Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder learning must be better than Fruit 2.2.....................

Author: Thomas Logan

Date: 19:14:09 10/20/05

Go up one level in this thread


On October 20, 2005 at 22:03:47, Ryan B. wrote:

>On October 20, 2005 at 21:16:11, Thomas Logan wrote:
>
>>On October 20, 2005 at 20:37:45, Jorge Pichard wrote:
>>
>>>On October 20, 2005 at 19:34:51, Thomas Logan wrote:
>>>
>>>Shredder learning must be better than Fruit 2.2, and after 75 games Shredder
>>>should be ahead  :-)
>>
>>Yes that is a possibility though I am not sure and tend to doubt it but
>>
>>How would you check it ?
>>
>>Other than running another match of equal duration
>>
>>Fruit appears to have been helped by the addition of df8.ctg to book.bin
>>
>>and was + 3 after the first 10 games
>>
>>thereafter fruit played with own book alone
>>
>>What I have done is put all the fruit-shredder games into chessbase clipbase
>>
>>and organized by eco codes
>>
>>I then look for duplicate openings to see if perhaps shredder was helped by a
>>previous game position
>>
>>There was no apparent evidence that Shredder benefitted from this
>>
>>Perhaps this is the wrong way to test - I am unsure as to how position learning
>>works and maybe it works negatively in the avoidance of a position
>>
>>I don't know you tell me
>>
>>Tom
>
>I think it would take a lot of games for position learning to be much factor.

I agree



 I
>cant see position learning being bad for an engine because the load into hash
>would be at init so no slow down during game play and if the engine avoids a
>position all together due to learning it means it could not find a good score
>from any of the moves it did not play in that position last time. Book learning however could be a controlling factor especially if the opposing engine does not
>have a well tuned book or is using a book not tuned for that engine.  I do not
>have windows so maybe someone can correct me on this but I think CB GUI handles
>book learning for each engine using a CB book.

If this is true than both
>engines would have equal learning chances so long as the book is good for the
>engine using it.  I would not pick the fritz book for Fruit however due to a
>much different engine play stile and see this as a disadvantage for Fruit.
>
>Ryan



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.