A Self-dual Polar Factorization for Vector Fields

TL;DR

This paper introduces a self-dual polar factorization for vector fields, representing them via measure-preserving involutions and a convex-concave Hamiltonian, extending Brenier's polar decomposition with a self-dual structure.

Contribution

It provides a novel self-dual polar decomposition for vector fields using measure-preserving involutions and a convex-concave Hamiltonian, generalizing Brenier's classical result.

Findings

Any non-degenerate vector field in L-infinity can be expressed via a measure-preserving involution and a Hamiltonian.

Monotone vector fields correspond to the involution being the identity, linking to classical convex analysis.

The decomposition can be reformulated as a self-dual mass transport problem.

Abstract

We show that any non-degenerate vector field $u$ in $ L^{\infty}(\Omega, \R^N)$, where $\Omega$ is a bounded domain in $\R^N$, can be written as {equation} \hbox{$u(x)= \nabla_1 H(S(x), x)$ for a.e. $x \in \Omega$}, {equation} where $S$ is a measure preserving point transformation on $\Omega$ such that $S^2=I$ a.e (an involution), and $H: \R^N \times \R^N \to \R$ is a globally Lipschitz anti-symmetric convex-concave Hamiltonian. Moreover, $u$ is a monotone map if and only if $S$ can be taken to be the identity, which suggests that our result is a self-dual version of Brenier's polar decomposition for the vector field $u$ as $u(x)=\nabla \phi (S(x))$, where $\phi$ is convex and $S$ is a measure preserving transformation. We also describe how our polar decomposition can be reformulated as a self-dual mass transport problem.