Mean-field neural networks-based algorithms for McKean-Vlasov control problems *

arXiv:2212.11518v2 Announce Type: replace-cross
Abstract: This paper is devoted to the numerical resolution of McKean-Vlasov control problems via the class of mean-field neural networks introduced in our companion paper [25] in order to learn the solution on the Wasserstein space. We propose several algorithms either based on dynamic programming with control learning by policy or value iteration, or backward SDE from stochastic maximum principle with global or local loss functions. Extensive numerical results on different examples are presented to illustrate …

abstract algorithms arxiv class companion control dynamic learn math.oc mean networks neural networks numerical paper programming q-fin.cp solution space stat.ml type via