Author: Dann Corbit
Date: 11:28:41 10/28/02
Go up one level in this thread
On October 28, 2002 at 01:54:35, Ricardo Gibert wrote: >On October 28, 2002 at 00:46:03, Dann Corbit wrote: > >>On October 27, 2002 at 01:37:13, Ricardo Gibert wrote: >> >>>On October 26, 2002 at 22:37:59, Russell Reagan wrote: >>> >>>>On October 26, 2002 at 07:41:41, Ricardo Gibert wrote: >>>> >>>>>No collision errors if the 64 bits are stored in the table. >>>> >>>>That *might* be true if you had a perfect hash key generation algorithm, where >>>>each pawn formation mapped to exactly 1 hash key, but that is not the case. >>>> >>>>Russell >>> >>> >>>I'm wasn't talking about collisions in general. I'm talking about collision >>>*errors* where 1 pawn structure is mistaken for another. When you stated, "It >>>makes me wonder if there are ever *errors* caused by this" is what I addressing. >> >>Then store the 64 bit formation described and a collision is impossible. > >Look at the bottom of http://www.talkchess.com/forums/1/message.html?261847 Well, obviously, you can calculate statistics for collisions easily with this information, since you understand it. It might be very useful to calculate the optimal hash size for a given set of search parameters.
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.