Maximal generalized solutions of Hamilton-Jacobi equations (Q2374342)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Maximal generalized solutions of Hamilton-Jacobi equations |
scientific article |
Statements
Maximal generalized solutions of Hamilton-Jacobi equations (English)
0 references
15 December 2016
0 references
Let \(U \subset \mathbb{R}\) be open and bounded. Take \(H(x,p): U \times \mathbb{R}^n \to \mathbb{R}\) measurable in \(x\) for all \(p\), and convex/superlinear in \(p\) for all \(x\). The author studies \[ (\ast) \begin{cases} H\left(x,Du(x)\right) = 0 & \text{ in } U, \\ u=v & \text{ on } \partial U, \end{cases} \] where \(v \in W^{1,\infty}\left(U\right)\) satistifes \[ H\left(x,Dv(x)\right) \leq 0 \;\text{ in } U \;. \] Here \(u\) is a maximal generalized solution meaning that \(u\) is maximal amongst functions in \(W^{1,\infty}\left(U\right)\) satisfying \((\ast)\) for almost every \(x \in \overline{U}\). For \(H(x,p)\) discontinuous in \(x\), there may not exist a viscosity solution. However for \(H(x,p)\) continuous in \(x\), the maximal generalized solution agrees with the unique viscosity solution. The author extends a study of \[ \begin{cases} \left|Du(x)\right|^2 - a(x) = 0 & \text{ in } U, \\ u=v & \text{ on } \partial U \end{cases} \] in [\textit{S. Zagatti}, J. Differ. Equations 257, No. 1, 231--263 (2014; 1288.35193)] and \[ \begin{cases} H\left(Du(x)\right) - a(x) = 0 & \text{ in } U, \\ u=v & \text{ on } \partial U \end{cases} \] in [\textit{S. Zagatti}, Adv. Pure Appl. Math. 7, No. 2, 123--141 (2016; 1338.35109)] for \(a \in L^{\infty}\left(U\right)\). These problems are motivated by problems in the calculus of variation [\textit{S. Zagatti}, Calc. Var. Partial Differ. Equ. 31, No. 4, 511--519 (2008; 1139.49017)]. Here the author determines a minimizer of \[ I\left[ u \right] = \int_U \, L\left(x,Du(x) \right) \,dx \] from a minimizer of \[ \widetilde{I}\left[ u \right] = \int_U \, \widetilde{L}\left(x,Du(x) \right) \,dx \] satisfying \[ (\ast \ast) \;\;\widetilde{L}\left(x,Du(x) \right) - L\left(x,Du(x) \right) = 0, \] where \(\widetilde{L}\) is the quasi-convex envelope of \(L\) with respect to \(p\). Since minimizers can be irregular, \((\ast \ast)\) should be cast as \((\ast)\). The approach involves viscosity solutions \(u^{(n)}\) of \[ \begin{cases} H^{(n)}\left(x,Du^{(n)}(x)\right) = 0 & \text{ in } U, \\ u^{(n)}=v & \text{ on } \partial U \end{cases} \] for \(H^{(n)}\) regularizations of \(H\). The author extracts a subsequence converging weak \(\ast\) in \(W^{1,\infty}\). Convexity of \(H(x,p)\) in \(p\) shows that the limit is a subsolution. Furthermore, convergence of \(Du^{(n)}\) almost everywhere shows that the limit is the maximal generalized solution.
0 references
maximality
0 references
strong convergence
0 references
fully nonlinear PDE
0 references
0 references
0 references
0 references
0 references