Author: blass uri
Date: 08:36:33 11/10/99
Go up one level in this thread
On November 10, 1999 at 10:09:33, Ratko V Tomic wrote: <snipped> >Namely, for sampling to work here at all, in order to generate >a "fair sample" of positions, you would have to have a good _model_ of what a >"fair sample" here is, i.e. how much material of which kind should you put in. >For example, it's not enough to pick a uniform odds for counts for pieces I did not think to pick a uniform odds for counts for pieces. It is possible to calculate the probability of every material configuration and choose 1000 random material configuration with these probabilities and after it to choose random position from everyone of the material configuration. The best way to do it is to use an array with these many numbers. You need to have almost 50 digits in the numbers and I am not expert in using long numbers in programs but it is possible to do it. If you have enough memory it is even possible to do it relatively fast(you can use an array of sum of probabilities and for every random numbernumber to find quickly the 2 numbers it is between them,you need only something like log(100,000,000) steps for every random number). 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.