Sharp Wasserstein estimates for integral sampling and Lorentz summability of transport densities (Q2108031)

In the main result of the paper, the author proves that for any two probability measures \(\mu,\nu\) on a convex, compact domain \(\Omega\in \mathbb{R}^d\), \(d\ge 2\), with \(\mu\in L^{d',1}\), there exist a constant \(C=C(d)\) and a geodesic \(\mu_t\) in the \(p\)-Wasserstein space, \(p\in [1,\infty]\), connecting \(\nu\) to \(\mu\) and such that \[ \left\|\int_0^1\mu_t\, dt\right\|_{ L^{d',\infty}}\le C\|\mu\|_{ L^{d',1}}. \] Here \(L^{p,q}\) is the Lorentz space of orders \(p\) and \(q\), endowed with the associated quasinorm \(\|\cdot\|_{ L^{p,q}}.\) A simple corollary of this result answers a conjecture of \textit{S. Steinerberger} [J. Math. Anal. Appl. 501, No. 2, Article ID 125185, 14 p. (2021; Zbl 1466.49043)] by providing an upper bound of the quantity \[ \left|\int \phi \,d\mu-\int \phi \,d\nu\right| \] in terms of Lorentz quasinorms of \(\mu\), \(\nabla \phi\) and the \(\infty\)-Wasserstein distance between \(\mu\) and \(\nu\).