L.
Nenna: From (optimal) fat plans to optimal maps for
the multi-marginal optimal transport with Coulomb
cost
Abstract:
In this talk I will present the
multi-marginals (Monge-Kantorovich) optimal transport
problem (with a generic cost function) and a numerical
method to solve this kind of problem based on the
discretization of the entropic functional (also known as
Kullback-Leibler distance). Then, I will introduce the
Density Functional Theory (DFT) which is an approximate
computational method for solving the many-electron
Schrödinger equation at a more reasonable cost.
One scenario of interest for the DFT is when the repulsion
between electrons largely dominates over the kinetic
energy. In this case the problem can be reformulated as a
multi-marginal OT problem with the Coulomb
cost. I will show that the entropic functional,
firstly used as a numerical trick, has actually an
interesting physical meaning. Finally, I will present some
numerical results highlighting that the solution of the OT
problem with Coulomb cost can be either a "fat" transport
plan or a transport map.
|