Non-symmetric intrinsic Hopf-Lax semigroup vs. intrinsic Lagrangian

Abstract

In this paper, we analyze the 'symmetrized' of the intrinsic Hopf-Lax semigroup introduced by the author in the context of the intrinsically Lipschitz sections in the setting of metric spaces. Indeed, in the usual case, we have that $d(x,y) =d(y,x)$ for any point $x$ and $y$ belong to the metric space $X$; on the other hand, in our intrinsic context, we have that $d(f(x),\pi^{-1} (y)) \ne d(f(y),\pi^{-1} (x)),$ for every $x,y \in X$. Therefore, it is not trivial that we get the same result obtained for the "classical" intrinsic Hopf-Lax semigroup, i.e., the 'symmetrized' Hopf-Lax semigroup is a subsolution of Hamilton-Jacobi type equation. Here, an important observation is that $f$ is just a continuous section of a quotient map $\pi$ and it can not intrinsic Lipschitz. However, following Evans, the main result of this note is to show that the "new" intrinsic Hopf-Lax semigroup satisfies a suitable variational problem where the functional contained an intrinsic Lagrangian. Hence, we also define and prove some basic properties of the intrinsic Fenchel-Legendre transform of this intrinsic Lagrangian that depends on a continuous section of $\pi$.