POMDP Policy Solver

A decision engine that models sequential decisions as a partially observable Markov decision process, explicitly representing history-dependent dynamics.

Responsibility. Computes policies for sequential decisions under partial observability and history dependence.

Also known as: Partially Observable MDP solver

Variant of Decision Engine abstract

When to choose. Choose when the problem is non-Markovian or history-dependent (e.g., treatment response depending on prior treatments) and state augmentation is insufficient.