Wasserstein Convergence Rates for the EMCEL and Perturbed EMCEL Algorithms

We discuss the convergence rates in every p-th Wasserstein distance of the EMCEL and related algorithms. For time marginals, we get the rate of 1/4; on the path space, any rate strictly smaller than 1/4. These rates apply also in irregular situations such as, e.g., an SDE with irregular coefficients, sticky Brownian motion, a Brownian motion slowed down on the Cantor set. In contrast to the previous talk, we need to impose an additional assumption on the strong Markov process to be approximated, which is essential, and altogether the treatment of the rates requires different techniques. This is a joint work with Stefan Ankirchner and Thomas Kruse.

Event Topic

Mathematical Finance, Stochastic Analysis, and Machine Learning