Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Instability thing...

Author: Uri Blass

Date: 12:22:35 09/18/04

Go up one level in this thread


On September 18, 2004 at 11:34:28, Sune Fischer wrote:

>On September 18, 2004 at 07:02:22, Andrew Platt wrote:
>
>>On September 18, 2004 at 04:05:38, Sune Fischer wrote:
>>
>>>
>>>I decided to try out the triangular PV thing Bob
>>>speaks so highly of, to see if it improves move ordering...
>>>
>>>I was careful to terminate the PV on all exact scores - of course.
>>>Still I was getting illegal moves in the PV.
>>>
>>>It turned out to be a hash/nullmove problem.
>>
>>There's no score that could be responsible for having illegal moves in the PV.
>>There are two main things I had to take care of to get rid of illegal moves:
>>
>>a) Make sure you truncate the PV for the ply you are entering before you start
>>doing a null move / iterating through the move list / internal iterative
>>deepening, etc.. In other words before you do anything that will increase the
>>current ply. Obviously since you are using a triangular array, you aren't
>>touching the PV stored higher up when you do this.
>
>This might work, but I think you have to update the PV instantly on all new PV
>nodes, otherwise you will overwrite the child PV at the next move.
>
>Currently I wait with the PV copying to the end of that node when all the moves
>have been searched, then I only have to update the PV once (or maybe not at all
>if there is a FH move before that).

I do not wait  with the pv copying and it is one of the things that I may change
in the future but I do not understand why waiting is a problem.

You can have a varaible for the best move that you found and a variable for the
first alpha when you began to search the ply and update the pv only when you
return alpha in the end of the search when alpha!=first_alpha.

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.