The paper “Convex Optimization for Parameter Synthesis in MDPs” by Murat Cubuktepe, Nils Jansen, Sebastian Junges, Joost-Pieter Katoen and Ufuk Topcu has been accepted for publication in the journal IEEE Transactions on Automatic Control (impact factor 5,8).
This paper formulates the feasibility problem for parametric MDPs (pMDPs) problem as a non-convex quadratically constrained quadratic program (QCQP) and develops two different solution procedures: sequential convex programming (SCP) and linear programming (LP). The latter technique is shown to scale to models with hundreds of thousands of states and thousands of parameters.