Optimal Transportation for Generalized Lagrangian

Abstract

In this paper, we study the optimal transportation for generalized Lagrangian $L=L(x, u,t)$, and consider the cost function as following: $$c(x, y)=\inf_{\substack{x(0)=x\\x(1)=y\\u\in\mathcal{U}}}\int_0^1L(x(s), u(x(s),s), s)ds.$$ Where $\mathcal{U}$ is a control set, and $x$ satisfies the following ordinary equation: $$\dot{x}(s)=f(x(s),u(x(s),s)).$$ We prove that under the condition that the initial measure $\mu_0$ is absolutely continuous w.r.t. the Lebesgue measure, the Monge problem has a solution, and the optimal transport map just walks along the characteristic curves of the corresponding Hamilton-Jacobi equation: \begin{equation*} \begin{cases} V_t(t, x)+\sup_{\substack{u\in\mathcal{U}}}<V_x(t, x), f(x, u(x(t), t),t)-L(x(t), u(x(t), t),t)>=0.\\ V(0,x)=\phi_0(x) \end{cases} \end{equation*}