Author: Vincent Diepeveen
Date: 05:59:12 04/06/03
Go up one level in this thread
On April 06, 2003 at 08:42:56, Ian Kennedy wrote:
>On April 06, 2003 at 08:37:13, Dan Andersson wrote:
>
>> The short answer is that there can be no real answer to your question except
>>for very small graphs. But there are ways to approximate:
>>http://www.cs.vu.nl/~aske/Papers/acm-final.pdf
>>http://citeseer.nj.nec.com/plaat96exploiting.html
>>http://www.cs.vu.nl/~aske/Papers/th-5-6.pdf.gz
>>
>>MvH Dan Andersson
>
>Thanks for these links, although the first one is the Schaeffer paper I was
>quoting from which sadly only refers to a node counting method without
>describing it and refers the reader to a book which is now out of print. I will
>check out the other two as well now.
>
>Ian
Ian i need to add 1 note and that's that only search times matters.
In DIEP i can do next:
if( depthleft == 1 ply && incrementalmaterial < alfa )
THEN do not try nullmove.
Now i need a lot less nodes in diep than without this nullmove prevention rule.
However DIEP needs more time to finish each ply. So which is better?
Right. The faster search time is better.
Reason is of course transpositiontable.
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.