Author: Robert Hyatt
Date: 13:25:46 02/02/98
Go up one level in this thread
On February 02, 1998 at 15:42:36, Frank Schneider wrote: >On February 01, 1998 at 17:36:42, Robert Hyatt wrote: > >>On February 01, 1998 at 17:32:35, Brian McKinley wrote: >> >>>I am writing my first chess program, and am at the point where I need to >>>determine how much time to spend in the search. Is there a common >>>formula? >>> >>>-Brian >> >>hard to answer question. Simple formula for normal time controls >>might be: >> >> target=(time_left-safety_fudge)/moves_remaining; >> > >In my opinion it is a good idea to search for example target*1.5 >when movew_remaining is large. For example if there are 50 moves >left it is quite likely that the game is decided earlier and you >don't need as much time for moves 40-50 than for 0-10. > >Another interesting point is that grandmasters usually spend most >of their time for the first moves after their prepared moves. >Maybe give some additional bonus shortly after out of book. > >Frank As I mentioned, the "Using time wisely" paper is worth reading. We did a lot of study on GM games, and came up with a "curve" that fit the usual GM time/move from the point he left his "book" (preparation) to the time control. We then made Cray Blitz use this formula for allocating time. It worked pretty well. We did, later on, decide to not use time so aggressively up front so that we had sufficient time to search through a fail low condition close to time control...
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.