Finitely additive mass transportation

Publication date 3 June 2022

Measures and integrals in product spaces (28A35) Integration theory via linear functionals (Radon measures, Daniell integrals, etc.), representing set functions and measures (28C05) Linear optimal control problems (49N05) Probabilistic measure theory (60A10) Probability distributions: general theory (60E05) Martingales with discrete parameter (60G42)

Abstract: Some classical mass transportation problems are investigated in a finitely additive setting. Let

O m e g a = p r o d_{i = 1}^{n} O m e g a_{i}

and

m a t h c a l A = o t i m e s_{i = 1}^{n} m a t h c a l A_{i}

, where

(O m e g a_{i}, m a t h c a l A_{i}, m u_{i})

is a (

s i g m a

-additive) probability space for

i = 1, l d o t s, n

. Let

c : O m e g a i g h t a r r o w [0, i n f t y]

be an

m a t h c a l A

-measurable cost function. Let

M

be the collection of finitely additive probabilities on

m a t h c a l A

with marginals

m u_{1}, l d o t s, m u_{n}

. If couplings are meant as elements of

M

, most classical results of mass transportation theory, including duality and attainability of the Kantorovich inf, are valid without any further assumptions. Special attention is devoted to martingale transport. Let

(O m e g a_{i}, m a t h c a l A_{i}) = (m a t h b b R, m a t h c a l B (m a t h b b R))

for all

i

and

M_1=�igl{Pin M:Pll P^* ext{ and }(pi_1,ldots,pi_n) ext{ is a }P ext{-martingale}}

where

P^{*}

is a reference probability on

m a t h c a l A

. If

M_{1} e e m p t y s e t

, then

int c,dP=inf_{Qin M_1}int c,dQquadquad ext{for some }Pin M_1.

Conditions for

M_{1} e e m p t y s e t

are given as well.

This page was built for publication: Finitely additive mass transportation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6401053)