Author: Robert Hyatt
Date: 12:31:59 03/08/02
Go up one level in this thread
On March 08, 2002 at 12:29:32, Uri Blass wrote: >On March 08, 2002 at 12:01:47, Robert Hyatt wrote: > >>On March 08, 2002 at 00:40:01, Dieter Buerssner wrote: >> >>>On March 07, 2002 at 22:50:05, Robert Hyatt wrote: >>> >>>>On March 07, 2002 at 22:45:55, Dieter Buerssner wrote: >>>> >>>>> 5b1k/3R4/5P2/p1rB4/8/8/6K1/8 b - - >>>>> >>> >>>>>Perhaps I am too naive, but when I watched this game, I first was confident, >>>>>that it should be a very easy draw. My (stupid?) reasoning was: A computer will >>>>>not lose KRBKR (when he does not see the exceptional win fast). The position >>>>>here is KRBPKRBP. The Bs are of opposite color. So, I thought, it should allways >>>>>be possible, to sac one B for a P, and have an easy draw. Well, I was totally >>>>>wrong, and so was the knowledge I coded into my engine. >>>>> >>>>>How long does your engine need, to avoid a4? How long does it need, to see a >>>>>convincing winning score after a4? (Kg3 is a forced win, other moves perhaps, >>>>>too). >>> >>>>Crafty decides a4 is not so good pretty quickly (on my laptop here)... >>> >>>Yes. >>> >>>> 12-> 58.63 1.45 1. ... Rb5 2. Rd8 Kh7 3. Be4+ Kg8 4. >>>> Rd7 Rb2+ 5. Kg3 Rb6 6. Bd5+ Kh8 7. >>>> Rf7 Bc5 8. Kf4 a4 >>> >>>This will draw. >>> >>>>If you force a4 and let it search for white: >>>> >>>> 11-> 19.99 2.18 2. Kf3 Rc1 3. Kf4 Rf1+ 4. Kg5 Re1 5. >>>> Kg6 Rg1+ 6. Kf7 Bc5 7. Rd8+ Kh7 8. >>>> Be4+ Kh6 9. Ke6 >>> >>>While the score does not look too convincing yet, Kf3 will win. >>> >>>Thanks for all the answers. However, I'd also be interested in comments about >>>the "knowledge" above. Was this just bad luck? Or is it just stupid to assume, >>>that KRBPKRBP with Bs of opposite color should be draw? >>> >>>Regards, >>>Dieter >> >>I don't assume draw if both sides have pawns. It is too easy to get a B >>pinned and let the pawn slip by If one side has a KRB and the other has >>a KRBP then opposite B's should guarantee that the P can't win. I don't >>try to get to fancy here because there are too many exceptions when both >>sides have pawns... > >I do not think that 0.00 is a good idea because it means that the program does >not care about the question if the king is in a bad square but I still believe >that most KRBP vs KRBP with opposite color bishops are drawn so dividing the >regular evaluation by 4 in this case to be closer to a draw does not seem to be >too dangerous. > >Uri I divide it by 2 in this case, which I consider to be perfectly safe, since it still lets one side with a bad king or strong rook build up a significant positional advantage in terms of score... I do this for bishops of opposite colors only, however...
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.