}{n+2} < p-1\) and letting
we have
To estimate the second term on the right-hand side of (4.8), we apply the Poincaré inequality with \(\theta \in (2n/((n+2)p),1)\) and Lemma 4.3 to obtain
where \(c=c( data )\). Lemma 3.2 implies that
By combining the estimates above and applying (4.8) and Young’s inequality, we obtain
The third condition in (4.3) implies that
This completes the proof. \(\quad \square \)
The following lemma will be used in the next section.
Lemma 4.6
Let u be a weak solution to (1.1). Then there exist constants \(c=c( data )\) and \(\theta _0=\theta _0(n,p,q)\in (0,1)\) such that for any \(\theta \in (\theta _0,1)\),
whenever \(Q_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.3). Here \(\Psi (\Lambda )\) and \(\Phi (\Lambda )\) are defined in (4.1) and (4.2).
Proof
The second condition in (4.3) implies that
By representing \(Q_{2\rho }^\lambda (z_0)\) as a union of \(Q_{2\rho }^\lambda (z_0)\cap \Psi ((4c)^{-1/\theta }\lambda ^p)\) and \(Q_{2\rho }^\lambda (z_0)\setminus \Psi ((4c)^{-1/\theta }\lambda ^p)\), we have
for any \(c > 0\). A similar argument gives
It follows from Lemma 4.5 that
By recalling the second and third conditions in (4.3), we obtain
Thus, we have
We note that
where we applied the first condition in (4.3). The estimate above implies that
Therefore, by replacing \(2K(4c)^{1/\theta _0}\) with c, (4.9) can be written as
This completes the proof. \(\quad \square \)
4.2 The (p, q)-intrinsic case
In this case we consider estimates in (p, q)-intrinsic cylinders as in (3.7) and assume that (4.4) holds. We remark that constants in the estimates depend only on \(n,N,p,q,\nu ,L\) since (1.1) reduces to a parabolic (p, q)-Laplace system in \(G_{2\kappa \rho }^\lambda (z_0)\). We denote
Lemma 4.7
Let u be a weak solution to (1.1). Then there exists a constant \(c=c(n,N,p,q,\nu ,L)\) such that
whenever \(G_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.4).
Proof
Let \(2\rho \leqq \rho _1<\rho _2\leqq 4\rho \). By Lemma 2.3, there exists a constant \(c=c(n,p,q,\nu ,L)\) such that
For the first term on the right-hand side of (4.10), we apply Lemma 3.4 together with the second and third conditions in (4.4) to obtain
where \(c=c(n,N,p,q,L)\).
For the second term on the right-hand side of (4.10), as in the proof of Lemma 4.3, we obtain
where \(c=c(n,N,p)\). By using Lemma 3.5 and (3.11), we obtain
where \(c=c(n,N,p,q,L)\). By combining estimates and arguing as in the proof of Lemma 4.3, we have
The conclusion follows by applying Lemma 4.2. \(\quad \square \)
Lemma 4.8
Let u be a weak solution to (1.1). Then there exists a constant \(c=c(n,p,q)\) such that for any \(\theta \in (n/(n+2),1)\),
whenever \(G_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.4).
Proof
From the second condition in (4.4), we obtain
By Lemma 4.1, there exists a constant \(c=c(n,p,q)\) such that for any \(\theta \in (n/(n+2),1)\), we have
and
Thus we conclude that
This completes the proof. \(\quad \square \)
Lemma 4.9
Let u be a weak solution to (1.1). Then there exist constants \(c=c(n,N,p,q,\nu ,L)\) and \(\theta _0=\theta _0(n,p,q)\in (0,1)\) such that for any \(\theta \in (\theta _0,1)\),
whenever \(G_{2\kappa \rho }^{\lambda }(z_0)\subset \Omega _T\) satisfies (4.4). Moreover, we have
where \(\Psi (\Lambda )\) and \(\Phi (\Lambda )\) are as in (4.1) and (4.2).
Proof
Once the first estimate in the statement holds, then the second estimate follows as in the proof of Lemma 4.6.
To prove the first estimate in the statement, we apply Lemma 2.3 to obtain
Using Lemmas 4.8, 3.4 and 4.7 for the first term on the right-hand side of (4.11), we obtain
As in the proof of Lemma 4.5, we obtain
and from Lemma 3.5 we conclude that
For the second term on the right-hand side of (4.11), we have
where
A similar argument for |F| gives
By collecting all the estimates above and applying Young’s inequality together with the second condition in (4.4), we obtain
We use the fourth condition in (4.4) to absorb the first term on the right-hand side. This completes the proof. \(\quad \square \)
5 The proof of Theorem 1.1
In this section, we will complete the proof of Theorem 1.1. We divide the section into three subsections. In the first subsection, we construct intrinsic cylinders which are either p-intrinsic or (p, q)-intrinsic, see (4.3) and (4.4). In the second subsection, we prove a Vitali type covering property for the system of intrinsic cylinders constructed in the first subsection. Note that the collection consists of two different types of intrinsic cylinders depending on the center point of the cylinders. Finally, in the last subsection, we complete the proof of gradient estimate by applying Fubini’s theorem together with Lemma 4.2.
5.1 Stopping time argument
Let
where \(Q_{2r}(z_0)=B_{2r}(x_0)\times (t_0-(2r)^2,t_0+(2r)^2)\). Moreover, let
With \(\Psi (\Lambda )\) and \(\Phi (\Lambda )\) as in (4.1)–(4.2) and \(\rho \in [r,2r]\), we denote
and
Next we apply a stopping time argument. Let \(r\leqq r_1<r_2\leqq 2r\) and
where \(\kappa \) is as in (5.2). For every \(w\in \Psi (\Lambda ,r_1)\), let \(\lambda _{w}>0\) be such that
where \(H_{w}\) is as in (3.2). We claim that
For a contradiction, assume that the inequality above does not hold. Then
which is a contradiction with (5.3). Therefore, for \(s\in [(r_2-r_1)/(2\kappa ),r_2-r_1)\), we have
Since \(w\in \Psi (\Lambda ,r_1)\) and (5.4) holds, it follows that \(w\in \Psi (\lambda _{w}^p,r_1)\). By the Lebesgue differentiation theorem there exists \(\rho _{w}\in (0,(r_2-r_1)/(2\kappa ))\) such that
and
for every \(s\in (\rho _{w},r_2-r_1)\). Observe that (5.6) and (5.5), imply that
For \(K>1\) as in (5.2), either
In addition, either
We consider three cases:
-
(1)
\(K\lambda _{w}^p\geqq a(w)\lambda _{w}^q\), that is, the first condition in (5.9) holds,
-
(2)
\(K\lambda _{w}^p\leqq a(w)\lambda _{w}^q\) and \(a(w)\geqq 2[a]_{\alpha }(10\rho _{w})^\alpha \), that is, the second condition in (5.9) and the first condition in (5.10) and
-
(3)
\(K\lambda _{w}^p\leqq a(w)\lambda _{w}^q\) and \(a(w)\leqq 2[a]_{\alpha }(10\rho _{w})^\alpha \), that is, the second condition in (5.9) and the second condition in (5.10) hold.
First we note that (1), together with (5.6)–(5.7), imply (4.3) for p-intrinsic cylinders by replacing the center point and radius with \(w\) and \(\rho _{w}\). Next we show that (2) implies (4.4) for (p, q)-intrinsic cylinders. From the second condition in (5.9) we obtain \(a(w)>0\) and \(G_{s}^{\lambda _{w}}(w)\subsetneq Q_s^{\lambda _{w}}(w)\). By (5.6)–(5.7), we obtain
for every \(s\in (\rho _{w},r_2-r_1)\). Recall that \(w\in \Psi (\Lambda ,r_1)\) and \(\Lambda =H_{w}(\lambda _{w})\). We find \(\varsigma _{w}\in (0,\rho _{w}]\) such that
and
for every \(s\in (\varsigma _{w},r_2-r_1)\). Moreover, it follows from the first condition in (5.10) that
and
Therefore,
Hence, (2) implies (5.11)–(5.12). This shows that (4.4) is satisfied by replacing the center point and radius with \(w\) and \(\varsigma _{w}\).
Finally, we prove that (3) never occurs due to (5.2). From the second condition in (5.9) and the second condition in (5.10), we have
By applying (5.6) and recalling that \(\gamma =\tfrac{\alpha p}{n+2}\), we obtain
Observe that
and
It follows from (5.2) that
Therefore, the second condition in (5.9) and the second condition in (5.10) cannot occur together.
5.2 Vitali type covering argument
In Section 4 we considered reverse Hölder inequality, and in Section 5.1 we discussed a stopping time argument, for p-intrinsic and (p, q)-intrinsic cylinders. For each \(w\in \Psi (\Lambda ,r_1)\), we consider
We prove a Vitali type covering lemma for this collection of intrinsic cylinders. We denote
Recall that \(l_{w}\in (0,R)\) for every \(w\in \Psi (\Lambda ,r_1)\), where \(R=(r_2-r_1)/\kappa \) and \(\kappa \) is as in (5.2). Let
We construct subcollections \(\mathcal {G}_j\subset \mathcal {F}_j\), \(j\in {\mathbb {N}}\), recursively as follows. Let \(\mathcal {G}_1\) be a maximal disjoint collection of cylinders in \(\mathcal {F}_1\). By (5.8) we observe that the measure of each cylinder in \(\mathcal {G}_1\) is bounded from below, which implies that the collection is finite. Suppose that we have selected \(\mathcal {G}_1,...,\mathcal {G}_{k-1}\) with \(k\geqq 2\), and let
be a maximal collection of pairwise disjoint cylinders. It follows that
is a countable subcollection of pairwise disjoint cylinders in \(\mathcal {F}\). We claim that for each \(\mathcal {Q}(w)\in \mathcal {F}\), there exists \(\mathcal {Q}(v)\in \mathcal {G}\) such that
where
For every \(\mathcal {Q}(w)\in \mathcal {F}\), there exists \(j \in \mathbb {N}\) such that \(\mathcal {Q}(w)\in \mathcal {F}_j\). By the construction of \(\mathcal {G}_j\), there exists a cylinder \(\mathcal {Q}(v)\in \cup _{i=1}^j \mathcal {G}_i\) for which the first condition in (5.14) holds true. Moreover, since \(l_{w}\leqq \tfrac{R}{2^{j-1}}\) and \(l_{v} \geqq \tfrac{R}{2^j}\), we have
In the remaining of this subsection, we will prove the second claim in (5.14). We note that if \(\lambda =\lambda _{w}=\lambda _{v}\) and either
or
then the second claim in (5.14) holds true if \(\kappa \geqq 5\). Indeed, once the scaling factor in the time interval of two intrinsic cylinders is the same, these cylinders are in the standard parabolic metric space. Thus the standard proof of Vitali’s covering lemma can be applied in these cases.
Regardless of (1) and (2), for \(i\in \{v,w\}\), there exist \(2\rho _i\geqq l_i>0\) and \(\lambda _i>0\) such that
and
We show that the second claim in (5.14) holds in all four possible cases that may occur:
-
(i)
\(\mathcal {Q}(v)=Q_{l_v}^{\lambda _v}(v)\) and \(\mathcal {Q}(w)=Q_{l_w}^{\lambda _w}(w)\),
-
(ii)
\(\mathcal {Q}(v)=G_{l_v}^{\lambda _v}(v)\) and \( \mathcal {Q}(w)=G_{l_w}^{\lambda _w}(w)\),
-
(iii)
\(\mathcal {Q}(v)=G_{l_v}^{\lambda _v}(v)\) and \( \mathcal {Q}(w)=Q^{\lambda _w}_{l_w}(w)\) and
-
(iv)
\(\mathcal {Q}(v)=Q_{l_v}^{\lambda _v}(v)\) and \(\mathcal {Q}(w)=G_{l_w}^{\lambda _w}(w)\).
Observe that in any of these cases, the first condition in (5.14) implies that \(Q_{l_{w}}(w) \cap Q_{l_{v}}(v) \ne \emptyset \) and
This will already imply that the second claim in (5.14) holds for the spatial part of the set by enlarging the radius by factor 5. In the rest of this subsection, we show the inclusion of the time intervals when enlarging the radius with factor \(\kappa \) by considering each case separately.
First we collect a few facts that will be applied in the argument. By (5.18) we have
On the other hand, from (5.17), we may deduce that
If \(\lambda _w \leqq \lambda _v\), we claim that
For a contraction, assume that (5.21) does not hold. It follows from (5.16) and (5.19) that
By (5.20) we obtain
since \(\lambda _w \leqq \lambda _v\) and \(q \leqq p + 2\alpha /(n+2)\). Substituting the negation of (5.21) and the above display into the right-hand side of (5.22) leads to a contradiction since
On the other hand, if \(\lambda _v\leqq \lambda _w\), we claim that
Otherwise, it follows from (5.20) that
In any case we have
Let \(v=(x_v,t_v)\) and \(w=(x_w,t_w)\) for \(x_v,x_w\in \mathbb {R}^n\) and \(t_v,t_w\in \mathbb {R}\).
(i): \(\mathcal {Q}(v)=Q_{l_v}^{\lambda _v}(v)\) and \(\mathcal {Q}(w)=Q_{l_w}^{\lambda _w}(w)\). For any \(\tau \in I_{l_w}^{\lambda _w}(t_w)\), we apply (5.15), (5.23) and \((p-2)/p\leqq 1\) to have
which implies \(I_{l_w}^{\lambda _w}(t_w) \subset \kappa I_{ l_v}^{\lambda _v}(t_v)\). Thus, we have \(Q_{l_w}^{\lambda _w}(w) \subset \kappa Q_{ l_v}^{\lambda _v}(v)\).
(ii): \(\mathcal {Q}(v)=G_{l_v}^{\lambda _v}(v)\) and \( \mathcal {Q}(w)=G_{l_w}^{\lambda _w}(w)\). For any \(\tau \in J_{l_w}^{\lambda _w}(t_w)\), we have
By (5.16), (5.23) and \(2/p\leqq 1\) we have
Therefore applying (5.15), we obtain
This implies that \(J_{l_w}^{\lambda _w}(t_w) \subset \kappa J_{ l_v}^{\lambda _v}(t_v)\). Thus, we have \(G_{l_w}^{\lambda _w}(w) \subset \kappa G_{ l_v}^{\lambda _v}(v)\).
(iii): \(\mathcal {Q}(v)=G_{l_v}^{\lambda _v}(v)\) and \( \mathcal {Q}(w)=Q^{\lambda _w}_{l_w}(w)\). For any \(\tau \in I_{l_w}^{\lambda _w}(t_w)\), we have from (5.16) that
Recalling \(K\lambda _w^p\geqq a(w)\lambda _w^q\), we apply (5.23), \(2/p\leqq 1\) and (5.16) to get
which, together with (5.24) and (5.15), implies
Therefore \(I_{l_w}^{\lambda _w}(t_w) \subset \kappa J_{ l_v}^{\lambda _v}(t_v)\) and \(Q_{l_w}^{\lambda _w}(w)\subset \kappa G_{l_v}^{\lambda _v}(v)\).
(iv): \(\mathcal {Q}(v)=Q_{l_v}^{\lambda _v}(v)\) and \(\mathcal {Q}(w)=G_{l_w}^{\lambda _w}(w)\). For any \(\tau \in J_{l_w}^{\lambda _w}(t_w)\), we apply (5.15), (5.23) and \((p-2)/p\leqq 1\) to have
Therefore \(J^{\lambda _w}_{l_w}(t_w)\subset \kappa I_{l_v}^{\lambda _v}(t_v)\) and \(G_{l_w}^{\lambda _w}(w) \subset \kappa Q_{l_v}^{\lambda _v}(v)\). Since we have covered every case, the proof of the second condition in (5.14) is completed.
5.3 Final proof of the gradient estimate
We write the countable pairwise disjoint collection \(\mathcal {G}\) defined in (5.13) as \(\mathcal {G}=\cup _{j=1}^\infty \mathcal {Q}_j\), where \(\mathcal {Q}_j=\mathcal {Q}(w_j)\) with \(w_j \in \Psi (\Lambda ,r_1)\).
Lemmas 4.6 and 4.9 imply that there exist \(c=c( data )\) and \(\theta _0=\theta _0(n,p,q)\in (0,1)\) such that
for every \(j\in \mathbb {N}\) with \(\theta = (\theta _0+1)/2\). By summing over j and applying the fact that the cylinders in \(\mathcal {G}\) are pairwise disjoint, we obtain
Moreover, since
we conclude from (5.25) that
For \(k\in {\mathbb {N}}\), let
and
It is easy to see that if \(\Lambda >k\), then \(\Psi _k(\Lambda ,\rho )=\emptyset \) and if \(\Lambda \leqq k\), then \(\Psi _k(\Lambda ,\rho )=\Psi (\Lambda ,\rho )\). Therefore, we deduce from (5.26) that
Recalling (5.3), we denote
Then for any \(\Lambda >\Lambda _1\), we obtain
Let \(\varepsilon \in (0,1)\) to be chosen later. We multiply the inequality above by \(\Lambda ^{\varepsilon -1}\) and integrate each term over \((\Lambda _1,\infty )\), which implies
We apply Fubini’s theorem to estimate \(\textrm{I}\) and obtain
Since
we have
Similarly, by Fubini’s theorem, we have
and
By combining the estimates above we obtain
We choose \(\varepsilon _0=\varepsilon _0( data )\in (0,1)\) so that for any \(\varepsilon \in (0,\varepsilon _0)\),
Then, by applying Lemma 4.2 we get
The claim follows by letting \(k\longrightarrow \infty \) and recalling (5.1).
Data Availability Statement
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
References
Arora, R., Shmarev, S.: Double-phase parabolic equations with variable growth and nonlinear sources. Adv. Nonlinear Anal. 12(1), 304–335, 2023
Baroni, P., Colombo, M., Mingione, G.: Harnack inequalities for double phase functionals. Nonlinear Anal. 121, 206–222, 2015
Bögelein, V., Duzaar, F.: Higher integrability for parabolic systems with non-standard growth and degenerate diffusions. Publ. Mat. 55(1), 201–250, 2011
Bögelein, V., Duzaar, F., Marcellini, P.: Parabolic equations with \(p, q\)-growth. J. Math. Pures Appl. (9) 100(4), 535–563, 2013
Bögelein, V., Duzaar, F., Marcellini, P.: Parabolic systems with \(p, q\)-growth: a variational approach. Arch. Ration. Mech. Anal. 210(1), 219–267, 2013
Bulíček, M., Gwiazda, P., Skrzeczkowski, J.: On a range of exponents for absence of Lavrentiev phenomenon for double phase functionals. Arch. Ration. Mech. Anal. 246(1), 209–240, 2022
Chlebicka, I., Gwiazda, P., Zatorska-Goldstein, A.: Parabolic equation in time and space dependent anisotropic Musielak–Orlicz spaces in absence of Lavrentiev’s phenomenon. Ann. Inst. H. Poincaré C Anal. Non Linéaire 36(5), 1431–1465, 2019
Colombo, M., Mingione, G.: Regularity for double phase variational problems. Arch. Ration. Mech. Anal. 215(2), 443–496, 2015
Colombo, M., Mingione, G.: Calderón–Zygmund estimates and non-uniformly elliptic operators. J. Funct. Anal. 270(4), 1416–1478, 2016
De Filippis, C.: Gradient bounds for solutions to irregular parabolic equations with \((p, q)\)-growth. Calc. Var. Partial Differ. Equ. 59(5), 171, 2020
De Filippis, C., Mingione, G.: A borderline case of Calderón–Zygmund estimates for nonuniformly elliptic problems. Algebra Anal. 31(3), 82–115, 2019
DiBenedetto, E.: Degenerate Parabolic Equations. Universitext. Springer, New York, 1993
Esposito, L., Leonetti, F., Mingione, G.: Sharp regularity for functionals with \((p, q)\) growth. J. Differ. Equ. 204(1), 5–55, 2004
Giannetti, F., Passarelli di Napoli, A., Scheven, C.: On higher differentiability of solutions of parabolic systems with discontinuous coefficients and \((p, q)\)-growth. Proc. R. Soc. Edinb. Sect. A 150(1), 419–451, 2020
Giusti, E.: Direct Methods in the Calculus of Variations. World Scientific Publishing Co., Inc., River Edge, 2003
Hästö, P., Ok, J.: Higher integrability for parabolic systems with Orlicz growth. J. Differ. Equ. 300, 925–948, 2021
Hästö, P., Ok, J.: Maximal regularity for local minimizers of non-autonomous functionals. J. Eur. Math. Soc. 24(4), 1285–1334, 2022
Hästö, P., Ok, J.: Regularity theory for non-autonomous partial differential equations without Uhlenbeck structure. Arch. Ration. Mech. Anal. 245(3), 1401–1436, 2022
Kim, W., Kinnunen, J., Särkiö, L.: Lipschitz truncation method for the parabolic double-phase system and applications. arXiv, 2023
Kinnunen, J., Lewis, J.L.: Higher integrability for parabolic systems of \(p\)-Laplacian type. Duke Math. J. 102(2), 253–271, 2000
Marcellini, P.: Regularity of minimizers of integrals of the calculus of variations with nonstandard growth conditions. Arch. Ration. Mech. Anal. 105(3), 267–284, 1989
Marcellini, P.: Regularity and existence of solutions of elliptic equations with \(p, q\)-growth conditions. J. Differ. Equ. 90(1), 1–30, 1991
Mingione, G., Rǎdulescu, V.: Recent developments in problems with nonstandard growth and nonuniform ellipticity. J. Math. Anal. Appl. 501(1), 125197, 2021
Ržička, M.: Electrorheological fluids: modeling and mathematical theory. Lecture Notes in Mathematics, Vol. 1748. Springer, Berlin, 2000
Singer, T.: Existence of weak solutions of parabolic systems with \(p, q\)-growth. Manuscr. Math. 151(1–2), 87–112, 2016
Zhikov, V.V.: Averaging of functionals of the calculus of variations and elasticity theory. Izv. Akad. Nauk SSSR Ser. Mat. 50(4), 675–710, 1986
Zhikov, V.V.: Lavrentiev phenomenon and homogenization for some variational problems. C. R. Acad. Sci. Paris Sér. I Math. 316(5), 435–439, 1993
Zhikov, V.V., Kozlov, S.M., Oleĭnik, O.A.: Homogenization of Differential Operators and Integral Functionals. Springer, Berlin, 1994. Translated from the Russian by G. A. Yosifian [G. A. Iosif’ yan]
Acknowledgements
This research was started during a visit of the first author at the Department of Mathematics of Aalto University. He would like to thank the Nonlinear Partial Differential Equations group for the kind and warm hospitality.
Funding
Open Access funding provided by Aalto University. No funding was received for conducting this study.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors have no competing interests to declare that are relevant to the content of this article.
Additional information
Communicated by G. Dal Maso.
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
Kim, W., Kinnunen, J. & Moring, K. Gradient Higher Integrability for Degenerate Parabolic Double-Phase Systems. Arch Rational Mech Anal 247, 79 (2023). https://doi.org/10.1007/s00205-023-01918-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00205-023-01918-0