Iterative Solution of Markov Decision Processes

This bachelor thesis discusses the application of different iterative solvers to the Policy
iteration, which is commonly used to optimize the so called Markov Decision Processes
(MDPs). In general, MDPs provide us with a mathematical framework used to model decision-making in stochastic environments, which we will be discussing in detail in this thesis.
Our focus lies on analysing and discussing the Jacobi, Gauss-Seidel, Richardson and Krylov subspace methods. We are especially interested in understanding which model is most efficient depending on the problem we want to solve and how the different iterators can be interpreted in regards to our model. Meaning, we aim to describe what happens with our MDP in the process of the linear system being solved.