[d]6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1
You may use 6-men egtb. Drawish score could be +/- 50 cp.
Can your favorite engine show a drawish score?
Moderators: hgm, Rebel, chrisw
-
- Posts: 4833
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
-
- Posts: 582
- Joined: Wed May 10, 2006 7:28 pm
- Location: Birmingham, England
Re: Can your favorite engine show a drawish score?
No, but Critter can.Can your favorite engine show a drawish score?
Marek Soszynski
-
- Posts: 4833
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
Re: Can your favorite engine show a drawish score?
Good CritterMarek Soszynski wrote:No, but Critter can.Can your favorite engine show a drawish score?
Code: Select all
22/27 -0.17 1. ... Kf8 2.Ke4 Bf6 3.Bd5 Bc3
-
- Posts: 6442
- Joined: Tue Jan 09, 2007 12:31 am
- Location: PA USA
- Full name: Louis Zulli
Re: Can your favorite engine show a drawish score?
Ferdy wrote:[d]6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1
You may use 6-men egtb. Drawish score could be +/- 50 cp.
Code: Select all
Stockfish 9 64 BMI2 by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
setoption name SyzygyPath value /home/louis/Documents/Chess/tb/src
info string Found 510 tablebases
position fen 6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1
go depth 5
info depth 1 seldepth 1 multipv 1 score cp 0 nodes 5 nps 454 tbhits 1 time 11 pv g8f8
-
- Posts: 27808
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: Can your favorite engine show a drawish score?
It is just a matter of whether there is code to recogize Rook-Pawn plus wrong Bishop, not just in isolation, but also within the context of unlike Bishops. So this would fall under material combinations where the Pawn on which your win potential hinges can be stopped by a Bishop sac, for which you would typically discount by a large factor.
Unlike forward search, retrograde analysis is not blind to 50-move draws. It would even get them flawlessly if there was a billion-move rule.
That is pretty obvious. Then the engine would not have solved it, but the EGT generator...zullil wrote:Maybe you meant "You may not use 6-men egtb?"
Unlike forward search, retrograde analysis is not blind to 50-move draws. It would even get them flawlessly if there was a billion-move rule.
Last edited by hgm on Tue Feb 06, 2018 1:01 pm, edited 1 time in total.
-
- Posts: 4833
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
Re: Can your favorite engine show a drawish score?
I only have 5-men, and got this.zullil wrote:Ferdy wrote:[d]6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1
You may use 6-men egtb. Drawish score could be +/- 50 cp.Maybe you meant "You may not use 6-men egtb?"Code: Select all
Stockfish 9 64 BMI2 by T. Romstad, M. Costalba, J. Kiiski, G. Linscott setoption name SyzygyPath value /home/louis/Documents/Chess/tb/src info string Found 510 tablebases position fen 6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1 go depth 5 info depth 1 seldepth 1 multipv 1 score cp 0 nodes 5 nps 454 tbhits 1 time 11 pv g8f8
Code: Select all
72/72 -2.15 1. ... Kf8 2.Kg4 Be5 3.Kg5
-
- Posts: 4833
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
Re: Can your favorite engine show a drawish score?
I have that knowledge of drawish kbp-k (rook pawn and wrong bishop color) but the score is not drawish even after a long search. The engine will be able to draw as the king will not leave that area, but the score is ridiculoushgm wrote:It is just a matter of whether there is code to recogize Rook-Pawn plus wrong Bishop, not just in isolation, but also within the context of unlike Bishops. So this would fall under material combinations where the Pawn on which your win potential hinges can be stopped by a Bishop sac, for which you would typically discount by a large factor.
-
- Posts: 10298
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: Can your favorite engine show a drawish score?
I think that the problem is that chess programs never use games against theirselves as part of the evaluation function and when they get more time they spend all the time for searching more nodes and do not spend time on having better static evaluation function.Ferdy wrote:[d]6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1
You may use 6-men egtb. Drawish score could be +/- 50 cp.
I believe that program may do better at long time control if we decide that the evaluation function is not only function of the position but also function of the iteration and we get smarter evaluations at higher iterations by games of the program against itself so at iteration 50 we may for example get evaluation based on the result of game between the program and itself at depth 10.
-
- Posts: 27808
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: Can your favorite engine show a drawish score?
But it lacks the knowledge that a dead-draw plus Pawn vs Bishop is almost always still a draw. Search cannot easily substitute for that knowledge, because the side that imagines itself to be winning can postpone conversion to the dead-draw it does recognize for a very long time without repeating. The same would be the case when you only recognize KBK as a draw. Then it would not see anything wrong with KBPKB or KBPKN, because it will happily stay a good passer ahead upto the horizon (but without making any kind of progress). You really have to recognize these material combinations where the Pawn that represents your entire winning chance can be easily sacrificed away by a 'surplus piece' of the opponent.Ferdy wrote: I have that knowledge of drawish kbp-k (rook pawn and wrong bishop color) but the score is not drawish even after a long search. The engine will be able to draw as the king will not leave that area, but the score is ridiculous
-
- Posts: 450
- Joined: Wed Nov 24, 2010 10:57 am
- Location: INDIA
Re: Can your favorite engine show a drawish score?
[Event "?"]Ferdy wrote:[d]6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1
You may use 6-men egtb. Drawish score could be +/- 50 cp.
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[FEN "6k1/8/6PP/5K2/2B5/2b5/8/8 b - - 0 1"]
[SetUp "1"]
1... Kf8 {(0:00:00) 0kN}{[%eval 0,99,Houdini 6.03 Pro x64-NTB][%meval TB:3]}
{[%eval 0,1,Houdini 6.03 Pro x64-NTB][%meval TB:3]} *
Always Expect the Unexpected