Author: Uri Blass
Date: 15:03:55 08/31/00
Go up one level in this thread
On August 31, 2000 at 17:31:32, Dan Andersson wrote: >A simple metodology. The simple statistical model hurts my mathematical >instincts, as well as my knowledge. 'Keep it as simple as possible but no >simpler.' The model is simple. The only problem is to construct the set of psuado legal positions not to be so big. My counting program calculated the number of possible positions for every possible material configuration and calculated the number of possible pseudo legal positions in every structure. Choosing a random pseudo legal position if you want equal probability for all the psuado-legal positions should be done by the following steps. 1)Choosing a random number of pseudo legal position. 2)Calculating the material configuration. 3)Calculating the exact position in the material configuration based on the number. It is possible that only considering material configuration is not enough and we need to reduce the size of the set of the pseudo legal positions. I have ideas how to do it(considering also some knowledge about the pawn structure and not only knowledge about the material condiguration) but in this case calculating the number of pseudo legal positions can be a more complicated task for the computer and it may need more time to do it or to choose random pseudo legal position. My Mathematical instincts say that the problem is solvable and that we can get a number that will be almost the exact number of legal positions when we can be sure with 95% confidence that the error is not more than 10% if we work hard about the problem. 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.