Parametrization of random vectors in polynomial chaos expansions via optimal transportation.

Polynomial chaos (PC) expansions are used for the propagation of uncertainty through dynamical systems as an alternative to Monte Carlo methods. Model parameters in a given dynamical system are assumed to have known expansions, which correspond to simple standard distributions, and one is usually interested in the polynomial expansion of the system solution. We are concerned with the problem of estimating the PC expansion of a parameter vector when only realizations from its distribution are given. To this end we apply ideas from optimal transportation theory and network optimization