arXiv reader

Lipschitz continuity results for a class of obstacle problems

Carlo Benassi and Michele Caselli

arXiv
2210.06318v1 · PDF · 12 October 2022
Categories
math.AP
Journal
Atti Accad. Naz. Lincei Cl. Sci. Fis. Mat. Natur. 31 (2020), no. 1, pp. 191-210
DOI
10.4171/rlm/885
Licence
CC BY 4.0 The site shows the text.

Abstract

We prove Lipschitz continuity results for solutions to a class of obstacle problems under standard growth conditions of p-type, p \geq 2. The main novelty is the use of a linearization technique going back to [28] in order to interpret our constrained minimizer as a solution to a nonlinear elliptic equation, with a bounded right-hand side. This leads us to start a Moser iteration scheme which provides the L^\infty bound for the gradient. The application of a recent higher differentiability result [24] allows us to simplify the procedure of the identification of the Radon measure in the linearization technique employed in [32]. To our knowledge, this is the first result for non-autonomous functionals with standard growth conditions in the direction of the Lipschitz regularity.

Items 3

Theorem

Theorem 2.1

GDQQ

Let u \in \mathcal{K}_{\psi, \Psi}(\Omega) be a solution to the obstacle problem (?), under the assumptions (?)-(?). If \psi \in W^{2, \infty}_{\rm loc}(\Omega), then u \in W^{1, \infty}_{\textnormal{loc}}(\Omega) and the following estimate

\|(1+|Du|^2)^{\frac{1}{2}}\|_{L^{\infty}(B_{\rho})} \le \, \frac{C}{(R-\rho)^{\beta}} \, \left ( \int_{B_{R}} (1 + |Du|^2)^{\frac{p}{2}} \, dx \right )^{\frac{1}{p}}

holds for every 0 < \rho < R and with positive constants C and \beta depending on n, p, \lambda, \Lambda, R, \rho and the local bounds for \|D \psi\|_{W^{1, \infty}}.

Remark

Remark 2.2

MFRE

Due to the local nature of our main result, we do not assume further regularity on the boundary datum \Psi.

Theorem

Theorem 6.1

H9N7

Let f satisfy the growth conditions (?), (?), (?), and let f(x, \xi) be a convex C^2 function in its second argument. Then the sequence f^{\ell k}: \Omega \times \mathbb{R}^n \rightarrow [0, + \infty) defined in (?), convex in the last variable is such that f^{\ell k} \to f pointwise as (\ell, k) \to + \infty for a.e. x \in \Omega and for all \xi \in \mathbb{R}^n. Moreover, f^{\ell k} \to f uniformly in \Omega_0 \times K for every \Omega_0 \subset \subset \Omega and K compact set of \mathbb{R}^n. Moreover:

  • There exist constants M_1, M_2 >0, independent of k,\ell such that

    M_1(1+|\xi|^2)^{\frac{p}{2}} \le f^{\ell k}(x, \xi) \le M_2 (1 + |\xi|^2)^{\frac{p}{2}} \qquad \textnormal{for a.e. $x \in \Omega$, for all $\xi \in \mathbb{R}^n$},
  • there exist constants \lambda_1, \Lambda_1>0 such that for all (x, \xi) \in \Omega \times \mathbb{R}^n and \mu \in \mathbb{R}^n

    \lambda_1 \, (1 + |\xi|^2)^{\frac{p-2}{2}} |\mu|^2 \le \sum_{i,j} f^{\ell k}_{\xi_i \xi_j}(x, \xi) \mu_i \mu_j ,

    and

    |f^{\ell k}_{\xi_i \xi_j}(x, \xi)| \le \Lambda_1 \, (1 +|\xi|^2)^{\frac{p-2}{2}}.
  • There exists a constant \Lambda_2>0 such that for all (x, \xi) \in \Omega \times \mathbb{R}^n

    |f^{\ell k}_{\xi x}(x, \xi)| \le \Lambda_2 (1 +|\xi|^{2})^{\frac{p-1}{2}}.

Source and licence

The text on this page is from arXiv:2210.06318v1 by Carlo Benassi and Michele Caselli, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2210/2210.06318.jsonl.