Author: Robert Hyatt
Date: 10:38:24 09/22/99
Go up one level in this thread
On September 22, 1999 at 10:54:04, Ricardo Gibert wrote: >Ok, so some programs do not examine underpromotions, since it seems to be a net >gain on average. I can understand why some would not like that, so I have a >simple compromise. Good underpromotions tend to occur in one of 2 cases: >1) Greatly reduced material. Both players have only zero or one pieces on the >board (I'm not counting Ks or Ps here of course). >2) Promotion to an N is with check. > >So the compromise would be to test for the above 2 cases to determine whether an >underpromotion should be examined. This should be more economical than examining >each of the underpromotions. The chance of overlooking a good underpromotion >should be nil. > >Reasonable? I think you will find the main objection to underpromotions is in the move generator, rather than having to search the resulting positions. Because you now have a loop to emit 4 under-promotions. And checking the conditions you mention would be a handicap in a move generator that is trying to cheat on the speed issue... I just produce them and go on...
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.