Weak KAM aspects of convex Hamilton-Jacobi equations with Neumann type boundary conditions (Q619894)

The author considers Hamilton-Jacobi equations with Neumann type boundary condition \[ \begin{aligned} H(x,Du(x))=a &\quad \text{in }\Omega,\tag{1}\\ D_\gamma u(x)=g(x) &\quad \text{on }\Gamma\tag{2} \end{aligned} \] and the initial-value problem with Neumann type boundary condition \[ \begin{aligned} u_t(x,t)+H(x,Du(x,t))=0 &\quad \text{for }(x,t)\in\Omega\times(0,T),\tag{3}\\ D_\gamma u(x,t)=g(x) &\quad \text{for }(x,t)\in\Gamma\times(0,T),\tag{4}\\ u(x,0)=u_0(x) &\quad \text{for }x\in\overline{\Omega},\tag{5} \end{aligned} \] where \(H\) is a convex and coercive Hamiltonian. The purpose of this paper is to establish some theorems concerning weak KAM theory for convex HJ equations. Indeed, the critical value (or the additive eigenvalue) and the Aubry set associated with (1), (2) are defined and some of basic properties of the Aubry set, representation formulas for solutions of (1), (2) and the existence of extremals (or minimizers) for variational formulas of certain types of solutions of (1), (2) are established As is somehow well known in optimal control, the solutions of (3)--(5) or (1), (2) should be given by the value function of the corresponding optimal control problem, whose dynamics is described by the Skorokhod problem. The author presents here a treatment of the Skorokhod problem suitable to the formulas for the solutions of (3)--(5) or of (1), (2), which is slightly different from usual treatments of the Skorokhod problems for applications to stochastic differential equations. One of main difficulties in the approach was how to recover the convexity of problems (1), (2) or (3)--(5) in the viscosity formulation. This difficulty is solved in two theorems and some comparison results for sub and supersolutions of (1), (2) and of (3)--(5) are also established.