Author: Robert Hyatt
Date: 06:56:22 08/28/00
Go up one level in this thread
On August 28, 2000 at 08:29:05, José Carlos wrote: >On August 27, 2000 at 21:52:05, Robert Hyatt wrote: > >>On August 27, 2000 at 17:33:15, Tom King wrote: >> >>>Hi all, >>> >>>a question for programmers on fail highs. >>> >>>what do you do in your program if a fail high is encountered, which on the >>>research fails low? >> >> >>Two answers. First I use PVS, so keep this in that context: >> >>1. if the null-window search fails high, but the re-search fails low, I >>ignore it totally. >> >>2. If the null-window search fails high, _and_ the re-search (with the >>aspiration window) fails high, then I keep the move as best, even if the >>re-search with beta,+infinity fails low... >> >> >> >>> >>>I've ignored this issue, because it doesn't seem to happen all that often (in my >>>program). So if my program finds a move which fails high, even if the research >>>indicates that it maybe shouldn't have failed high, it thinks the move is good. >>>Maybe this is bad? At the WMCCC recently, I noticed a couple of these fail high/ >>>fail low moves cropping up at critical, complex positions. Often I was unhappy >>>with the move my program chose in these cases. Perhaps these fail high/ fail low >>>moves need to be treated with suspicion? >> >> >> >>Perhaps. I have found that the null-window fail high can't be >>trusted, but since that is verified with a subsequent alpha,beta >>re-search, it might fail low and get ignored, or fail high and get >>re-searched a third time with +infinity for beta. Seems safe enough. >>But when this is happening, strange things are going on. If you turn null- >>move off, most of these fail high/fail lows go away... so that is the source >>of the problem. > > Yes, in most cases, but I don't have null-move in Averno, and experience those >fh-fl from time to time. Some people here (I think Ulrich Tuerke and some other) >explained how this can come from hashing too, and seemd very clear to me. > > José C. > It can definitely happen without null-move. With null-move, it happens _far_ more frequently. >>> >>>Cheers, >>>Tom
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.