The Setting
For a prescribed particle distribution, the Coulomb transport problem compares arbitrary joint laws with transports determined by the position of the first particle.
The Definition
Definition 1. The particle space is Euclidean three-space, as in (1). (1)✦ A three-particle configuration is a point of the space in (2); its coordinate projections are , and . (2)✦
For , the extended inverse distance is (3). (3)✦ The Coulomb cost is the sum in (4), with each reciprocal interpreted as in (3). (4)✦
For a real function , its density measure is (5), where and denotes Lebesgue measure; for nonnegative write . (5)✦
Write for the Borel probability measures on a Euclidean space . A three-marginal coupling of a Borel measure on is a probability measure on satisfying the marginal conditions in (6). (6)✦
A measure-preserving map for is a Borel map satisfying the pushforward condition in (7); denotes this class. (7)✦ For maps , the graph cost is the nonnegative extended integral (8). (8)✦
The Kantorovich value is the infimum over couplings in (9). (9)✦ The Monge value is the infimum over preserving maps in (10). (10)✦
A smooth compact probability density on is a real function belonging to the class in (11). (11)✦
The Kantorovich value for is attained when there exists a coupling satisfying (12). (12)✦ A Borel measure has no Monge optimizer when every pair of preserving maps satisfies the strict cost comparison (13). (13)✦
A Coulomb counterexample is a real function satisfying (14). (14)✦
In the paper, the result reads: for a Borel probability measure on , consists of probability measures on with all three marginals equal to , the cost is with value at every collision, is its infimum over , an admissible Monge plan is for Borel maps satisfying , and the density class requires , and .
What Is Checked
The counterexample condition (14) is stated for real densities on , with couplings from (6) and preserving maps from (7).