Author: Peter McKenzie
Date: 12:07:39 11/08/98
Go up one level in this thread
On November 08, 1998 at 11:06:50, Robert Hyatt wrote: >On November 08, 1998 at 10:22:26, Heiko Mikala wrote: > >>Hi everybody! >> >>A few days ago I started to let my program play at FICS. As most of you >>may know, most of the games at FICS are played using Fischer Time >>Control, like for example 2 12 (meaning 2 minutes for the whole game >>plus 12 seconds after every move). >> >>Consequently I had to implement support for this time control, but in the >>first step only made a quick hack: I use normal Blitz time-usage for every >>move, then add the x seconds to the remaining time and use normal >>Blitz time-usage again for the next move. Obviously this is sub-optimal. >> >>It should be easy to find a better solution, but as I know that many other >>programmers let their programs play at the servers, my question is: How >>did others implement Fischer-Time Control in their programs? Any good >>ideas/tricks? >> >>Thank you all, >> >>Heiko. > > > >I use target=INC + REMAINING/XXX > >XXX is currently 25 I think, but you may want to experiment to find a setting >you like. I've been as high as 30, as low as 20. Note that 25 is very good, >*if* you are thinking on the opponent's time... because lots of times will find >you moving very quickly due to a correct prediction. This formula will make you >use more time up front where it is important, letting you save time by pondering >and moving it to the end... I do the same thing, my XXX is 30 and my program isn't pondering. Seems to work pretty well. I do one other thing: if the total time remaining is less than 15 seconds I subtract 1 from the thinking target time. This is a bit of paranoia to prevent it from getting too close to losing on time. Peter
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.