Author: Robert Hyatt
Date: 02:21:42 08/03/98
Go up one level in this thread
On August 02, 1998 at 23:55:02, Ed Schröder wrote: >>I've always done this. The problem is that on some occasions, the null- >>move search can cause a "false-fail-high". The re-search produces a score >>that was worse than the old search, or, more commonly, actually fails low. > >>I reject such fail highs if the re-search is worse than the old best (I >>catch this by searching with (old-best, +infinity), so that if I fail low, >>I know it is worse than the old best move and it gets rejected outright. > >>I've had versions where this was "broken" for various reasons, and they >>always produced an occasional bizarre-looking move that could be an outright >>blunder, or just a slightly weaker move... > >Hi Bob, > >Glad to hear a "false-fail-high" is common practise -:) > >Like to add that I often have noticed such a "false-fail-high" is rewarded >in the next iteration after all. It has puzzled me for years. A few years >ago I made a test-version that always took a "fail-high" as best move, >false or not. Results were not better but also not worse. > >Similar experiences? > >- Ed - I believe you said you don't use null-move if I recall. If you don't, you won't see the *ugly* problems I have seen... The reason I do this as I do now is that I saw games where crafty played *terrible* moves here and there. When I looked at the output, I saw a fail high but no score. One was against Bruce, in a probably winning position, where Crafty played a move after a fail high, and then when "pondering" it failed low, by over a piece. I traced it to a fail high caused by a null-move search and hash table interaction. I didn't have this problem in Cray Blitz so far as I know, but I only used null move R=1, with no recursion (one null move in any path). Crafty clearly plays better with this kludge in place. I have encountered positions (pathological cases for sure) where *every* move would fail high then fail low on the research. You should try playing a fail high move *there*. :)
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.