Author: Andrew Dados
Date: 22:41:59 09/23/00
Go up one level in this thread
On September 23, 2000 at 16:28:09, William Bryant wrote: >I am adding a specific Mate search to my program and trying various >combinations of search settings. I was curious about the setting others are >using. > >For a search for mate (in an unknown number of moves) >I was going to set Beta to Infinity (32768), Alpha to a mate limit (32768 - 300) >and use iterative deepening, no PVS, ignore fail lows at the root >(which means it didn't find a mate at that depth). >I was also going to use NULL moves but no PVS since until mate is found, >you do not have a best move. > >For a search for mate in N, I was going to change alpha to Infinity - plydepth >(32768 - (N * 2)), and set the initial iterative depth to N*2 rather than use >iterative deepening. This should say if there is a mate at this depth. > >Does this sound like a reasonable approach? > >Any comments would be appreciated. > >William >wbryant@ix.netcom.com It is probably easier to write a dedicated searcher for all it is worth. Note you need no qsearch and eval=0 all the time... so you just need an alpha-beta loop, move generator and in-check indicator... -Andrew-
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.