Author: Iddo Bentov
Date: 03:28:17 07/26/01
Go up one level in this thread
>>we're thinking about ideas for a project in computer chess, we tried
>>to focus on ideas that involve chess endings..
>>
>>so far we haven't found interesting ideas that haven't been tried
>>before..
>>
>>our request.. if anyone has any new/interesting ideas in computer
>>chess that he thought about and are worth trying.. please reply
>>and tell us? the ideas don't have to be of the kind that actually
>>improve the current strong chess programs.. just ideas that are
>>interesting..
>
>Get a KP vs K database, figure out how to dump all the values, and write a
>function that produces the same basic values ("win" for won positions, "draw"
>for drawn positions) in every case, by invoking rules.
>
>This has already been done by Don Beal a long time ago, but it might be worth
>trying to see if you could do it with fewer rules. Also, I'm not certain he
>checked his against an endgame database.
>
>You could do more complicated cases if you felt up to it, but I think the odds
>are high that you'd end up producing nothing. The interesting cases are 5-man
>endings, but I don't think anyone has had real success with rules for even 4-man
>cases (KR vs KN, for instance).
>
>Another possibility is to figure out a way to solve "wild 7". This was
>supposedly done by Hans Berliner, but I've never seen an article.
>
>[D] 4k3/5ppp/8/8/8/8/PPP5/3K4 w - - 0 1
>
>White to move wins.
>
>bruce
>
>
>bruce
hello..
your post is very interesting.... thanks very much for your reply..
your first ideas is simple and interesting.. exactly the kind we
were looking for.. we'll look into it.. if there is an article by
Don Beal we could look up.. it could help.. i tried www search but
nothing much came up.. i wonder how important minimizing the number
of rules is.. and wonder what percent of correctness we should
expect from such a function.. e.g. for KP vs K and KR vs KN
about wild7 (i play it on icc) and in general.. what sort of methods
can be used to solve a position whose minimax tree (with alpha-beta
and using hashtable etc) is still intractable..? it seems that for
wild7 patterns recognition could help? but in general.. are there
any interesting methods to solve positions?
btw it seems that computer programs aren't good at wild7? at least
on my old p133 cpu.. is it because of null-moves ? i notice that
crafty on icc accepts many wild types.. but not wild7
also i'm interested to know, are there any chess positions that have
a known solution proved by humans, but none of the computer programs
can solve no matter how long they think? i think that in checkers
there is known opening (white doctor) where all computer programs
make the wrong move.. i wonder if it is always possible to create
a position that will trick all chess programs.. if such positions
exist in chess.. then i wonder what are such positions with the least
amount of pieces?
any comments would be appreciated....
thanks again....
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.