Author: Uri Blass
Date: 04:45:27 01/18/04
Go up one level in this thread
On January 18, 2004 at 07:31:08, martin fierz wrote: >On January 18, 2004 at 06:42:50, Uri Blass wrote: > >>On January 18, 2004 at 05:29:59, Kurt Utzinger wrote: >> >>>On January 18, 2004 at 04:44:59, martin fierz wrote: >>> >>>>On January 18, 2004 at 02:27:35, Kurt Utzinger wrote: >>>> >>>>>On January 18, 2004 at 01:25:52, Kurt Utzinger wrote: >>>>> >>>>>>[D]8/8/5pk1/5b2/7K/3R3p/7P/8 w - - 0 1 >>>>> >>>>>How on earth should White win this position? I see no way >>>>>to do this. Some reasons for my assumption: >>>>> >>>>>1.) >>>>>If White at some time takes the h3-pawn with the rook, >>>>>the pawn ending is a draw. >>>>> >>>>>2) >>>>>The only try would be to win Black's f6-pawn. But >>>>>this can't be done. >>>>> >>>>>3) >>>>>But even in the case that due to some Zugzwang >>>>>White wins the f6-pawn, the position still looks >>>>>drawn. >>>>> >>>>>Any counter arguments. >>>>>Kurt >>>> >>>>hi kurt, uri, >>>> >>>>this is an interesting position. the first thing to note are a couple of >>>>theroetically known endgames: >>>> >>>>-> if white succeeds in winning both black pawns, he wins >>>[D]7k/R7/7P/6K1/8/3b4/8/8 w - - 0 1 >>>> >>>>-> however, if he is careless then, he only draws >>>>[D]7k/R7/7P/6K1/8/3b4/8/8 w - - 0 1 >>>> >>>>these two endings are just interesting so i posted them anyway. but they are >>>>irrelevant. the relevant ending is if white wins the f-pawn: >>>> >>>[D]8/8/6k1/5b2/7K/3R3p/7P/8 w - - 0 1 >>>>this is supposed to be a win in endgame theory AFAIK, and that is the >>>>counter-argument to kurt's argument #3 - if white succeeds in winning f6, he >>>>will in fact win the game. in addition, the wins of this type of endgame usually >>>>are very very deep, so even searching this position to 30 or 40 ply might not be >>>>enough to see the win of the h-pawn. >>>> >>>>if somebody has the KRPKBP tablebase and time to spend he could run the original >>>>position over night and se what happens... >>>> >>>>cheers >>>> martin >>> >>> Hi Martin >>> Indeed a very interesting position in which the main >>> question remains: can White succeed in winning Black's >>> f6-pawn? >>> Kurt >> >>It seems that white can do it(or to win the h3 pawn) >>Otherwise I see no explanation for Yace increase in the score. >> >>Uri > >uri, wouldn't you be seeing a huge jump in the score if it really saw the win of >a pawn? but you only see +0.02 or so on next iterations. i let shredder 8 >(unfortunately without TBs) run overnight on the starting position, it reached >depth 38, and the score on d30 was +1.50 and at d38 +1.52 (or something like >that). i would guess that it sees it can get a tiny bit better piece placement >which explains +.02 difference, but nothing more. i added the 4-man TBs now, and >am trying a shootout at 1min/move, and will repeat the deep search later - but i >really don't think your interpretation of yace's score is right here - why >shouldn't it evaluate this position as +1.xx? it's clear that only white has >winning chances... >i also would guess that this position is a draw - but who knows? > >cheers > martin The point is that with the same material Yace always evaluates the position near 0.8 pawns for white even when I follow Peter's line. I need explanation for the significantly bigger score of Yace not by 0.02. I prefered to use yace and not other programs because I am afraid that other programs may never see zugzwang because of null move pruning. I am not sure at this point if white is winning. There are lines that white can win the pawn but it is not clear if black is forced to go to one of them. Uri
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.