Author: KarinsDad
Date: 14:45:05 01/20/99
Go up one level in this thread
On January 20, 1999 at 16:37:41, Bruce Moreland wrote: > >On January 20, 1999 at 12:01:14, KarinsDad wrote: > >>On January 20, 1999 at 10:08:15, Robert Hyatt wrote: >> >>>On January 19, 1999 at 20:20:48, Larry Griffiths wrote: >>> >>>>I ran my program with Alpha-Beta and without Alpha-beta. >>>>The scores and moves chosen were different. >>>> >>>>Is this normal for Alpha-Beta or is this an indication of an >>>>error in my implementation of Alpha-Beta pruning? >>>> >>> >>>It's a bug. Alpha/beta _must_ produce the exact same result as minimax, >>>this was proven in "An analysis of Alpha/Beta Pruning" written by Knuth and >>>Moore, 1975.... >> >>Actually, it is not necessarily true that there is a bug in the code. There is >>an assumption on that. Since the min max search should have been exhaustive to a >>certain depth, you cannot run the program to a deeper depth with the alpha beta >>or the scores could change which in turn could result in different moves as >>well. >> >>If the alpha beta was able to search deeper than the min max, then there is a >>bug in the testing as opposed to a bug in the code. > >I think the assumption is that he's running to the same depth, otherwise of >course it won't return the same score and PV, otherwise we'd all search to depth >= 1 and quit. > >bruce It's the little things that sometimes trip us up. Before Larry started changing a lot of code, it seemed prudent to remind him of the obvious. KarinsDad
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.