Author: Uri Blass
Date: 01:25:52 06/13/01
Go up one level in this thread
On June 13, 2001 at 04:03:00, Dieter Buerssner wrote:
>On June 13, 2001 at 02:57:57, Uri Blass wrote:
>
>>On June 13, 2001 at 01:51:34, Dieter Buerssner wrote:
>>
>>>On June 11, 2001 at 01:31:14, Jouni Uski wrote:
>>>
>>>> 8/4k3/2K5/6p1/6B1/7P/8/8 w - -
>>>
>>>[...]
>>>>Hmm. Should this already be solved poblem?
>>>
>>>I think, if you can give a rule, under which conditions such positons -
>>>characterized as one side has a rook pawn and a "wrong" bishop and the other
>>>side has a pawn in the file next to the rook pawn - are drawn, it will soon be a
>>>solved problem for many chess engines :-)
>>>
>>I believe that every position when white has the wrong bishop when black has not
>>more than one pawn is a draw if black can get into the corner or near the
>>corner because4 capturing the pawn with the pawn means stalemate.
>
>I think, I can see your point. However look at this position:
>
>[D]8/7k/4K1p1/8/7P/3B4/8/8 w - -
>
>With 4-men TBs:
>
> 492 0.084 Mat10 4t 1.h5 Kh8 2.hxg6 {EGTB} 2...Kg7! 3.Be4 Kf8 4.Bc6
> Kg7! 5.Kf5! Kg8 6.Kf6! Kh8 7.g7+! Kg8 8.Bd5+!
> Kh7 9.g8=Q+ Kh6 10.Qh8# {1351}
>
>
>I think, it fits your stated rule (or I have misunderstood ...). Black is
>already on one of the usually "safe" places with his king. It is laso two steps
>nearer to h8 than the white king (and pawn). This will without a black pawn
>allways be suffisient for a draw (as you explained to me very well some time
>ago). But still here, it is lost for black. I have no idea, if this has any
>practical relevance for a chess engine. Probably, in almost all case, such a
>rule will still work, and in the exceptional cases, a very shallow search will
>help.
>
>Regards,
>Dieter
You are right so it seems that if you want a practical rule to decide about draw
without search then you need to use a better idea(I know that at least old
versions of Junior use a wrong rule to decide when to stop to search and the
result is that it is possible to beat them in drawn KBP vs K without
tablebases).
The rule is wrong only in rare cases but a human who know about the weaknesses
of Junior can lead it to exactly the rare positions when it has problems.
I believe that if the king is on the corner and it is black to move or if it is
white to move and white cannot play pxp it is a draw but I did not think about
it to find if there are no exception.
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.