A Choquet theory of Lipschitz-free spaces
Richard J. Smith
TL;DR
The paper develops a Choquet-like framework for Lipschitz-free spaces by leveraging De Leeuw representations on $\beta\widetilde{M}$ and introducing a Choquet-type quasi-order $\preccurlyeq$ on Radon measures. It builds a function cone $G$, constructs a corresponding compactification $\widetilde{M}^G$, and analyzes minimal measures (instead of the classical focus on maximal ones) to obtain well-behaved representations and shadow/extended-support structure. Key results include a characterization of minimality via a Choquet-type test, the possibility of mutually singular marginals for optimal/minimal representations, and a plan to decompose free-space elements into convex integrals of molecules. These tools culminate in establishing that every free-space element has optimal representations concentrated on $\mathfrak{p}^{-1}(M\times M)$, enabling a robust extreme-point analysis and laying groundwork for a complete description of extreme points in $B_{\mathcal{F}(M)}$. The framework thus provides a deeper structural understanding of Lipschitz-free spaces with potential applications to convex decompositions and point-extremality questions.
Abstract
Let $(M,d)$ be a complete metric space and let $\mathcal{F}(M)$ denote the Lipschitz-free space over $M$. We develop a ``Choquet theory of Lipschitz-free spaces'' that draws from the classical Choquet theory and the De Leeuw representation of elements of $\mathcal{F}(M)$ (and its bidual) by positive Radon measures on $β\widetilde{M}$, where $\widetilde{M}$ is the space of pairs $(x,y) \in M \times M$, $x \neq y$. We define a quasi-order $\preccurlyeq$ on the positive Radon measures on $β\widetilde{M}$ that is analogous to the classical Choquet order. Rather than in the classical case where the focus lies on maximal measures, we study the $\preccurlyeq$-minimal measures and show that they have a host of desirable properties. Among the applications of this theory is a solution (given elsewhere) to the extreme point problem for Lipschitz-free spaces.
