Author: Dan Newman
Date: 16:49:42 01/12/00
Go up one level in this thread
On January 12, 2000 at 13:41:29, m.d.hurd wrote: >On January 12, 2000 at 12:09:08, John Merlino wrote: > >>On January 11, 2000 at 22:15:50, Imran Hendley wrote: >> >>>On January 11, 2000 at 20:06:00, John Merlino wrote: >>> >>>>On January 11, 2000 at 03:11:42, m.d.hurd wrote: >>>> >>>>>On January 11, 2000 at 03:07:19, m.d.hurd wrote: >>>>>Correction : >>>>> >>>>>Fritz 6 solves this in seconds and keeps announcing the mate up to where it >>>>>promotes to a knight then draws by three fold repitition. >>>>> >>>>> >>>>>Start position >>>>> >>>>>[D] 8/8/8/2p5/1pp5/brpp4/1pprp2P/qnkbK3 w - - 0 1 >>>>> >>>>>Regards >>>>> >>>>>Mike >>>> >>>>Chessmaster falls into the same trap as Fritz (et. al.). It announced mate in 15 >>>>in 8 seconds (on a PII-450), and proceeded to promote to a queen for a 3-fold >>>>draw. >>>> >>>>But, if I take it past the knight promotion, it DOES work out the mate in 9 from >>>>there. >>>> >>>>jm >>> >>>I don't see the logic behind this. When it announces mate does it see the knight >>>promotion? If so why does it promote to a Queen? And if when it announces mate >>>it sees a Queen promotion, how can this be, because there is obviously no mate >>>with a Queen. Do you know why this happens? >> >>I have no idea why it happens. It does appear to defy logic. But there must be >>SOME programmatic reason for it, since Fritz appears to do its own version of >>the same thing (announce mate, promote to knight, draw by 3-fold). >> >>jm > >And not just Fritz 6, if you take the position after the pawn has promoted to a >Knight, neither Fritz 6 or 5.32 or Hiarcs 7.32 or Crafty 17.4 find the mate. >Doctor 3 does and also comet b10. So some solve it and some do not, maybe >someone else knows the answer why it should be so difficult. > >Mike Well, my program (Shrike) doesn't find the mate after searching 9+ hours unless I turn off null-move, in which case it finds it in 15 seconds... In order to mate you have to put the knight on b3, but null move allows black to pass and leave its queen on a2, preventing this. So it looks like there could be a null-move problem with these other programs too. Perhaps they turn it off when white has only a king and pawn but turn it back on for king and knight, or perhaps the large amount of black material keeps it on for one side or something... Anyway, it's a very interesting position. -Dan.
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.