Expand description
StateMap – adaptive state-to-probability mapper.
Maps each of 256 states to a 12-bit probability (1-4095) that adapts based on observed bits. Uses variable learning rate:
- Fast learning for low-count states (early observations matter more)
- Slow learning for high-count states (well-established statistics)
Initialization uses the StateTable’s static probability as the prior.
Structs§
- State
Map - Maps 256 states to adaptive 12-bit probabilities.