Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: automated book generation in chess/checkers

Author: martin fierz

Date: 16:45:11 12/12/01

Go up one level in this thread


On December 12, 2001 at 18:44:10, Dann Corbit wrote:

>On December 12, 2001 at 18:31:14, martin fierz wrote:
>[snip]
>>[D] 3q1rk1/pp1bppbp/3p1np1/8/2rNP1P1/2N1BP2/PPPQ4/2KR3R w - - 0 16
>>
>>this was our starting position. i had hoped that in a sharp position as this,
>>dropout expansion would also work for chess. it didnt :-(
>>the problem was that crafty's main line went like 16. Qh2? h5? etc. the correct
>>answer to 16. Qh2? is Rxc3! which every dragon player will play in a slpit
>>second. the problem is that even if you can take the queens off with white now,
>>your position is bad. there have been too many discussions here about
>>"understanding", and i don't really want to say anything about it to avoid
>>further discussion.
>
>How long did you let crafty run?  Did you try other engines?
>
>Did you run against the forward positions?
>
>Did you analyze against all positions in all games (including future moves) from
>here and minimax that?
>
>I think you won't be able to get the right position from the root except by
>blind chance.
>[snip]

i'll try to answer as good as i can, tom did the work...
crafty was set to run at 1-3 minutes/move. the point about dropout expansion
would be that you do not need to know which move is best with this time limit,
just to get a set of good moves which you can explore further.
what the program did was to generate all successor positions to the diagram,
score them all, and continue this process for all "promising" successors, like
Qh2, but also for many other moves. then it backs up the values it gets to the
root of the tree, and can also change it's mind about which successors it should
look at at the root, if a different score than the original is backed up. also,
when the most promising successors have been expanded, it goes back to the less
promising ones and expands them a bit, just much less than the more promising
ones.

since you have much more positions in your database, you might want to look up
what it says on this position (score/best move)?

cheers
  martin

i don't know in what way to improve this as long as the engine does not know
that the Qh2 Rxc3 position is not good for white.



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.