Abstraction-Guided Truncations for Stationary Distributions of Markov Population Models

To understand the long-run behavior of Markov population models, computation stationary distribution is often a crucial part. We propose truncation-based approximation that employs state-space lumping scheme, aggregating states in grid structure. The resulting approximate used to iteratively refine relevant and truncate irrelevant parts state-space. This way, algorithm learns well-justified finite-state projection tailored behavior. demonstrate method’s applicability wide range non-linear problems with complex behaviors.