On the strong maximum principle for degenerate parabolic equations (Q2460012)

The classical strong maximum principle states that if a (sub)solution to the equation \[ u_t+F( t,x,u,D_xu,D_x^2u) =0,\quad ( t,x) \in \Omega \subset \mathbb{R}^{d+1}, \] achieves a maximum at an interior point \(( t_0,x_0) \) in \(\Omega, \) then the maximum is achieved on every curve in \(\Omega \) with endpoint \( ( t_0,x_0) \) along which the first coordinate \(t\) is nondecreasing. In this paper, the author establishes variants of the strong maximum principle for subsolutions to degenerate parabolic equations for which the standard version of the strong maximum principle does not hold. The author formulates his results for equations of the form \[ G( t,x,u,u_t,D_xu,D_x^2u) =0,\quad ( t,x) \in \Omega \subset \mathbb{R}^{d+1}. \] In the paper the solutions are considered in the viscosity sense, i.e., for \(u\) to be a solution of \(( 1) \) it is required that it is both sub and supersolution, and \(u\) is a subsolution in \(\Omega \) provided \(u \) is upper semicontinuous and \[ G( t_0,x_0,u( t_0,x_0) ,\varphi _t( t_0,x_0) ,D_x\varphi ( t_0,x_0) ,D_x^2\varphi ( t_0,x_0) ) \leq 0, \] whenever \(\varphi \) is twice continuously differentiable and the function \( u-\varphi \) has a local maximum at the point \(( t_0,x_0) \in \Omega .\)