Author: David Rasmussen
Date: 03:49:33 02/09/01
Go up one level in this thread
On February 07, 2001 at 11:31:12, Robert Hyatt wrote: >On February 07, 2001 at 10:59:31, Pat King wrote: > >>I have seen it written here that with 64 bit Zobrist hashing, the perfect key >>should change 32 bits. When I had what I thought to be hashing problems, I >>captured some stats on my hash keys. I found that most of them changed 28-36 >>bits (within 4) with a few outliers as far as 13 bits from "perfection". I also >>checked that I was not generating duplicate keys. How good or bad is this? >>Should I work on the average, or the outliers? Any comments appreciated :) >> >>Pat > >The main issue is hamming distance between any two positions you search. >If each move changes 10 bits, then after 6 moves, you have potentially >changed 60. After 12 you _could_ be back to where you started. The place >to start working is on your random numbers. When I first did mine, I simply >checked the hamming distance between any two of the numbers and if it was >unacceptably low (say < 16 bits different) I culled one of them. I doubt >you can do really bad random numbers unless you make the classic mistake of >using two 32-bit floating point numbers and sticking them together to make >one 64 bit random number. The problem with this is that the 'exponent' part >of each number will be close to the same since FP random number generators >usually produce a number N such that 0 <= N < 1.0 and that will mean your >64 bit numbers are really maybe on 44 bits of significant bits. The best 64-bit linear code, with a dimension of 10 (for 1024 codewords which is enough for the appx. 800 codewords we need), is an extended BCH code with hamming distance 28. Now as for nonlinear codes, I don't know. It would be nice to find the 64-bit non-linear code with highest hamming distance. This would be the best that could be done.
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.