Author: Robert Hyatt
Date: 09:46:58 04/01/00
Go up one level in this thread
On April 01, 2000 at 10:32:52, blass uri wrote: >On March 31, 2000 at 23:17:45, Robert Hyatt wrote: > >>On March 31, 2000 at 12:53:43, Stephen Ham wrote: >> >>>Dear Readers, >>> >>>I know many of you are actively following my ongoing match games versus Fritz 6a >>>and Nimzo 7.32. For those of you not familiar with the event, please visit: >>> >>>http://correspondencechess.com/campbell/index.htm >>> >>>Anyway, a frequent poster here (name is withheld) wrote to Mr. Campbell stating >>>that since the chess engines are displaying their top 3 choices, they are being >>>weakened "a lot". No explanation was given for that claim. >>> >>>Would somebody here please provide a detailed explanation regarding whether this >>>claim is correct and why? >> >> >>It depends on how they compute these variations. Done correctly, it is >>_horribly_ inefficient. If you watch a normal search, the first move will >>usually take over 50% of the total time. The remaining N-1 moves take the >>remaining 50% of the time. If you have it display two 'best'moves, you >>increase the total search time by roughly 50%. The first move takes the >>same time as before. The second move also takes the same time as before, >>and the final N-2 moves take just a tad less than before. Net loss is >>ugly. If you have it display the best 3 moves, you slow it down by exactly >>a factor of two... >> >> >> >> >>> >>>The chess engines are on settings recommended by ChessBase USA as their optimal >>>settings for this event. My extremly limited understanding is that displaying >>>the top 3 choices does indeed affect the chess engines, but it causes them to >>>spend more time on what it believes to be the 3 best moves. As such, this sounds >>>like an enhancement to me. Given that the chess engines are allowed about 24 >>>hours calculation time on weekdays and are searching to 16-18 ply, I can't >>>imagine that this weakens them in any way. >> >> >>Each iteration will take about 2x longer than the previous. Rather than >>a branching factor of 3x, you raise it to 6x. This will cost several plies >>over 24 hours. > >My experience with chessbase engines show that it is not the case. >I also see no reason that the branching factor will change. > >The program is going to be slower by a constant factor by calculating the first >3 moves. >The only difference relative to calculating only the best move is that you need >to use time to calculate the second best move and the third best moveand I do >not see a reason to be more than 3 times slower. > >The only case when the branching factor is going to be bigger is if you generate >a tree of moves and not only the best 3 moves. > >Uri The term "branching factor" has two uses: (a) the correct one is that it is the number of branches at the 'typical' node in a tree. No one uses it like this any longer although it is the correct usage. (b) the typical one is that it is the ratio of the time for T(n)/T(n-1) (time for iteration n divided by the time for iteration N-1). I inaccurately used a combination... clearly iteration N will take 2x (or more) longer than normal. But then so will iteration n-1. But in any case, a factor of 2-3 is a ply. And every ply helps...
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.