New chess engine: Serendipity 0.3.1 (java)
Serendipity
Serendipity, an UCI chess engine written in Java.
Author - Xu-shawn
v.0.3.1:
Three new usability features:
Resizable Hash
WDL Output
Eval Normalization
Thanks to work from the Stockfish team, Serendipity's evaluation has now been fit to a Win-Draw-Loss probability model. As a result, not only has Serendipity gained the ability to predict the probability of winning, drawing, or losing a position under testing conditions, but Serendipity's centipawn output has also been normalized. In other words, Serendipity's evaluation output has been rescaled in such a way that 100 centipawns represent 50% expected win rate at testing conditions. This will make Serendipity's evaluation more comparable to other engines who has adopted similar normalization schemes.
To run Serendipity, install the Java Runtime Environment (version >= 17), and make sure that the JRE is under system path. Then, run the program either through directly running the exe or through the command java -jar Serendipity-0.3.1.jar.
Serendipity 0.1 results:
Sloth 1.7 | 2/2 | +2 | 2 Partii |
Tcheran 2.2 | 2/2 | +2 | 2 Partii |
Zagreus 5.0 | 2/2 | +2 | 2 Partii |
Victoire 1.3.0 | 2/2 | +2 | 2 Partii |
Cadabra 2.0.1 | 2/2 | +2 | 2 Partii |
Yakka 1.0 | 2/2 | +2 | 2 Partii |
Akimbo 1.0.0 | 1/2 | +0 | 2 Partii |
Critter 1.6a | 1/2 | +0 | 2 Partii |
Clarity 6.0.0 | 1/2 | +0 | 2 Partii |
Arasan 24.2 | 0.5/2 | -1 | 2 Partii |
Rebel 16.3 | 0.5/2 | -1 | 2 Partii |
Vafra 14.12.1 Stockfish | 0.5/2 | -1 | 2 Partii |
Peacekeeper 2.40 | 0.5/2 | -1 | 2 Partii |
Obsidian 12.0 | 0/2 | -2 | 2 Partii |
Berserk 13 | 0/2 | -2 | 2 Partii |
Clover 6.1.19 | 0/2 | -2 | 2 Partii |
Caissa 1.18 | 0/2 | -2 | 2 Partii |
Alexander 1.1 | 0/2 | -2 | 2 Partii |
Velvet 7.3.0 | 0/2 | -2 | 2 Partii |
Delocto 0.7 | 0.5/1 | +0 | 1 Partii |
Windows executable & jar file.
Comments
Post a Comment