Low regularity results for degenerate Poisson problems

Marta Calanchi, Massimo Grossi · 2025

Read Low regularity results for degenerate Poisson problems by Marta Calanchi, Massimo Grossi (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-2503-08649.txt.

Introduction

Marta Calanchi and Massimo Grossi study a degenerate Poisson problem in a 2025 mathematics paper. The equation models diffusion where the conductivity fades near the boundary of a smooth bounded domain, at a rate set by distance to the edge and an exponent below one. They describe how the solution behaves near the boundary and discuss its regularity properties, that is, how smooth the solution can be expected to be under a continuous source term.

Catalog record

CREATORMarta Calanchi, Massimo Grossi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-08649

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Low regularity results for degenerate Poisson problems Marta Calanchi, Massimo Grossi Abstract . In this paper we study the Poisson problem, { − div ( d β ∇ u ) = f in ​ Ω u = 0 on ​ ∂ Ω , \begin{cases}-{\rm div}(d^{\beta}\nabla u)=f&{\rm in}\ \Omega\\ u=0&{\rm on}\ \partial\Omega,\end{cases} where Ω ⊂ ℝ N \Omega\subset\mathbb{R}^{N} , N ≥ 2 N\geq 2 is a smooth bounded domain, f f is a continuous function, β < 1 \beta<1 , and d ⁡ ( x ) = d ​ i ​ s ​ t ​ ( x , ∂ Ω ) d(x)=dist(x,\partial\Omega) . We describe the behaviour of u u near ∂ Ω \partial\Omega and discuss some of its regularity properties. 1. Introduction The study of partial differential equations plays a fundamental role in mathematical analysis, with applications ranging from physics to engineering, and has profound implications in understanding the regularity of solutions in various settings. In this work, we focus on a particular class of Poisson-type problems that involve a degenerate/singular elliptic operator. In many cases, these problems exhibit degeneracy at the boundary, where the behaviour of the solution becomes less regular or singular. We consider the weighted operator, denoted by L_{\beta} , defined as (1.1) L_{\beta}(u):=-{\rm div}(d^{\beta}\nabla u). where the weights w(x)=d^{\beta}(x) with \beta<1 are powers of the distance function d(x)=dist(x,\partial\Omega), and \Omega\subset\mathbb{R}^{N} is a bounded domain with \partial\Omega\in C^{2} . We are interested principally in the regularity but mainly in the precise behaviour at the boundary of classical solutions of the following Poisson-type problem (1.2) \begin{cases}L_{\beta}(u)=f&{\rm in}\ \Omega\\ u=0&{\rm on}\ \partial\Omega,\end{cases} where f is a given function that acts as a source term within the domain \Omega . Although the operator is degenerate, if f\in C^{0,\alpha}(\Omega) for some \alpha\in(0,1) , the classical local Schauder estimates give at least that u\in C^{2}(\Omega) , since L_{\beta} is uniformly elliptic in every compact set \Omega^{\prime}\subset\Omega . The weight introduces a degeneracy into the operator, which affects the behaviour of the solutions, that in general become less regular. We aim to refine the understanding of the regularity of solutions to ( 1.2 ), particularly near the boundary of \Omega . Since the operator is in divergence form, it is natural to also consider the problem from a variational point of view, addressing both the exitence and the possible regularity of the weak solutions of ( 1.2 ). To this aim it is necessary to introduce the natural (weighted) Sobolev space in which to frame the problem. More generally, for w\in L_{loc}^{1}(\Omega) , w(x)\geq 0\ a.e. , we denote with L^{p}(\Omega,w) the weighted Lebesgue space of measurable functions u on \Omega with finite norm (1.3) ||u||_{L^{p}(\Omega,w)}=\left(\int_{\Omega}|u(x)|^{p}w(x)\ dx\right)^{\frac{1}{p}} and with W^{k,p}(\Omega,w) the weighted Sobolev space of all measurable functions u whose distributional derivatives belongs to L^{p}(\Omega,w) , i.e. for which the norm (1.4) ||u||_{W^{k,p}(\Omega,w)}=\left(\sum_{|\alpha|\leq k}||D^{\alpha}u||^{p}_{L^{p}(\Omega,w)}\right)^{\frac{1}{p}} is finite. It is well-established (see e.g [ 7 ] ) that if w^{-\frac{1}{p}}\in L^{p^{\prime}}_{loc}(\Omega) , then W^{k,p}(\Omega,w) are Banach spaces, and C_{0}^{\infty}(\Omega)\subset W^{k,p}(\Omega,w) . Therefore, one can also define the space W_{0}^{k,p}(\Omega,w)=\overline{C_{0}^{\infty}(\Omega)}^{||\cdot||_{W^{k,p}(\Omega,w)}} as the closure of C^{\infty}_{0}(\Omega) with respect to the norm defined in ( 1.4 ). An interesting reference for weighted Sobolev spaces and related inequalities is the work by Edmunds & Opic, ( [ 4 ] , see also the work of Kufner and Opic [ 7 ] ), which focuses on the weighted Poincaré inequality treating in details the special case of weights w which are powers of the distance function d(x). It will be exactly this work that we will refer to in order to discuss some preliminary results. We say that u\in W^{1,2}_{0}(\Omega,d^{\beta}) is a weak solution to ( 1.2 ), if (1.5) \int_{\Omega}\nabla u\nabla\phi\ d^{\beta}dx=\int_{\Omega}f\phi\ dx,\quad\forall\phi\in C^{\infty}_{0}(\Omega). It is standard to prove that, if f/d^{\beta}\in L^{2}(\Omega,d^{\beta}) , then the problem ( 1.2 ) admits a weak solution u\in W^{1,2}_{0}(\Omega,d^{\beta}) (see Section 2 for more details). The present work is partially inspired by the celebrated paper of Fabes, Koenig and Serapioni ( [ 5 ] ), which investigates the local regularity properties of solutions to degenerate elliptic partial differential equations. Specifically, it focuses on the behaviour of solutions in regions where the ellipticity condition of the equation may fail or degenerate. The authors develop and apply methods to demonstrate that, under certain conditions, solutions to these equations possess local regularity. Their results involve more general weights that belong to the Muckenhoupt classes A_{p} (see e.g. [ 8 ] ), i.e. weights w defined as follows: for \ p>1 , w\in A_{p} if \sup_{B\subset\Omega}\left(\frac{1}{|B|}{\int_{B}w\ dx}\right)\left(\frac{1}{|B|}{\int_{B}w^{-\frac{1}{p-1}}}dx\right)^{p-1}<+\infty. (For example the weight w=d^{\beta}\in A_{2} when \beta\in(-1,1) , see Section 2 ). In [ 5 ] the authors proved some regularity result about weak solutions, including L^{\infty}(\Omega) regularity and Hölder regularity, under specific assumptions on f . Among these results, we highlight the one most relevant to our discussion, and we state it in the case w=d^{\beta} , although it holds for more general weights as well: Theorem 1.1 (Theorem 2.4.8 in [ 5 ] ) . Let u\in W^{1,2}_{0}(\Omega,d^{\beta}) be a weak solution of L_{\beta}(u)=-div\overrightarrow{f} , where |\overrightarrow{f}|/d^{\beta}\in L^{p}(\Omega,w) and p>2n-\varepsilon , for some \epsilon>0 . Then, u is Hölder continuous in \Omega . In this paper we do not concern ourselves with the optimal assumptions on f to ensure that the solution of ( 1.2 ) is continuous or bounded. Instead, we simply assume this regularity and then derive further properties from it. We will derive explicit estimates that describe the asymptotic behaviour of the solution as it approaches the boundary. These estimates provide upper and lower bounds for the solution u , highlighting the maximal regularity of the solution in terms of the parameters of the problem, in particular of the exponent \beta . The main result in this direction is stated in our main theorem, which follows. Theorem 1.2 . Suppose that u\in C^{2}(\Omega)\cap C(\overline{\Omega}) solves ( 1.2 ) with f\geq 0 , f\not\equiv 0 in \Omega , f\in C(\overline{\Omega}) and \beta<1 . Then, for any \eta_{1},\eta_{2}>0 , there exist \sigma\in(0,1) and two positive constants D_{1} and D_{2} such that (1.6) D_{1}d^{1-\beta}(x)(-\log d(x))^{-\eta_{1}}\leq u(x)\leq D_{2}d^{1-\beta}(x)(-\log d(x))^{\eta_{2}},\quad\forall x\in\Gamma_{\sigma}^{\circ}. where \Gamma_{\sigma}=\{x\in\overline{\Omega}:\ d(x)<\sigma\}. Moreover, if \Omega is convex, then (1.7) D_{1}d^{1-\beta}(x)(-\log d(x))^{-\eta_{1}}\leq u(x)\leq D_{2}d^{1-\beta}(x),\quad\forall x\in\Gamma_{\sigma}. The estimate ( 1.7 ) provides a precise control of the solution u on the boundary \partial\Omega . In our opinion it is interesting in itself, and in any case, we will use it to deduce some properties of the solution. A first immediate consequence concerns the maximal Hölder regularity that we may expect from the solution. As mentioned before, in [ 5 ] , under some assumptions on f , it was proved that the solution u\in C^{0,\alpha}(\overline{\Omega}) , for some \alpha\in(0,1) . We give an upper bound to the value \alpha : Theorem 1.3 . Suppose that u\in C^{2}(\Omega)\cap C^{0,\alpha}(\overline{\Omega}) solves ( 1.2 ) with f\geq 0 , f\not\equiv 0 in \Omega , f\in C(\overline{\Omega}) and \beta<1 . Then \alpha\leq\min\{1,1-\beta\}. We point out that the result is sharp, as proved in Remark 4.1 . Another consequence concerns the regularity of this solution within the framework of Sobolev spaces. The solution to equation ( 1.2 ) minimizes the functional F(u):=\frac{1}{2}\int_{\Omega}|\nabla u|^{2}d^{\beta}\ dx-\int_{\Omega}fu\ dx in the space W^{1,2}_{0}(\Omega,d^{\beta}) (see Section 2 ).

Read the complete text