Author: Oliver Roese
Date: 17:35:17 05/24/00
Go up one level in this thread
On May 24, 2000 at 18:05:24, Robert Hyatt wrote: >On May 24, 2000 at 10:28:57, Oliver Roese wrote: > >>Hi all! >> >>This question is about pondering during the opponents time... >>Crafty does the following: >>It predicts the oppononts move, assuming "optimal" play and then starts to >>work until the opponent moves. >>If it predicts the opponents move correctly it has a great edge, otherwise >>only some hashtableentries. >>If it wouldnt predict the opponents move it would gain a small contribution to >>_every_ move. >>Obviously the better it predicts the opponents move, the better is the first >>method. > >Point 1. Against strong opposition, it correctly predicts about 50% of >the time, roughly. Which is not terribly surprising... > > >>From my experience as a mere chessplayer i would say the following: >>-Predicting the opponents move is very difficult even in games of the >>highest value (disregarding trivial cases and extraordinary circumstances). >>-Intuitively i would judge a small contribution to every move as more >>worthfully than an extremly big one that occurs seldomly >>To say it exaggerated: If you have 20 moves to made and distribute >>your resources evenly, you may have a chance. If you invest all in the first >>move, making the other 19 moves very bad, you are dead for sure. >>In more general terms: >>The relative benefit of predicted moves decreases rapidly with increasing >>searchdepth, i think. >> >>Maybe one could use a hybrid approach? >>What is the reason to having this in crafty? > >Easy. At present, about 1/2 of the time it correctly predicts the opponent's >move, and can make a move using little of its own time. saving about 1/2 of >the total time. How would you improve on that? If you pick the best 4 moves, >and searched them equally (during pondering) then after your opponent moves, >you have spent 1/4 of the normal time on the move he played. You save 25%. >25% is < 50%. If you pick the best 2, you could save 50% total. Which is >what it is already saving. > > > >>Thanks in advance for any input and giving me some of your time. >> >>Oliver Roese Thanks for all input! I will shortly recapitulate: Let me call p(i) the probability with which on catches the opponents move if one searches i "good" moves. (i=1,2,3,...) We can then expect to make use of p(i)/i time from the opponent in the mean. It is better to search two "good" moves, if p(1) < p(2)/2. p(1) = 1/2 is obviously unbeatable. I never expected such a high hitrate. But what if this chance drops down??? This _may_ be the case on extremly short time controls. Or in positions, which are generally not very good understood by chessmachines (i.e. closed positions). Another objection is, that it is in general better to have a less volatile payoff structure, which favors to search more moves (as i already mentionted). How to devise i good moves is of course a problem for itself. Oliver Roese
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.