The Setting
Three-particle transport with a common marginal compares probability measures on configurations with configurations prescribed by two maps of the first particle. The configuration space is , with points ; write , and for its first, second and third coordinate projections.
The extended Coulomb cost on is the function given by (1), with value at every collision: (1)✦ The inverse distance between points is the reciprocal of their Euclidean distance, with value when the points coincide.✦
A three-marginal self-coupling of a Borel probability measure on is a probability measure on whose three coordinate marginals equal .✦ A measure-preserving map for is a Borel map with .✦ The coupling class and the class of admissible pairs of maps are given by (2): (2)
For a Borel probability measure on , the coupling cost of and the Monge transport cost of are the nonnegative extended integrals in (3). The Kantorovich value and the Monge value of a Borel probability measure on are the infima of the coupling and Monge transport costs, respectively, given by (4). (3) (4)
For , a configuration in is -separated when the inequalities in (5) hold: (5)✦ A Borel probability measure on admits finite-cost Monge approximation if, for every real , there is a pair with finite cost at most .✦
Counterexample Densities
The density measure of a real function on is the measure obtained by weighting Lebesgue measure by ; for nonnegative write .✦ A smooth compact probability density on is a nonnegative real function such that and are both smooth and compactly supported, and .✦
A Borel probability measure on has Kantorovich attainment if some has .✦ A Borel probability measure on has Monge nonattainment if for every .✦ A Coulomb counterexample density is a smooth compact probability density whose measure is a probability measure with finite , Kantorovich attainment and Monge nonattainment.✦ A smooth compact probability density on satisfies the full Coulomb conclusion if is a Coulomb counterexample density and its measure satisfies and admits finite-cost Monge approximation.✦
The Claim
Theorem 1. Let be a metric space whose Borel -algebra makes it a standard Borel space, and let be a nonempty standard Borel space. Let be a finite atomless measure on , a finite measure on , and a measurable map with . Suppose is compact and . For every there is a measurable map satisfying (6): (6)✦
Let be a Borel probability measure on , , and measurable with . Put and . Suppose , , is concentrated on -separated configurations, and for -almost every . The pair satisfies (7), where : (7)✦
Let be an atomless Borel probability measure on . Suppose is a compact set with , and satisfies and is concentrated on -separated configurations for some . Then (8) holds: (8)✦
For any Borel probability measure on admitting finite-cost Monge approximation, the infima satisfy (9): (9)✦
Let be a Coulomb counterexample density and . Suppose satisfies and is concentrated on -separated configurations for some . Then (10) holds: (10)✦
In the paper, the result reads: for the probability measure constructed in Theorem 1.1, the infimum of over Borel maps with equals ; every admissible pair has cost strictly above this common infimum, but finite-cost pairs approach it.
What Is Checked
The Coulomb assertions are stated over Borel self-couplings and measure-preserving maps of , including the density measures , and the lifting assertion is over finite standard Borel measures with compact concentration of the atomless marginal.