Multilevel methods for dynamic programming

The author studies multilevel methods for the solution of systems of linear equations arising from discrete probabilistic optimal control problems with a special emphasis of problems related to recommendation engines. This book features a sound and rigorous matrix-theoretic foundation, and novel convergence analyses of the proposed procedures. Moreover, the theoretical discussion is supported by numerical experiments. The research presented in this book has been carried out as a part of a joint project involving TU Hamburg-Harburg, TU Berlin, and the realtime analytics company prudsys AG.