flok wrote: ↑
Sat Oct 26, 2019 9:22 pm
Ferdy wrote: ↑
Fri Oct 25, 2019 11:57 pm
client.py is working.
Is your program Atakador?
Plays very well! (ignore the neural nets because they always win)
Tried viewing some games of ataxxzero, below is interesting, ataxxzero as black (white dots) just played a7. It does not try to gain material but seemed trying to allow its opponent to expand in the center.
FEN: 1oxx3/oxxx3/oxx4/7/7/1x3o1/x5o o 0 11
Here is Atakador's analysis, its a python engine. Just curious what it thinks.
C:\ataxx>py -3 atakador.py
info string This is Atakador an engine that can play standard Ataxx game.
info string Atakador is a Filipino word meaning likes to attack.
info string Credits: Python-ataxx modules
id name Atakador v0.11.5
id author fsm
option name UCI_Opponent type None 1000 computer Atakador v0.11.5
option name NullMovePruning type check default true
Code: Select all
position fen 1oxx3/oxxx3/oxx4/7/7/1x3o1/x5o o 0 11
a b c d e f g
7║ ║O║X║X║ ║ ║ ║7
6║O║X║X║X║ ║ ║ ║6
5║O║X║X║ ║ ║ ║ ║5
4║ ║ ║ ║ ║ ║ ║ ║4
3║ ║ ║ ║ ║ ║ ║ ║3
2║ ║X║ ║ ║ ║O║ ║2
1║X║ ║ ║ ║ ║ ║O║1
a b c d e f g
1oxx3/oxxx3/oxx4/7/7/1x3o1/x5o o 0 11
go movetime 60000
info depth 1 score cp 10 time 497 nps 66 pv b7d5
info depth 2 score cp -10 time 3820 nps 63 pv b7d5 b7
info depth 3 score cp -110 time 27039 nps 65 pv b7d5 b5d4 a5c4
info score cp -110 time 27039
How deep are these other engines are searching?
It is fun implementing different ideas even with this slow searcher
nullmove is ok
lmr is ok
hashtable just even (remove it for a moment)
qsearch is ok
move ordering with static board evaluation is ok
move ordering from pv is ok
Will release it once it is stable and other improvement ideas are tried.
Found a nice article on ataxx at http://www.linuxonly.nl/docs/4/0_Ataxx.html#page96
Perhaps it is better if you add this link on your website, something like programming resources section.