Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question for Bob: Crafty opening book

Author: Robert Hyatt

Date: 18:02:36 06/08/98

Go up one level in this thread


On June 08, 1998 at 19:35:17, Edward Screven wrote:

>On June 08, 1998 at 18:42:30, Robert Hyatt wrote:
>
>>On June 08, 1998 at 16:47:20, Edward Screven wrote:
>>
>>>On June 08, 1998 at 07:36:10, Robert Hyatt wrote:
>>>
>>>>On June 08, 1998 at 02:58:23, Roland Pfister wrote:
>>>>
>>>>>... would it be possible for you to create a Crafty book from your
>>>>>large PGN file (you call it decent :-) and then create a PGN file
>>>>>from the opening book that does not contain the singular lines
>>>>>that are not used anyway ...
>>>>>
>>>>>Roland
>>>>
>>>>yes.  I'm going to do this, in fact.  The only issue is that I am
>>>>going to use the {comment} stuff to include wins/losses/draws, so it
>>>>will remain standard PGN ...
>>>
>>>why not represent the result summary as a normal pgn header tag?
>>>that way existing general purpose pgn processing software won't
>>>have to be modified to understand special comments.  something
>>>like
>>>
>>>    [Result-Summary "231/224/304"]
>>>    1. e4 e5 *
>>>
>>>(hopefully with longer lines!)
>>>
>>>  - edward
>>
>>
>>that will lose a lot of information.  IE 1. e4 will have a big number
>>of games, but by the time you get to move 10, the number of times that
>>move is played will shrink.  I don't see another easy way to do this,
>>although most programs should ignore { xxx yyy zzz } type comments and
>>they would still get a usable book.  Other ideas are welcome, however.
>
>let's say the original pgn file contains only these games:
>
>    1. e4 e5  2. Nf3  1-0
>    1. e4 e5  2. Nc3  0-1
>    1. e4 d6  2. Nc3  1/2-1/2
>    1. e4 d6  2. d4   1-0
>
>now you generate a summary pgn file, including only moves that are
>played at least twice.  (twice instead of thrice or more makes the
>example easier.)  what i was proposing is that the output would be
>
>    [Result-Summary "1/1/0"]
>    1. e4 e5 *
>
>    [Result-Summary "1/0/1"]
>    1. e4 d6 *
>
>what information (other than what you wanted to discard) is lost?
>
>  - edward


aha... you are thinking of producing one game for *each* alternative
in the book?  I'm looking at doing something much shorter, as I did in
Cray Blitz...   your idea will certainly work, but it will produce a
really large pgn file I'm afraid...  the CB format was more of a "tree"
type format without having 1. e4 in the file thousands of times...

Bob



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.