English

A note on duality theorems in mass transportation

Probability 2019-07-17 v1

Abstract

The duality theory of the Monge-Kantorovich transport problem is investigated in an abstract measure theoretic framework. Let (X,F,μ)(\mathcal{X},\mathcal{F},\mu) and (Y,G,ν)(\mathcal{Y},\mathcal{G},\nu) be any probability spaces and c:X×YRc:\mathcal{X}\times\mathcal{Y}\rightarrow\mathbb{R} a measurable cost function such that f1+g1cf2+g2f_1+g_1\le c\le f_2+g_2 for some f1,f2L1(μ)f_1,\,f_2\in L_1(\mu) and g1,g2L1(ν)g_1,\,g_2\in L_1(\nu). Define α(c)=infPcdP\alpha(c)=\inf_P\int c\,dP and α(c)=supPcdP\alpha^*(c)=\sup_P\int c\,dP, where inf\inf and sup\sup are over the probabilities PP on FG\mathcal{F}\otimes\mathcal{G} with marginals μ\mu and ν\nu. Some duality theorems for α(c)\alpha(c) and α(c)\alpha^*(c), not requiring μ\mu or ν\nu to be perfect, are proved. As an example, suppose X\mathcal{X} and Y\mathcal{Y} are metric spaces and μ\mu is separable. Then, duality holds for α(c)\alpha(c) (for α(c)\alpha^*(c)) provided cc is upper-semicontinuous (lower-semicontinuous). Moreover, duality holds for both α(c)\alpha(c) and α(c)\alpha^*(c) if the maps xc(x,y)x\mapsto c(x,y) and yc(x,y)y\mapsto c(x,y) are continuous, or if cc is bounded and xc(x,y)x\mapsto c(x,y) is continuous. This improves the existing results in \cite{RR1995} if cc satisfies the quoted conditions and the cardinalities of X\mathcal{X} and Y\mathcal{Y} do not exceed the continuum.

Keywords

Cite

@article{arxiv.1907.07059,
  title  = {A note on duality theorems in mass transportation},
  author = {Pietro Rigo},
  journal= {arXiv preprint arXiv:1907.07059},
  year   = {2019}
}

Comments

To appear

R2 v1 2026-06-23T10:22:17.553Z