State Space Disaggregation
Python code for progressive state-space disaggregation in infinite-horizon dynamic programming.
Code associated with progressive state-space disaggregation methods for solving large Markov Decision Processes. The project studies how coarse abstractions can be refined to recover useful state distinctions while keeping dynamic programming computations tractable.
The implementation supports numerical experiments around abstraction refinement, Bellman updates, and benchmark MDPs used in the ICAPS and EWRL work.
Area: approximate dynamic programming, state abstraction, Markov aggregation/disaggregation.
Link: GitHub profile fallback.