Reversing Inhomogeneous and Imprecise Markov Chains

Reversal of a Markov chain means observing the same process backwards in time. A class of Markov chains whose reverses adhere to the same probabilistic laws as the original processes are called reversible Markov chains. In this article, a reversal of imprecise Markov chains is proposed. This is done by reversing a set of compatible inhomogeneous Markov chains. The reversed processes can be explicitly characterized if the interpretation of strong independence is adopted. However, the reversed processes generally lack some important properties such as the Markov property.