Author: Uri Blass
Date: 12:36:07 04/06/03
Go up one level in this thread
On April 06, 2003 at 15:26:34, Vincent Diepeveen wrote: >On April 06, 2003 at 11:00:25, Uri Blass wrote: > >>On April 06, 2003 at 09:55:19, Vincent Diepeveen wrote: >> >>>On April 06, 2003 at 09:48:56, Ian Kennedy wrote: >>> >>>>There is no qsearch. What exactly do you mean by 'what evaluation function do >>>>you use'? >>> >>>then this renders your experiments to useless, because experiments with a >>>gametree that suffer from horizon effects which get backupped to the root are >>>completely useless for obvious reasons. >>> >>>Why not measure something in the game of 4 connect? No need for a qsearch there. >>>No need for nullmove. >> >>That game was solved in 1988. >> >>If you want to play it in demensions that were not solved then >>I do not understand why no need for qsearch and null moves. >> >>I think that qsearch can help in that game and when the side that moved threats >>to win in the next move then you can extend preventing him to win in the qsearch >>until no side threats to win or can win in the next move. > >that's too slow in 4 connect. Faster is searching at 10 million nps a second >fullwidth using iterative deepening. I do not think that it is faster if the target is to play and not solve the game and the dimension of the board is not so small to do it a solved problem. by searching 10 miliion nps a second. suppose for the discussion that the board is 99*99. 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.