Global higher integrability for systems with p p -growth structure in noncylindrical domains
Kristian Moring, Leah Schätzler, Christoph Scheven · 2025
Read Global higher integrability for systems with p p -growth structure in noncylindrical domains by Kristian Moring, Leah Schätzler, Christoph Scheven (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-02553.txt.
Introduction
Kristian Moring, Christoph Scheven, and Leah Schätzler prove a regularity result for a class of parabolic equations whose domain moves in time. The systems have p-growth structure, modelled on the parabolic p-Laplace equation, and are posed with Cauchy-Dirichlet data in a bounded noncylindrical domain. Such a region is one in space-time whose spatial slice can change shape as time advances. The question is whether the gradient of a solution is more integrable than the equation alone guarantees.
The result needs growth exponents above an explicit dimension-dependent threshold. The domains must satisfy regularity conditions and neither grow nor shrink too fast. Then the authors prove global higher integrability of the gradient, up to the boundary and the initial time. The conclusion is new even in the linear-flavoured case where the exponent equals two, the borderline that classical theory left open in this generality.
Catalog record
| CREATOR | Kristian Moring, Leah Schätzler, Christoph Scheven |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 85 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-02553
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Global higher integrability for systems with p p -growth structure in noncylindrical domains Kristian Moring, Leah Schätzler, Christoph Scheven Abstract . We consider the Cauchy–Dirichlet problem to systems with p p -growth structure with 1 < p < ∞ 1<p<\infty , whose prototype is ∂ t u − div ( | D u | p − 2 D u ) = div ( | F | p − 2 F ) , \partial_{t}u-\Div\big(|Du|^{p-2}Du\big)=\Div\left(|F|^{p-2}F\right), in a bounded noncylindrical domain E ⊂ ℝ n + 1 E\subset\mathds{R}^{n+1} . For p > 2 ( n + 1 ) n + 2 p>\frac{2(n+1)}{n+2} and domains E E that satisfy suitable regularity assumptions and do not grow or shrink too fast, we prove global higher integrability of D u Du . The result is already new in the case p = 2 p=2 . 2020 Mathematics Subject Classification 35B65, 35K51, 35K55, 35R37 1. Introduction For n\in\mathbb{N} and T>0 , we suppose that E\subset\mathbb{R}^{n}\times[0,T) is a bounded relatively open domain, and we consider the Cauchy–Dirichlet problem to partial differential equations (PDEs) of parabolic p -Laplace type \partial_{t}u-\Div\big(|Du|^{p-2}Du\big)=\Div\left(|F|^{p-2}F\right)\qquad\text{in }E, where 1<p<\infty , u\colon E\to\mathbb{R}^{N} for some N\geq 1 , and F\in L^{p}(E,\mathbb{R}^{Nn}) . In particular, in the case N>1 , we are dealing with systems of PDEs. More generally, denoting the initial and boundary values by g_{o} and g , respectively, we are concerned with systems with p -growth structure of the form \displaystyle\left\{\begin{array}[]{rl}\partial_{t}u-\Div\mathbf{A}(x,t,u,Du)=\Div\left(|F|^{p-2}F\right)&\text{ in }E,\\ u=g&\text{ on }\bigcup_{t\in[0,T)}\partial E^{t}\times\{t\},\\ u=g_{o}&\text{ on }E^{0}\times\{0\}.\end{array}\right. For the precise assumptions on E , \mathbf{A} , g_{o} , and g , we refer to Section 2.1 . Moreover, the time slice of E at a fixed time t\in[0,T) is given by E^{t}=\{x\in\mathbb{R}^{n}:(x,t)\in E\}, such that E=\bigcup_{t\in[0,T)}E^{t}\times\{t\}. Viewing E^{t} as a spatial set that changes over time instead of focusing on E as a noncylindrical domain in space-time, the terms time-varying or time-changing domains have also been used in the literature. For the existence theory for PDEs of parabolic p -Laplace type, and some generalizations in such domains, we refer to [ 3 , 5 , 17 , 23 , 27 , 26 ] . The motivation for considering PDEs in noncylindrical domains is twofold. On the one hand, they appear naturally in some physical and biological applications, e.g. the flow of fluids through a container whose walls can be moved, or pattern formation in a growing organism. For an overview of applications of different PDEs involving a noncylindrical setting, we refer to [ 6 , 15 , 16 , 21 , 28 , 29 ] . On the other hand, from a purely mathematical point of view, the noncylindrical formulation truly takes the evolutionary nature of the parabolic system ( 1 ) into account, whereas the cylindrical case E=\Omega\times[0,T) constitutes a simpler special case. In the present paper, we establish the global higher integrability of the gradient of weak solutions u to ( 1 ). This means that while a priori we only know that |Du| is integrable to power p , under suitable conditions on E , g_{o} , g , and F we prove that |Du| is actually integrable to a higher exponent than p . To the best of our knowledge, our result is already new for the case p=2 in the noncylindrical situation. Before we state the precise result, we briefly discuss the history of the problem. For elliptic PDEs of p -Laplace type in a domain \Omega\subset\mathbb{R}^{n} , local higher integrability is due to Meyers & Elcrat [ 20 ] , while global higher integrability was established by Kilpeläinen & Koskela [ 11 ] . In particular, the authors of the latter article assume that the complement of \Omega satisfies a uniform p -fatness condition (see Definition 2.1 below), and show that this condition is essentially optimal. A key component in their proof is the self-improving property of the uniform p -fatness condition that was first established by Lewis [ 18 ] in the more general context of Riesz potentials (see Lemma 2.7 below). For PDEs of parabolic p -Laplace type with quadratic growth p=2 , Giaquinta & Struwe [ 7 ] prove local higher integrability. However, their technique does not carry over to the case of more general p . The breakthrough was achieved almost 20 years later by Kinnunen & Lewis [ 13 , 14 ] , who were able to deal with the whole range \max\big\{1,\frac{2n}{n+2}\big\}<p<\infty . Note that the critical exponent \frac{2n}{n+2} appears naturally in different areas of regularity theory for the parabolic p -Laplace equation. Their key idea was to use a suitable intrinsic geometry to balance the different scalings in the evolution and diffusion terms, i.e., to consider cylinders of the form Q_{\varrho}^{(\lambda)}(x_{o},t_{o}):=B_{\varrho}(x_{o})\times(t_{o}-\lambda^{2-p}\varrho^{2},t_{o}+\lambda^{2-p}\varrho^{2}) , whose length depends on the integral average of |Du|^{p} over the same cylinder via \lambda^{p}\approx\mathchoice{\vbox{\hbox{$\textstyle-\!-$}}\kern-8.47218pt}{\vbox{\hbox{$\scriptstyle-\!-$}}\kern-5.86113pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-4.56949pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-3.81949pt}\!\iint_{Q_{\varrho}^{(\lambda)}(x_{o},t_{o})}|Du|^{p}\,\mathrm{d}x\mathrm{d}t. The main tool in the proof of higher integrability is a reverse Hölder inequality for solutions in such intrinsic cylinders, meaning that the integral average of |Du|^{p} over an intrinsic cylinder can be estimated from above in terms of the integral average of |Du|^{q} with a smaller exponent q<p . Subsequently, Parviainen [ 24 , 25 ] established global higher integrability in the cylindrical setting \Omega\times[0,T) under the condition that \mathbb{R}^{n}\setminus\Omega is uniformly p -fat. Since this condition is analogous to the elliptic case, it is optimal. In the noncylindrical setting, we are able to deal with the same level of generality concerning the spatial regularity of E . In addition, we give sufficient assumptions on the speed at which E may grow or shrink in time. It is an interesting question for future research to determine the optimal condition on E with respect to the time variable that allows for global higher integrability. To state our main result, we need the following notation. For z_{o}=(x_{o},t_{o})\in\mathbb{R}^{n+1} and R>0 , we denote standard parabolic cylinders by Q_{R}:=Q_{R}(z_{o})=B_{R}(x_{o})\times\big(t_{o}-R^{2},t_{o}+R^{2}\big). Further, throughout the paper, let (1.4) G:=|Dg|^{p}+|\partial_{t}g|^{\hat{p}^{\prime}}+|F|^{p}, where \hat{p}^{\prime} is defined in ( 2.8 ). Moreover, letting \beta given by ( 2.5 ) be the exponent in the two-sided condition ( 2.4 ), the scaling deficit is given by (1.5) d=\left\{\begin{array}[]{ll}\frac{p}{2}&\text{for }p\geq 2,\\[5.0pt] \frac{p(2\beta-1)}{\beta(p(n+2)-2n)-2}&\text{for }\frac{2(n+1)}{n+2}<p<2.\end{array}\right. Then, our main theorem is the following. Theorem 1.1 . Let \frac{2(n+1)}{n+2}<p<\infty , suppose that \mathbb{R}^{n}\setminus E^{t} is uniformly p -fat in the sense of Definition 2.1 with a fatness constant \alpha>0 for every t\in(0,T) , and assume that E satisfies the conditions ( 2.1 ) with \ell given by ( 2.1 ), and ( 2.4 ) with some \beta in the range given by ( 2.5 ). Furthermore, let the operator \mathbf{A} be given according to ( 2.1 ), and assume that for some \sigma>0 , the lateral and initial boundary values g\in\mathfrak{G}^{\sigma} and g_{o}\in\mathfrak{G}^{\sigma}_{o} ( see ( 2.10 ) and ( 2.11 ) for the definition of these spaces ) satisfy ( 2.12 ), and the source term F\in L^{p+\sigma}(E,\mathbb{R}^{Nn}) . Finally, let u be a weak solution to ( 1 ) according to Definition 2.3 . Then, there exists \varepsilon_{o}=\varepsilon_{o}(n,p,C_{o},C_{1},M,\alpha,\beta)>0 such that Du\in L^{p+\varepsilon_{1}}(E,\mathbb{R}^{Nn}), where \varepsilon_{1}=\min\{\varepsilon_{o},\sigma\} . Furthermore, for every \varepsilon\in(0,\varepsilon_{o}] and any cylinder Q_{2R}(z_{o})\subset\mathbb{R}^{n+1} with z_{o}\in\overline{E} , the bound \displaystyle\mathchoice{\vbox{\hbox{$\textstyle-\!-$}}\kern-8.47218pt}{\vbox{\hbox{$\scriptstyle-\!-$}}\kern-5.86113pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-4.56949pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-3.81949pt}\!\iint_{Q_{R}}|Du|^{p+\varepsilon}\chi_{E}\,\>\!\mathrm{d}x\>\!\mathrm{d}t \displaystyle\leq c\left(1+\mathchoice{\vbox{\hbox{$\textstyle-\!-$}}\kern-8.47218pt}{\vbox{\hbox{$\scriptstyle-\!-$}}\kern-5.86113pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-4.56949pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-3.81949pt}\!\iint_{Q_{4R}}(|Du|^{p}+G)\chi_{E}\,\>\!\mathrm{d}x\>\!\mathrm{d}t\right)^{1+\frac{\varepsilon d}{p}} \displaystyle\quad+c\mathchoice{\vbox{\hbox{$\textstyle-\!-$}}\kern-8.47218pt}{\vbox{\hbox{$\scriptstyle-\!-$}}\kern-5.86113pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-4.56949pt}{\vbox{\hbox{$\scriptscriptstyle-\!-$}}\kern-3.81949pt}\!\iint_{Q_{2R}}G^{1+\frac{\varepsilon}{p}}\chi_{E}\,\>\!\mathrm{d}x\>\!\mathrm{d}t \displaystyle\quad+c\left(\mathchoice{\vbox{\hbox{$\textstyle-$}}\kern-4.86108pt}{\vbox{\hbox{$\scriptstyle-$}}\kern-3.43057pt}{\vbox{\hbox{$\scriptscriptstyle-$}}\kern-2.908pt}{\vbox{\hbox{$\scriptscriptstyle-$}}\kern-2.76045pt}\!\int_{B_{2R}}|Dg_{o}|^{{p}_{*}+\varepsilon}\chi_{E^{0}}\,\>\!\mathrm{d}x\right)^{\frac{\hat{p}+\varepsilon}{\hat{p}_{*}+\varepsilon}} holds true for