Author: Robert Hyatt
Date: 05:37:46 10/02/98
Go up one level in this thread
On October 02, 1998 at 08:02:09, blass uri wrote: > >On October 01, 1998 at 17:46:47, blass uri wrote: > >> >>On October 01, 1998 at 14:15:54, Robert Hyatt wrote: >> >>>On October 01, 1998 at 11:40:46, blass uri wrote: >>> >>>> >>>>On October 01, 1998 at 08:21:43, Robert Hyatt wrote: >>>> >>>>>On October 01, 1998 at 01:11:12, blass uri wrote: >>>>> >>>>>> >>>>>> >>>>>> >>>>>>8/1r5p/6p1/8/P2k4/K6P/1R4P1/8 black to move >>>>>> >>>>>>Many programs are playing Rxb2 because they computes that white is losing its >>>>>>pawn in the pawn endgame and do not understand that after Rxb2 Kxb2 Kc4 Kc2 Kb4 >>>>>>Kd3 Kxa4 white is winning. >>>>>> >>>>>>Rxb2 is not the first move they want to play because they need time to see that >>>>>>black wins a pawn by this stupid mistake after Rxb2 Kxb2 Kc4 Ka3 >>>>>> >>>>>>Is there a program that understand by static evaluation that white is winning >>>>>>with the king at d3?(static evaluation at least 1.5 pawn advantage for white) >>>>>> >>>>>>Uri >>>>> >>>>>I get this: >>>>> >>>>> +---+---+---+---+---+---+---+---+ >>>>> 8 | | | | | | | | | >>>>> +---+---+---+---+---+---+---+---+ >>>>> 7 | | *R| | | | | | *P| >>>>> +---+---+---+---+---+---+---+---+ >>>>> 6 | | | | | | | *P| | >>>>> +---+---+---+---+---+---+---+---+ >>>>> 5 | | | | | | | | | >>>>> +---+---+---+---+---+---+---+---+ >>>>> 4 | P | | | *K| | | | | >>>>> +---+---+---+---+---+---+---+---+ >>>>> 3 | K | | | | | | | P | >>>>> +---+---+---+---+---+---+---+---+ >>>>> 2 | | R | | | | | P | | >>>>> +---+---+---+---+---+---+---+---+ >>>>> 1 | | | | | | | | | >>>>> +---+---+---+---+---+---+---+---+ >>>>> a b c d e f g h >>>>> >>>>>Black(1): sc >>>>>end-game phase >>>>> clearing hash tables >>>>>note: scores are for the white side >>>>>material evaluation................. 1.00 >>>>>development......................... 0.00 >>>>>pawn evaluation..................... 0.12 >>>>>passed pawn evaluation.............. 0.24 >>>>>passed pawn race evaluation......... 0.00 >>>>>interactive piece evaluation........ -0.23 >>>>>total evaluation.................... 1.13 >>>> >>>>I did not ask about the static evaluation of the originial position but about >>>>the static evaluation of the position after the line Rxb2 Kxb2 Kc4 Kc2 Kb4 kd3 >>>>kxa4 >>>> >>>> >>>>> >>>>> >>>>> >>>>>when I do a search, I never get "rxb2" as reasonable: >>>> >>>>programs get Rxb2 as reasonable because of the following line >>>>Rxb2 Kxb2 Kc4 Ka3 g5 g3 h5 g4 h4 Kb2 Kb4 kc2 kxa4 >>>>they need time to find this line(more than 1 or 2 seconds). >>>>white also can win this pawn ending but programs do not understand this. >>>> >>>>Uri >>> >>> >>>first, I let crafty search for > 6 minutes. *no* Rxb2. >> >>I tried the position with crafty15.20 without tablebases(48M hash tables and 6M >>hashpawn tables) and it found Rxb2 after more than 12 minutes at depth 12 with >>evaluation -0.39 >>I used pentium200MMX > >I "cheated" with the time when I changed the hash tables during the search >it is more than 30 minutes with evaluation of -0.54 pawn >whan I changed the hash tables in the beginning of the search. > >The question is if crafty with tablebases do not find Rxb2 at depth 12 >I suspect that more than 6 minutes is not enough to finish depth 12 and you need >a little more time. > >I do not see how crafty can use the tablebases from the original position at >depth 12 > >Uri on my machine, it goes beyond 15 plies here in 6 minutes. But the tablebases prevent Rxb2 if it searched to 50 plies. It simply won't play a move that gets it mated, if there is another move that doesn't...
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.