Author: Robert Hyatt
Date: 21:19:40 09/10/99
Go up one level in this thread
On September 10, 1999 at 18:47:49, William Bryant wrote: >On September 10, 1999 at 00:19:37, Robert Hyatt wrote: > >>Here is an interesting position given to me by Steffen Jakob: >> >> /p/P5p/7p/7P/4kpK/// w >> > >I tried this on Screamer (G3 266) and got the PV below. > >Welcome to Screamer (Beta) 1.00 Build 36 >Current Hash Table Size: 8192 K bytes >Debug Control On >PVS Search >NULL move Search > > >New Game ... >Move 1 W>> >The Current Search Time: 300 seconds > >New Board Position >8/p7/P5p1/7p/7P/4kpK1/8/8 w - - 0 0 >Move 1 W>> >PLY NODES TIME SCORE PV >____________________________________________ > 5 665 0:00 -10.95 Kh2 f2 Kg2 Ke2 Kh3 f1=Q+ Kg3 > 6 1185 0:00 -10.95 Kh2 f2 Kg2 Ke2 Kh3 f1=Q+ Kg3 > 7 1904 0:00 -10.95 Kh2 f2 Kg2 Ke2 Kh2 f1=Q Kg3 > 8 4177 0:00 -12.39 Kh2 f2 Kg2 Ke2 Kh2 Kf3 Kh3 f1=Q+ Kh2 Qxa6 > 9 9010 0:00 -13.27 Kh2 f2 Kg2 Ke2 Kh3 f1=Q+ Kh2 Qf4+ Kh3 Qg4+ Kh2 >Qxh4+ Kg2 > 10 22310 0:00 -13.57 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 Qf4+ > 11 107384 0:01 -13.57 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 Qf4+ > 12 359601 0:04 -99.88 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 g5 hxg5 Kf3 g6 Qg2+ > 13 988159 0:11 -99.88 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 g5 hxg5 Kf3 g6 Qg2+ > 14 2764683 0:31 -99.88 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 g5 hxg5 Kf3 g6 Qg2++ > 15 7232763 1:21 -99.88 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 g5 hxg5 Kf3 g6 Qg2++ > ----- Search Stopped ----- > 15 8745125 1:30 -99.88 Kh2 f2 Kg2 Ke2 Kg3 f1=Q Kh2 g5 hxg5 Kf3 g6 Qg2++ > Total Search Time: 1:30 > Move: Kh2 > > >I don't find mates shorter than 6 moves (12 plies). > >If I understand you solution correctly, if you fail high (regular search or >null move search) with a score that is creater than the Mate score limit, >you adjust that score to the mate score limit for fail high stores in the >hashtable. Also, don't store a best move in this case. > >ie in sort-of-pseudocode > > if (Score>alpha) { > if (Score>beta) { > long StoreScore = (Score>MateScoreLimit)?MateScoreLimit:Score; > HashStore(CurHashSig, NoMove, StoreScore, LOWER): > } > } > >William >wbryant@ix.netcom.com That seems to be backward. If score > beta, the _current move_ you just searched failed high. That move should be stored in the hash table to try it first the next time you search this position. The problem we are discussing is the _other_ case.. the one where you search all movs and still end up with best == alpha, which means all moves were bad...
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.