Author: Miguel A. Ballicora
Date: 17:07:27 01/09/02
Go up one level in this thread
On January 09, 2002 at 19:43:54, Russell Reagan wrote: >On January 09, 2002 at 17:40:39, Miguel A. Ballicora wrote: > >>On January 09, 2002 at 17:11:36, Russell Reagan wrote: >> >>>On January 09, 2002 at 12:50:12, David Hanley wrote: >>> >>>>I have seen it claimed somewhere that with perfect move ordering, an eight ply >>>>search would only consume a thousand nodes or so, even only alphabeta ( no >>>>hashing or forward pruning ). >>>> >>>>Is this so? >>>> >>>>dave >>> >>>Let's think about this for a moment. If you knew for a fact that your program >>>had perfect move ordering, then you could always assume that the move at the top >>>of your list was correct. Therefore you wouldn't even have to search, but if you >>>did, an 8 ply search would take a handful of milliseconds and consume a whopping >>>8 nodes (depending on how you count your nodes). You wouldn't even have to use >>>alpha-beta (or any other kind of search) if you _KNEW_ that your move ordering >>>was perfect. You could always choose the move at the top of the list, and your >>>program would play perfect chess. Maybe this isn't what you were thinking when >>>you said "perfect move ordering". >>> >>>Russell >> >>It was a theoretical question. >> >>Miguel > >In "theory" the game would play perfect chess without any search. I answered the >question from a theoretical standpoint. Everyone else answered it from a >practical viewpoint. No, because it was a theoretical question with a potential practical use. It was intended to know how many nodes you would have to search even though you know already that the first you pick is the correct one. It ask how long it is be the trip to prove it, not which move is the correct since you already know it. If you know how many nodes you will have to traverse a perfect ordering tree you will know a "bound". That will tell you right away if it is worth it or not to pursue more work on ordering. Am I searching 10 times more than the optimal or just 30% more? An engine with a perfect ordering will not play perfect chess. I will play the best chess you could possibly play with the evaluation function of that engine. Regards, Miguel
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.