Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Can this position be won under a 50-move rule?

Author: José de Jesús García Ruvalcaba

Date: 12:07:43 09/16/00

Go up one level in this thread


On September 16, 2000 at 14:33:51, guy haworth wrote:

>Consider the following position from Ken Thompson's 6-man EGT site:
>
>8/8/7B/5B2/3K4/8/2k5/n6n b - - 0 1
>
>It is a 'win' for White and also a maxDTC and maxDTM position.
>DTC = 76 plies (38 moves) and DTM = 212 plies (106 moves).
>
>Black cannot protract the first phase beyond 50 moves and therefore might be
>best advised to maximise DTM and (within that, should a choice occur) minimise
>DTC.
>
>Similarly, White might be best advised to do the following in order:
>    keep the first phase down to 50 moves or less,
>    minimise DTM (Depth to Mate), and
>    maximise DTC (Depth to Conversion).
>
>These White/Black strategies leadd to what I call an SGM-C-_SM+C+ line.
>However, White's (few) options and a Black defence along these lines can lead to
>
>[D] 6B1/1K6/1B6/8/8/8/3nk3/8 b - - 0 40
>
>and a KBBKN 2nd-phase allowing a 50-move draw claim.
>
>So, the question is "Can this position be won under a 50-move rule, and if so,
>how, given 'best play' by Black?".
>
>G

	The way to answer this question (and several others) is to produce a new set of
tablebases, which are neither DTM nor DTC.
	For positions which are won but might (or might not) infringe the 50-move rule,
two numbers are stored: the number of conversions needed to deliver mate and a
new number, let's call it Marimar number.
	Now the way to compute the Marimar number for a given won (except posible for
the 50-move rule) position X. If it is a mate, then the Marimar number is zero
and the number of conversions is also zero. Otherwise, since the tablebases are
computed by backwards analysis, this means that we have a position Y which can
be reached by X and that is won (again, save the 50-move rule). If the move
which leads to Y from X is not a conversion, then the number of conversions for
X is the same as Y, and Marimar number for X is the Marimar number for Y plus
one. If the move from X to Y is a conversion, then we have two cases:
a) If Marimar number for Y is less than 100, then for X it is 1.
b) If it is no less than 100, then for X it is 101.
In either case increase the number of conversions by one.
	(It is needed a decision procedure when X leads to more than one winning
position, and it is to follow the win with LOWER Marimar number).
	In these tablebases, won positions have a Marimar number lower than 100; while
position which would be won without the 50-move rule but are a draw with it have
a Marimar number no less than 100.
	Of course a special value for draws and a similiar reasoning for lost positions
is also required.
	I am surprised nobody has tried this simple solution before.
José.



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.