Author: blass uri
Date: 12:01:26 05/02/99
Go up one level in this thread
On May 02, 1999 at 12:23:57, Dave Gomboc wrote: >On May 02, 1999 at 09:04:43, blass uri wrote: > >> >>On May 02, 1999 at 08:15:59, Francis Monkman wrote: >> >>> >>>On May 02, 1999 at 05:54:47, blass uri wrote: >>> >>> >>>>Maybe black cannot save itself because after giving white 19.f6 g6 >>>>Junior found 20.Nxe6 and the evaluation goes up. >>>> >>>>Uri >>> >>>I'm sure 20 Nf3 was right, as Black now has only self-destructive moves >>>to make, eg your suggestion 20 ...h5 21 Nxh5 or 20 ...g5 21 Nxg5. King moves >>>also lose immediately. I'm pretty sure this is watertight, that's why this is a >>>serious question -- a 9-ply (let alone 10 or 11) should find this. >> >>21.Nxh5 is also good in this line and Junior suggests Rad8 because in this >>position it can see that 21...gxh5 is leading to mate against black. >> >>I think that you need to see more than 9,10,11 plies to see that 21.Nxh5 gxh5 is >>losing to Qd2 >>For example >>21.Nxh5 gxh5 22.Qd2 Rfc8 23.Qh6 Bf8 24.Qg5+ Kh7 25.Qxh5+ Kg8 26.Ng5 Bc5 27.Qh7+ >>Kf8 28.Qh8# is a 15 ply line and black has material advantage before the mate. >> >>It is more than 15 plies from the original position(exactly 21 plies) >> >>program may see it faster than 21 plies because of some extensions but probably >>not in 9,10,11 plies search. >> >>Uri > >Amir once said (paraphrased) that Junior's extensions are larger than even Deep >Blue's. If that were true for _this particular position_, it would be possible >to pick up the loss on an even lower iteration than 9. Junior cannot find f5 at iterations 17,18(usually eqvivalent to depth 9 in the brute force search) other programs also cannot solve it at depth 9. Genius3 shows at depth 7 after 1.f5 hxg5 2.f6 the move Qc7 with positive evaluation (1.21 pawns for black) it fail low at depth 8 and shows Qc7 with 0.00 evaluation main line 2...Qc7 3.fxg7 Kxg7 4.Nh5+ Kg8 5.Nf6+ Kg7 6.Nh5+ with perpetual check. white can get a decisive advantage in this line by 4.Bg5 Genius3 cannot see 3 plies after the original position that white is winning at depth 8. 8+3=11 so 11 plies are probably not enough for it to see that white is winning from the original position. The problem of most programs is that they do not smell the mate danger for black(otherwise they would do more extensions and find a solution at depth 9). I believe there are programs that can find the solution at depth 9 chessmaster and diep(use mate extensions) are the natural candidates. Uri
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.