KTH Mathematics

Optimal control and zero-sum games for Markov chains of mean-field type B Djehiche, S E Choutri, H Tembine arxiv.org/abs/1606.04244 Summary: When controlling a Markov chain of mean-field type, one actually controls a change of measure. By starting out from a class of markov chains that have a strong solution, a change of measure gives a pure jump process with intensities depending on the law and a control. Note that the pure jump process need not be a Markov chain, since the intensities are allowed to depend on the path of the process. The change of measure is defined by a Girsanov transformation and in order to this to be well defined, extra care had to be taked to ensure it had enough integrability. The usual condition was not satisfied by the unbounded intensities considered. List of papers Front page

Published by: Alexander Aurell Updated: 09-02-2018