Notice: Unexpected clearActionName after getActionName already called in /var/www/html/includes/context/RequestContext.php on line 338
A general class of free boundary problems for fully nonlinear parabolic equations - MaRDI portal

A general class of free boundary problems for fully nonlinear parabolic equations (Q2353832)

From MaRDI portal
(Redirected from Item:Q2453517)
scientific article; zbMATH DE number 6302105
  • A general class of free boundary problems for fully nonlinear elliptic equations
Language Label Description Also known as
English
A general class of free boundary problems for fully nonlinear parabolic equations
scientific article; zbMATH DE number 6302105
  • A general class of free boundary problems for fully nonlinear elliptic equations

Statements

A general class of free boundary problems for fully nonlinear parabolic equations (English)
0 references
A general class of free boundary problems for fully nonlinear elliptic equations (English)
0 references
0 references
0 references
9 July 2015
0 references
6 June 2014
0 references
Let \(\Omega \subset \mathbb{R}^{n+1}\) be unknown open set, \ \(Q_r(X):=B_r(x) \times (t-r, t)\subset \mathbb{R}^{n}\times \mathbb{R}\) -- parabolic ball of radius \(r\) with a center at a point \(X = (x,t)\) and \(Q_r=Q_r(0)\). There is studied free boundary problem with \(u \in W^{2,n}_x \cap W^{1,n}_t \) satisfying \[ \mathcal{H}(u) = 1 \;\text{{a.e. \;in}} \;Q_1\cap \Omega, \;\;|\widetilde{D}{}^2 u| \leq K \;\text{{a.e. \;in}} \;Q_1 \setminus \Omega, \eqno(1) \] where \(\mathcal{H}(u) := F(D^2 u) - \partial_t u\) is a fully nonlinear parabolic operator, \ \(\widetilde{D}{}^2 u = (D^2_x, \, D_t u) \subset \mathbb{R}^{n+1}\), \ \(K > 0\). For the solution of the problem (1) \(\widetilde{D}{}^2 u = 0\) a.e. inside \(\{u=0\}\) and \(D^2_x u = 0\) a.e. inside \(\{\nabla_x u=0\}.\) Let \[ \delta_r(u,X^0):= \inf_{t\in [t_0-r^2,\, t_0+r^2 ]}\!\frac{MD\big(\Lambda \cap \big(B_r(x^0)\times \{t\}\big)\big)}{r}, \] where \(\Lambda:=Q_1\setminus \Omega, \;X^0=(x^0,t^0)\),\ \(MD(E)\) is the smallest distance between two parallel hyperplanes containing \(E\). The authors prove that if \(u \in W^{2,n}_x \cap W^{1,n}_t\) is a solution of the problem (1), {\parindent=6mm \begin{itemize} \item{ 1)} then \(|\widetilde{D}{}^2 u| \leq \widetilde{C}\) \ in \ \( Q_{1/2}\), where \(\widetilde{C}=\widetilde{C}(n,\,\lambda_0,\,\lambda_1,\, \parallel u \parallel_\infty) >0\), \ \(\lambda_0,\,\lambda_1\), \ (\(0 < \lambda_0 \leq \lambda_1 < \infty\)) are ellipticity constants;\item [2)] if also \(\Omega \supset \{ u \not= 0\}\) and there exists \(\varepsilon > \) such that \( \delta_r(u,z) > \varepsilon \;\forall r < 1/4, \;\forall z\in \partial\Omega\cap Q_r, \) then \(\partial\Omega \cap Q_{r_0}\) is a \(C^1\) -- graph in space-time, where \(r_0\) depends on \(\varepsilon\) and the data. \end{itemize}}
0 references
The aim of the paper is to prove optimal regularity for \(W^{2,n}\) solutions \(u:B_1\to\mathbb R\) to the free boundary problem \(F(D^2u)=1\) a.e. in \(B_1\cap\Omega\), \(|D^2u|\leq K\) a.e. in \(B_1\setminus\Omega\), where \(K\) is a positive constant, \(B_1\) is the unit ball in \(\mathbb R^n\), \(\Omega\) is an unknown open set in \(\mathbb R^n\), \(F(0)=0\), \(F\) is uniformly elliptic and either convex or concave. This class of equations covers both the obstacle problem (right hand side given by \(\chi_{[u\neq0]}\)) and the more general free boundary problem with the right hand side of the form \(\chi_{[\nabla u\neq0]}\) studied by \textit{L. Caffarelli} and \textit{J. Salazar} [Trans. Am. Math. Soc. 354, No. 8, 3095--3115 (2002; Zbl 0992.35101)]. The two main results concern the interior \(C^{1,1}\) regularity of \(u\) and the free boundary regularity. The former one says that under the above assumptions the solution \(u\) possesses the interior \(C^{1,1}\) regularity, namely \(\|D^2u\|\leq C\) with \(C\) dependent on \(K\), the ellipticity constants of \(F\) and \(\|u\|_{L^\infty(B_1)}\) only. To obtain the free boundary regularity result the authors introduce the following concept of thickness of the set \(B_1\setminus\Omega\) in \(B_r(x)\) for \(r>0\): \(\delta_r(u,x)=MD(B_r(x)\cap B_1\setminus\Omega)r^{-1}\) where \(MD(E)\) is the smallest possible distance between two parallel hyperplanes containing the set \(E\). They prove that if \(F\) is convex, either \(\Omega\supset\{\nabla u\neq0\}\) and \(F\in C^1\) or \(\Omega\supset\{u\neq0\}\), and \(\delta_r(u,x)>\varepsilon\) for some \(\varepsilon>0\) and for all \(x\in\partial\Omega\cap B_r(0)\), \(r<1/4\), then there exists \(r_0>0\) dependent only on \(\varepsilon\) and the data such that \(\partial\Omega\cap B_{r_0}(0)\) represents a \(C^1\)-graph.
0 references
free boundary problem
0 references
fully nonlinear parabolic equation
0 references
regularity of the solution
0 references
free boundary
0 references
fully nonlinear elliptic equations
0 references
uniform ellipticity
0 references
thickness
0 references
regularity
0 references
0 references
0 references
0 references
0 references
0 references

Identifiers

0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references