Optimal transport of vector measures (Q2236836)

From MaRDI portal





scientific article
Language Label Description Also known as
default for all languages
No label defined
    English
    Optimal transport of vector measures
    scientific article

      Statements

      Optimal transport of vector measures (English)
      0 references
      26 October 2021
      0 references
      The motivations of this paper come from Chapter \(6\) of \textit{B. Klartag} [Needle decompositions in Riemannian geometry. Providence, RI: American Mathematical Society (AMS) (2017; Zbl 1457.53028)]. In the present paper and in [the author, J. Math. Pures Appl. (9) 154, 212--244 (2021; Zbl 1472.52005)] the author develops and studies a theory of optimal transport for vector measures, in connection with the localisation technique. In particular, it is proven here that optimal transport for \(\mathbb{R}^m\)-valued Borel measures admits a dual formulation and it is studied the Wasserstein space \(\mathcal{W}(X,\mathbb{R}^m)\) of vector-valued measures defined on a metric space \(X\). In contrast with the \(1\)-dimensional case, the author also shows that given a vector measure on the Euclidean space with total mass zero, the mass of a transport set may be different from zero, thus resolving in the negative a conjecture of Klartag. However, the mass balance condition for a measure \(\mu\in \mathcal{W}(\mathbb{R}^n,\mathbb{R}^m)\) turns out to be true under the assumption that there exists an optimal transport of \(\mu\) with marginals of its total variation that are absolutely continuous with respect to the Lebesgue measure.
      0 references
      vector measures
      0 references
      optimal transport
      0 references
      localisation technique
      0 references
      mass balance.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references