Author: Uri Blass
Date: 15:29:49 05/28/04
I make progress in having a better book. I already have procedures and functions that do the following jobs: 1)Translating the moves of the book that movei has in his public book to an array of struct with position and move and frequency(frequency 1 in this step for every position). 2)Sorting the book based on lexisographic order of position hash key and move(the book is still in an array at that point) 3)deleting doubles from the array and increase the frequency in these cases from 1 to bigger values. 4)deleting moves with frequency that is too small if I want to do it (it is planned to be optional based on the user's decision). 5)translating the array to a binary file 6)doing a binary search for a specific zobrist key that return -1 if it cannot find it or the place to first find the position in the file. I still need to do the following tasks: 1)deciding based on the place in the file which move to choose in book(probability based on the frequency if some positions has the same learning value when initial learning value is 0). 2)increasing learning value of positions that happened in the game by winning_value[ply] in case that the program wins and reducing the learning value by losing_value[ply] in case that the program lose when both winning_value and losing_value are arrays that the user can define. 3)adding a parameter that tell the program the maximal ply to search in book so the program does not try to look at book after enough plies. 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.