Module region_mdp

Region MDP: Markov Decision Process over HNSW semantic regions (RFC-013 Part A).

HNSW regions define a discrete state space. Episode trajectories define transitions. Rewards define outcomes. This module learns P(success | region, action_type) from observed trajectories.

States are HNSW hub node IDs at a given level. Actions are abstract types represented as string labels (e.g., “navigate”, “take”, “place”).

Structs

RegionMdp

A transition model over HNSW regions.