Luna is an alpha-beta pruning based chess engine that uses traditional evaluation methods.
It currently supports protocol, so it is possible to embed it into a GUI.
It currently supports protocol, so it is possible to embed it into a GUI.
Rating CEDR=2201
v.1.1:
Release Notes
Because of an oversight, I ended up disabling null move pruning from version 1.0, which causes that version to be weaker than it should be. Version 1.1 simply re-enables this search feature. Besides that, the missing priorities.json file from the last release sources was now added.
This release is about 13 Elo points stronger than version 1.0 in STC (LTC not measured yet).
Comparison with previous version
For now, only short time control tests were ran against version 1.0.
Test configuration
Time control: 8+0.08
Opening book: 8moves_v3.pgn
Test results
Score of Luna 1.1 vs Luna 1.0: 3690 - 3219 - 5721 [0.519] 12630
... Luna 1.1 playing White: 2009 - 1461 - 2846 [0.543] 6316
... Luna 1.1 playing Black: 1681 - 1758 - 2875 [0.494] 6314
... White vs Black: 3767 - 3142 - 5721 [0.525] 12630
Elo difference: 13.0 +/- 4.5, LOS: 100.0 %, DrawRatio: 45.3 %
Changelog
Re-enabled null move pruning
Comments
Post a Comment