Moment approximations in optimal transport
In this talk we study finite moment approximations of linear optimization problems over measures, building on the work of Alfonsi, Coyaud, Ehrlacher and Lombardi on moment-constrained optimal transport. Under suitable assumptions, these approximations admit discrete minimizers whose number of support points is controlled by the number of constraints, but they may fail to preserve important geometric properties.
We show that this framework accommodates several transport problems, including partial transport, capacity-constrained transport and martingale transport. For suitable moment functions, we establish a $W_p$--$W_\infty$ estimate analogous to that of Bouchitté, Jimenez and Rajesh, with an additive error. For repulsive costs, under suitable assumptions on the prescribed marginal, we prove that minimizers remain separated from the diagonal, uniformly in the number of moment constraints. Finally, under a moment-correction assumption and for a fixed family of moment functions, we show that the minimal fermionic energies converge in the semiclassical limit to the minimum value of the corresponding moment-constrained optimal transport problem with Coulomb cost.
This talk is based on joint work with Luca Nenna and Luigi De Pascale.

