Author: Uri Blass
Date: 21:40:11 03/31/02
Go up one level in this thread
On March 31, 2002 at 19:59:15, Jorge Pichard wrote: >On March 31, 2002 at 17:00:08, Uri Blass wrote: > >>On March 31, 2002 at 16:28:04, Jorge Pichard wrote: >> >>>Uri sent me his program movei00_6 and I started a match of possible 10 games at >>>45 Minutes per sides. Since movei00_6 doesn't have a Book Opening, I decided to >>>make the match even and play Chess Tiger 14.9 beta2 without its Book, but I also >>>forced the first move for CT 14.9 1.h3.... You will observed one of the best >>>game ever at this Low level probably both programs are rated closed to 2125. I >>>love these two moves from CT 14.9 70.Rxe6 and 76. Nxb5! >> >>Thanks for the game >> >>2 notes: >> >>1)movei has no learning function so you need to give tiger a different opening >>move in the rest of the games in order not to see the same game twice(for >>example 1.a3) >> >>2)I tested movei mainly in x minutes/y moves so it's time management is not >>optimal for x minutes/game(it is possible to improve it's time management in all >>time controls but it is possible to improve it more in x minutes/game). >> >>When the time control is x minutes/game movei simply asssumes every move that >>the time control is x minutes for the next 30 moves and I know that it is a >>mistake but the previous tournaments that it played were x minutes/y moves and >>not x minutes/game. > > >Sorry Uri but I was testing your program with the white pieces playing 1.h3 to >force the first move, I just read your message and the next two games tomorrow >I would NOT play the 1.h3 again probably 1.a3 and you can sugest any other >opening, but I did disable CT 14.9 Opening Book. If you want I just let CT 14.9 >and movei select their first opening move from now on. Plus in the next game I >will also play X minutes/Y moves instead of X minutes/games. I am using Arena >0.78 can you sugest a good time control between 45 or 90 minutes/Y moves. The >ending was well played by CT 14.9 and movei opened its king side too much with >the Clemenz Opening. If you want to use similiar time control to 2700+2 then you can use 30 minutes/40 moves. You can of course use 45 minutes/40 moves but in this case I guess that the fact that the time control is slower than 2700+2 is going to help palm tiger. My tests against faile on p800 were without opening book but the games did not repeat twice becuase I used different time control I used 9 short games of 1-5 minute/40 moves(white and black in 4 of them) and too long games of 120 minutes/40 moves and 180 minutes/40 moves(my program was white in both of these games). I told faile no information about hash tables but faile use hash tables by the default parameters so I do not think that it had a big influence on the results. Total result: My program beated faile 4-3 with white and 2.5-1.5 with black. My program lost the 120 minutes/40 moves but won the 180 minutes/40 moves(faile lost the 180/40 on time but my program continued to play the game under winboard and won also thanks to checkmate) In the game against palm tiger one of the main problems of my program is the blunder of going to a lost pawn endgame I suspect that it may have chances for a draw without that blunder. I am not an expert of queen endgames but I believe that at least in comp-comp games there are chances for a draw without Qf5. [D]k7/5Qp1/6P1/pp2q3/6K1/8/P7/8 w - - 0 55 Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.