Author: Uri Blass
Date: 03:57:40 01/09/02
Go up one level in this thread
On January 09, 2002 at 05:08:29, Jouni Uski wrote: >On January 09, 2002 at 05:03:19, Terry Ripple wrote: > >>This is a mate in 98 using all 5 man Namilov bases! I used Hiarcs 7.32 but Tiger >>14.0 says mate in 97 because i believe it announces the number to mate after it >>made it's first move of the solution! >> >>N6k/7p/8/8/8/8/8/K6N w - - 0 1 >> >>For those who can't read the FEN here is the ASCII position. (White to move) >>wKa1,Na8,Nh1/bKh8,Ph7 >> >>PS. I'am interested to know what the longest distance to mate is in a 5 man >>position just for curious reasons. >> >>Best regards, >>Terry > >There are mate in 127(128?) moves positions posted here just last week. They are >PP-P of course. I remember that mark young shows that one of this position is drawn by the 50 move rules if you give programs to play against itself. Note that it could be more easy to build tablebases that are based on the 50 moves rules and use them but for some reason it seems that Nalimov was not interested in the game of chess but in another game that does not include the 50 move rule so he did not build the right tablebases. Building the right tablebases will take time and for practical purpose it may be better to use the exisiting tablebases to get an estimate for probability to win instead of building the right tablebases. It may be intersting to get probability for a draw by the 50 move rule based on the distance to mate. programmers may take 1000 random positions of mate in x for x=60,61,62,...120 and give Crafty with the full 5 piece tablebases to play these positions against itself in order to find the number of draws. It may be possible after doing it to translate the distance to mate into a probability to win that is going to be translated to advantage in pawns. program may translate evaluation of mate in 100 to a lower evaluation based on their estimate of the probability to win and stop the search at tablebases positions with static score that is not distance to mate when the distance to mate is too big. If the remaining depth is big enough they can get even a better estimate by playing against themselves and translate the evaluation to something that is even closer to a draw if the game is drawn. 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.