Author: J. Wesley Cleveland
Date: 10:32:52 05/26/00
Go up one level in this thread
On May 25, 2000 at 12:31:44, Bruce Moreland wrote: >On May 24, 2000 at 19:23:16, Roberto Waldteufel wrote: > >>My program does precisely what you describe - it ponders by simply searching as >>if it were to play for the oponent. Have you noticed a significant improvement >>by focusing only on the predicted move? My reasoning in adopting this approach >>was that the predicted move would be the root PV move in our method, which takes >>the lion's share of the search time anyway, so the time investment in the other >>moves is only a modest, and some advantage is achieved whatever the oponent >>plays. Maybe I should switch to Crafty's method? > >The PV takes most of the time, but you still have to get a ply less. If you >search for 3 minutes, on position P, you'll get to D plies. If you choose a >successor of position P and search for three minutes, you'll get to >approximately D plies again. > >So what we've really got to choose between here is: > >1) Picking the best move and searching to D plies. > > - or - > >2) Searching all of them to D-1, then relying on the hash table to have anough >information that we'll get through D plies somewhat faster than we would have >when it comes time to search it. This is a clearly inferior solution if the >predicted move is made, but you have to balance this against the idea that you >might have some efficiency gain if the opponent makes a move that wouldn't have >been predicted. > how about 3) Searching the best move until you would reply instantaneously if your opponent made that move, and then switch to searching the next best move. >I haven't done any experiments but I think the first way must be better. You >predict a lot of the time, and therefore you receive the full benefit. You >throw your opponent off with an instantaneous reply in some cases. > >Weighed against this some increased efficiency getting to depth D-1. Maybe it's >better but I doubt it. > >bruce
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.