Author: Robert Hyatt
Date: 20:45:37 10/15/99
Go up one level in this thread
On October 15, 1999 at 19:29:17, Antonio Dieguez wrote: >On October 15, 1999 at 15:28:55, Robert Hyatt wrote: > >... >>>>No.. he makes a pass over the entire move list to pick the move with the >>>>highest 'priority'. Then the next time he wants a move, he does this again. >>>>One pass per move is the only way I can see to do it unless they are actually >>>>sorted by priority first, which is a waste of cpu cycles... >>> >>>but he has only 7 level prioritis, or he said another thing? >> >>sure, but he generates moves in random order and attaches a 'priority' to each >>one. You have to either (a) scan the entire list and pick the remaining move >>with the highest priority or (b) know which priority you are looking for and >>scan (on average) 1/2 of the list to find it, assuming there is a move with such >>priority. >> >>still N^2 basically... no matter how you cut it... > >ok, first of all I was no counting ataching the prioritys, anyway he knows what >he do. > >>>well, another question, is good to maintain 1 killer capture? I think yes... >>>who maintains a killer capture? >>> >> >> >>nobody I know of... why? because we try captures _before_ killers anyway, >>at least those that don't appear to instantly lose material. So by the time >>we get to killers, the captures are already rejected. > >sure! I was not saying that. > >(killers moves->before others moves , killer capture->before others captures) > >>>me. >>>(the prince of amy mizuno, sailor mercury.) > >me again. Killer captures have been tried many times. They don't work. The reason is that often the capture you want to try is the capture that rips the piece moved at the previous ply since that move hung the piece (common in full-width search). A killer capture doesn't do this. And doesn't work well as a result....
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.