Author: Dann Corbit
Date: 12:06:29 01/19/05
Go up one level in this thread
On January 19, 2005 at 11:49:13, Ricardo Gibert wrote: >This idea is half baked, but bear with me. > >What if we were to define the partial solving of chess as showing that White has >at least a draw. In other words, instead of trying to show it to be 1 of 3 >possibilities (i.e. win, draw, loss), try to reduce the possibilties to just 2 >of 3 possibilities (e.g. win, draw). > >Now how could we modify a perhaps material only search of the beginning position >to show this? > >What if were to score repetitions, stalemate & draw by insufficiency of material >the same way as a checkmate given by White. Then it is just barely possible that >a 25 ply or so search of the beginning position would return a "mate" score >indicating White has at least a draw in chess. Failing that, perhaps one of the >960 FRC start positions can be shown to be at least a draw for White. > >If the above modification is flawed, perhaps someone can think of different >modification that would do the trick. The point is that my lowering our sights, >we just might possibly come up with a partial solution to chess or FRC today. A >long shot for sure, but maybe worth a try. > >BTW, a draw by insufficiency of material can be loosened somewhat by requiring >only that Black have insufficient material to mate though I would be surprised >if this change would prove useful. That's an awful lot of captures to squeeze in >within the search horizon of todays programs. It will happen during a search but >it would probably not really be consequent. There are 14,482 forced direct mates easily seen within the first 5 plies of chess. (I found it a bit surprising that there are that many). So we can close off these branches as dead. That process could be extended until there are no more branches left. It might be a slower way to do it than a search from the root though. Here is a list of detected forced mates within the first 5 plies: ftp://cap.connx.com/pub/chess-engines/new-approach/r12345dm.epd.bz2
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.