Author: Vincent Lejeune
Date: 01:54:43 06/15/04
Go up one level in this thread
On June 14, 2004 at 22:29:03, Anthony Cozzie wrote: >On June 14, 2004 at 21:49:38, Mike Byrne wrote: > >>On June 14, 2004 at 20:42:45, Anthony Cozzie wrote: >> >>>On June 14, 2004 at 18:11:17, Derek Paquette wrote: >>> >>>>On June 14, 2004 at 15:38:12, Daniel Jackson wrote: >>>> >>>>>With todays top hardware/software, can the winning move e5!! be found? >>>>> >>>>>[D]r2qrb1k/1p1b2p1/p2ppn1p/8/3NP3/1BN5/PPP3QP/1K3RR1 w - - 0 1 >>>>> >>>>>Daniel >>>> >>>> >>>>I think that these position tests are absolutely useless, >>>>an IM can draw a program, so who cares if it can find 'specific' moves >>>> >>>>if it doesn't have a 'game plan' and can only draw, >>>> >>>>whats the difference? >>> >>>Actually I'm pretty happy with this result. This makes at least 5 of the 11 >>>nolot positions that Zappa can solve in reasonable (a few hours) time. Of >>>course, I am a long way from it playing the move over the board at tournament >>>time control! >>> >>>1. Nc3-e2 e6-e5 2. Nd4-f5 g7-g5 3. Bb3-f7 Bd7-e6 4. Bf7xe8 Qd8xe8 5. h2-h4 g5-g4 >>>6. Nf5xh6 Bf8xh6 7. Rf1xf6 Bh6-g7 8. Rf6-f1 Qe8-a4 9. Ne2-c3 >>> = (1.87) Depth: 17/38 00:58:56.78 2666999kN >>>1. e4-e5 d6xe5 2. Nc3-e4 Nf6-h5 3. Qg2-g6 Nh5-f4 4. Rf1xf4 e5xf4 5. Nd4-f3 >>>Qd8-b6 6. Rg1-g5 Qb6-d8 7. Nf3-e5 Qd8-e7 8. Ne5-f7 Kh8-g8 9. Ne4-f6 Qe7xf6 10. >>>Nf7xh6 Kg8-h8 11. Nh6-f7 Qf6xf7 12. Qg6xf7 >>> = (2.28) Depth: 17/34 01:16:27.50 3451482kN >>> >>>Positions like these show that (IMO) more depth is *always* better. Maybe not a >>>lot better, but Zappa is still finding new tactics at 17 ply. >>> >>>anthony >> >>The key to e4e5 is that the program must see Ne4 is good , leaving the Nd4 >>hanging by design -- just like Zappa shows. But even pv's can be misleading - >>after 7. ... Qe7 , the above position is Mate-in-7 with Rh5 >> >>[d]r3rb1k/1p1bq1p1/p3p1Qp/4N1R1/4Np2/1B6/PPP4P/1K6 w - - 0 6 > >Don't whine about my 3 ply search Mike ;) > >1. Ne5-f7 Kh8-g8 2. Ne4-f6 Qe7xf6 3. Nf7xh6 Kg8-h8 4. Nh6-f7 Qf6xf7 5. Qg6xf7 > = (2.28) Depth: 3/11 00:00:00.02 3kN >1. Rg5-h5 Re8-c8 2. Ne4-g5 Qe7xg5 3. Ne5-f7 Kh8-g8 4. Nf7xg5 Bf8-c5 5. Qg6-f7 >Kg8-h8 6. Rh5xh6 g7xh6 7. Qf7-h7 > = MAT7 Depth: 9/25 00:00:14.51 10595kN > >In fact Zappa makes a mistake much earlier, when it plays Qg6 (Spassky's move) >instead of the stronger Qg4. [d]r2qrb1k/1p1b2p1/p3p2p/4p2n/3NN3/1B6/PPP3QP/1K3RR1 w - - 0 1 I let Shredder 8 think for 45 min (depth 18/51) on this position in the 2-best-moves-mode (Centrino 1.7 Ghz, 96 MB hash) and it preferes Qg6 : 1. +- (7.37): 23.Dg6 Cf4 24.Txf4 exf4 25.Cf3 Fa4 26.Ce5 Dd4 27.Cf7+ Rg8 28.Cf6+ Dxf6 29.Dxf6 Te7 30.Cxh6+ Rh8 31.Dg6 f3 32.Fxa4 gxh6 33.Dg8# 2. +- (5.53): 23.Dg4 Fe7 24.Dxh5 Tf8 25.Dxe5 Txf1+ 26.Txf1 Db8 27.Dxb8+ Txb8 28.Tf7 Te8 29.Cxe6 g6 30.Fd5 Fxe6 31.Fxe6 Fd8 32.Fd5 > If I ever get my parallel version working, I'm >going to run over all the Nolot positions. > >anthony
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.