Author: Gian-Carlo Pascutto
Date: 07:58:55 04/20/02
Go up one level in this thread
On April 20, 2002 at 10:45:59, Uri Blass wrote:
>This is wrong.
>I am sure that sjeng cannot search 601 nodes in less than 0.000001 seconds.
>
>I see that the times are always a.bc0000 and it is not logical.
>I think that it is better to write a.bc...
It just does printf("%f", solutiontime);
The timing code I am using right now does not have more than centisecond
precision.
>Note that I expect to solve it relatively fast when I implement the right
>algorithm in my program but there are other things with higher priority for me
>and I think that the null move algorithm may give it more rating points(my
>latest version use it and seems to play better but at this moment it can lose on
>time because of some bug that I know when it needs to play at the middle of the
>iteration).
>
>Note that the solution should be easy for programs with the right algorithm
>because all the moves are forced.
>
>White needs to play the move in order to prevent promotion and black has only
>one legal move in the variation.
You are right. If you handle forced variations in a smart way, you will fly
on positions like this. Unfortunately, this has very little influence on how
good your program actually plays chess. I wouldn't mind Sjeng kicking Rebel
around like this in real games :)
--
GCP
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.