Author: Pat King
Date: 12:32:08 04/16/04
Go up one level in this thread
On April 14, 2004 at 17:43:26, Joachim Rang wrote: >On April 14, 2004 at 15:36:24, Pat King wrote: > >>I track what I call the 'practical depth limit' or PDL. If I run out of time, I >>make the current best move and decrement the PDL. If I reach the PDL, I make the >>current best move and increment the PDL. This doesn't explicitly speed up the >>selection of obvious moves, but it seems to work out that way. >> > >I do not understand. For example your are currently on depth 14 and run out of >time. You make the move and decrease PDL to 13. In your next move you will make >your move after depth 13 and increase PDL to 14 and so on? >Why should that beneficial? > >regards Joachim > You understand correctly, that's how it works. I did this when I was using MTD, because with MTD, you can't trust a PV based on an incomplete search at a given depth, and had to revert to the shallower search result upon running out of time. With my scheme, I only ended up wasting search time on an incomplete MTD search every other move. I kept this when I switched to PVS because I liked the way the engine played. It seemed more "human" to me in its use of time. Again, FWIW Pat >>>Thanks, >>>Eric Oldre (new chess programmer) >> >>FWIW >> >>Pat King (perpetual newbie chess programmer)
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.