Author: Russell Reagan
Date: 12:29:19 01/02/03
Go up one level in this thread
On January 02, 2003 at 06:57:15, Steve Maughan wrote: >1) Try a situaltion where no castling is allowed e.g. move king to d1 and d8. >See if the problem still exists. This will tell you if it is a castling >problem. I did try the same position with the kings on the d-file as you said, and it calculated perft correctly for 5 or 6 ply, so yes, I'm thinking it has something to do with castling. >2) Are you doing repetition detection in the perft routine? I'm not sure what >the convention is - is it the norm? Clearly the 4th ply would be when it comes >to light. I didn't think of this. No, I'm not doing repetition detection yet. I am getting a lot of incorrect perft(4) data, but so far all errors seem to be less than the correct values, which would lead me to believe that the current problems are not with repetition detection (since the correct engine would have less nodes due to NOT continuing after the repetition, where mine would get to the repetition and just keep on crunching). Also, is 4 ply where repetition would become a factor? I can't see how you could get 3 repetitions in 4 ply. It seems like you would need 8 ply to detect a full 3-fold repetition. Maybe you are talking about how some engines assume it is a draw after 2 repetitions? >3) Are you doing castling through check i.e. Black rook on d8 threaten d1 and >you castle - probably not the problem since you're not finding enough positions. I'm checking to make sure that (for example, in queenside castling for white) the squares b1, c1, and d1 are not occupied, that castling rights for the particular side exist, and that the squares c1, d1, and e1 are not attacked by the enemy. >Hoep you find the problem - let us know what it is! Me too :) Russell
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.