Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Gothmog goes commercial

Author: Uri Blass

Date: 11:23:31 04/02/04

Go up one level in this thread


On April 02, 2004 at 13:53:03, Tord Romstad wrote:

>On April 02, 2004 at 13:01:02, Matthias Gemuh wrote:
>
>>Hi Tord,
>
>Hi Matthias!
>
>>I think the next version of Gothmog shall be "of commercial strength".
>
>Thanks for the kind words, but I think you are too optimistic.  And even if it
>will some
>day achieve "commercial strength", it will never be of "commercial quality".
>
>>You deserve the money.
>
>I don't want money.  I want people to have fun and be happy, and computer chess
>in general (not only my engine) to advance.  :-)

In that case I hope that you will support my idea of general search code.
The idea is to have code(not chess related) that can be used for chess and for
other games when people can use it for chess and for other games.

I plan to think later again about improvement for that code and to add
information.

I think that it is important to have a small tree with probabilities for every
move that the program remembers and modify during the search.

I think that evaluation of every node in terms of probabilities and not in a
number is also important(not only in chess) and I may add in the general search
code something about it.

I can give an example

A good evaluation can evaluate that black has 0 chances to win here(I think that
it is a draw but assume for the discussion that there is no knowledge to be sure
about it)

[D]5rk1/6pp/8/8/8/8/5PPP/5RK1 w - - 0 1

If you evaluate the root position as equal you may want to prune this position
based on your probabilities when the remaining depth is not important when
another position when black is a piece down may be interesting to search because
you evaluate it as 90% win for white but 5% win for black so continue to search
may reveal this information.

see http://www.talkchess.com/forums/1/message.html?357887 for my last post in
this subject.

My code still does not code special code that is using some function to evaluate
probabilities of win,draw,loss but I think that general code should allow this
possibility(there is also need to add general hash code and general iterate
code).

I plan to think about the problem and I may post more code later and if you can
help me by looking at the code and post your ideas it may be productive.

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.