Author: Koundinya Veluri
Date: 20:32:14 11/02/03
Go up one level in this thread
On November 02, 2003 at 14:58:24, Dieter Buerssner wrote: >I think, it is useful. Specially when you are in a debugging situation, it is >nice to be able to reconstruct the conditions. Different initial numbers for >Zorbrist hashing will make it more difficult. For a fully debugged stable engine >(if it exists) it might be no problem at all, to use different numbers in >different runs. I mentioned, that one might want to use hash-keys for book >access. But one could do it in a different manner, too. I use random zobrist keys every time the program starts, but the keys I used to create the book are written to the book so there is no problem. I tried once to find a good set of zobrist keys. I found a good set, but as soon as I modified the program even slightly, I had to find another set. This was too much of a hastle so I just decided to use random numbers as I was doing before. When I debug though sometimes I set a constant seed so that the output is exactly reproducable. Koundinya
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.