Ciorap Bot - UCI chess engine
Author: Vlad Ciocoiu
Removed null-move and futility pruning (will probably add them back later, but right now they are worsening its performance)
Added evaluation parameter tuning code
Fixed small bugs in information display
Improved the source code with more asserts, and less useless imported libraries
Expected elo gain: ~150
Added evaluation parameter tuning code
Fixed small bugs in information display
Improved the source code with more asserts, and less useless imported libraries
Expected elo gain: ~150
Features
- Hybrid move generation using bitboards and mailbox
- Piece evaluation using piece-square tables and mobility bonus
- Pawn structure evaluation that recognizes passed / weak / doubled pawns
- King safety evaluation
- All evaluation parameters tuned using supervised learning
- Move searching using alpha-beta algorithm and Principal Variation Search
- Iterative deepening and aspiration windows
- Transposition table using Zobrist hash
- Move ordering using PV-move and MVV-LVA, killer move and history heuristics
- Late move reductions
- Quiescence search with delta-pruning
Comments
Post a Comment