Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fritz endgame CD

Author: José de Jesús García Ruvalcaba

Date: 07:36:46 04/05/99

Go up one level in this thread


On April 02, 1999 at 17:52:03, James T. Walker wrote:

>On April 02, 1999 at 10:51:27, José de Jesús García Ruvalcaba wrote:
>

[big snip]

>>
>>	To make the right choice with a pawn on the seventh rank (which can promote to
>>four different pieces), the promotion files are required. That is because
>>"probing at the root" really means probing after a one-ply full width search.
>>José.
>
>Hello Jose,
>Thanks for the info.  I understand what you are saying but I don't understand
>why "the promotion files are required".

Because the root position is not probed, what is probed is the positions after a
one-ply search. If there is a pawn on the seventh rank which can move, it can
promote and thus the score for the positions after the promotion are required.

> I know that would make the promotion
>easy but why couldn't you just include the promotions in the database you are
>using?

I am not sure I understand. The positions after the promotion can be reached by
many other ways (without any promotion), and the score should be the same.
Including only the position after the promotion seems cumbersome (and its score
can not be computed by itself, anyway), and it has not been done (at least in
the major tablebase sets). It is posible, but it would require different probing
code.

>I see no reason why you can't include these promotions since you are not
>planning on having the next set of tablebases.

Until now it is usually assumed that you have the next set of tablebases.

> Wouldn't a little planning allow
>this?

Many people think so. Somebody has even modified crafty to handle missing
promotion or capture classes. I do not know if those efforts have been
succesful. Anyway, it is easier to have the tablebases (but they are big!).

> Or is it impossible to add these to the end after the tablebases are
>generated?

I do not understand this question.

> This begs the question- How do you decide which tablebase to go to
>when there are 4 possibilities?  Do you probe all 4 tablebases before making the
>promotion?

Yes.

> If that's true then it seems like putting the promotion in the
>tablebase already in use is more sensible.

As I said, it has not been done.

>Jim Walker

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.