Author: blass uri
Date: 22:22:46 05/03/00
Go up one level in this thread
On May 03, 2000 at 23:33:17, Robert Hyatt wrote: >On May 03, 2000 at 22:14:48, blass uri wrote: > >>On May 03, 2000 at 21:40:09, Robert Hyatt wrote: >> >>>On May 03, 2000 at 20:33:38, blass uri wrote: >>> >>>>On May 03, 2000 at 18:29:24, Robert Hyatt wrote: >>>> >>>>>On May 02, 2000 at 09:07:42, Stefan Meyer-Kahlen wrote: >>>>> >>>>>> >>>>>>In my set of tactical test positions there is the following example: >>>>>> >>>>>>[D] 2rqk2r/pb1nbp1p/4p1p1/1B1n4/Np1N4/7Q/PP3PPP/R1B1R1K1 w k - bm Rxe6 >>>>>> >>>>>>I can?t remember where I have found this position, but the solution is supposed >>>>>>to be Rxe6. Until know every version of Shredder has found this move sooner or >>>>>>later and kept it as a best move until the end of my test time. >>>>>> >>>>>>Last night I have tested my latest version and found something interesting. >>>>>>Shredder played Rxe6 after a while as usual, but got a terrible fail low for >>>>>>that move at ply 11. >>>>>> >>>>>>So here is my question: >>>>>>Is Rxe6 really the best move and I have introduced a bug somewhere in my code or >>>>>>is Rxe6 a mistake? I didn?t find an improvement for white after 1. Rxe6 fxe6, >>>>>>but this can be due to my limited chess knowledge. >>>>>> >>>>>>Here is Shredder?s output: >>>>>> >>>>>> 6.01 0:00.71 -0.23 1.Bh6 a6 2.Bd3 Bf6 3.Nf3 a5 (63.557) 88.3 >>>>>> 6 -> 0:01.76 -0.23 1.Bh6 a6 2.Bd3 Bf6 3.Nf3 a5 (159.299) 90.2 >>>>>> 7.01 0:02.71 -0.09 1.Bh6 a6 2.Bd3 Bf6 3.Nb3 N5b6 4.Nxb6 Qxb6 >>>>>> (240.167) 88.3 >>>>>> 7 -> 0:04.90 -0.09 1.Bh6 a6 2.Bd3 Bf6 3.Nb3 N5b6 4.Nxb6 Qxb6 >>>>>> (435.029) 88.6 >>>>>> 8.01 0:07.96 -0.01 1.Bh6 a6 2.Bf1 N7f6 3.Rad1 Qa5 4.Bg7 Nf4 >>>>>> (704.124) 88.3 >>>>>> 8 -> 0:10.15 -0.01 1.Bh6 a6 2.Bf1 N7f6 3.Rad1 Qa5 4.Bg7 Nf4 (911.624) 89.7 >>>>>> 9.01 0:17.93 -0.17 1.Bh6 a6 2.Be2 Bg5 3.Bf3 Nf4 4.Qg3 Bxh6 5.Bxb7 >>>>>> (1.594.580) 88.8 >>>>>> 9 -> 0:32.29 -0.17 1.Bh6 a6 2.Be2 Bg5 3.Bf3 Nf4 4.Qg3 Bxh6 5.Bxb7 >>>>>> (2.918.536) 90.3 >>>>>>10.01 0:42.76 -0.14 1.Bh6 a6 2.Be2 Qa5 3.b3 e5 4.Nf3 e4 5.Nd2 f5 >>>>>> (3.876.621) 90.6 >>>>>>10.08 2:02.76 -0.13++ 1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ >>>>>> Kf8 6.Nxb7 (11.210.309) 91.3 >>>>>>10.08 2:26.60 +0.22++ 1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ >>>>>> Kf8 6.Bh6+ Kf7 7.Nxb7 a6 8.Nxa5 Rc1+ 9.Rxc1 >>>>>> (13.366.273) 91.1 >>>>>>10.08 3:25.78 +0.40 1.Rxe6 Rc7 2.Re2 O-O 3.Bh6 Re8 4.Rae1 a6 5.Bd3 a5 6.b3 >>>>>> (18.827.259) 91.4 >>>>>>10 -> 3:45.35 +0.40 1.Rxe6 Rc7 2.Re2 O-O 3.Bh6 Re8 4.Rae1 a6 5.Bd3 a5 6.b3 >>>>>> (20.541.368) 91.1 >>>>>>11.01 5:35.76 +0.15-- 1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ >>>>>> Kf7 6.Nxb7 Rc7 7.Nd6+ Kg7 8.Ne8+ Rxe8 9.b3 Rxc1+ >>>>>> (31.352.209) 93.3 >>>>>>11.01 9:18.76 -0.35-- 1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ >>>>>> Kf7 6.Nxb7 Rc7 7.Nd6+ Kg7 8.Ne8+ Rxe8 9.Qxe8 Nf6 >>>>>> 10.Qb8 Bd6 (54.470.299) 97.4 >>>>>>11.01 14:41.53 -1.36 1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ >>>>>> Kf7 6.Nxb7 Rc7 7.Nxa5 Rxd7 8.Nc6 Rc7 9.Nxe7 >>>>>> (86.129.099) 97.7 >>>>>>11.02 14:52.57 -1.35++ 1.Bh6 a6 2.Be2 Qa5 3.b3 Nf4 4.Bxf4 Bxg2 5.Qxg2 >>>>>> (87.133.005) 97.6 >>>>>>11.02 15:07.65 -1.00++ 1.Bh6 a6 2.Bxd7+ Qxd7 3.Nxe6 Qxa4 4.Ng7+ Kd8 5.Qd7+ >>>>>> Qxd7 (88.458.785) 97.4 >>>>>>11.02 16:23.21 -0.10 1.Bh6 a6 2.Bxd7+ Qxd7 3.b3 e5 4.Qxd7+ Kxd7 5.Rxe5 Bf6 >>>>>> 6.Rxd5+ Bxd5 7.Nb6+ Kd6 8.Nxc8+ Kd7 9.Re1 >>>>>> (95.283.049) 96.9 >>>>>> >>>>>>(I have formatted it a bit, hopefully it helps...) >>>>>> >>>>>>Stefan >>>>> >>>>> >>>>>I had a chance to study this... Crafty likes Re6 from about one second on. It >>>>>isn't a tactical thing, it is more of a positional issue. I let it search to >>>>>15+ plies and the score didn't change much, being right around zero. I didn't >>>>>get any fail low, but in looking at your analysis, I think your fail low is >>>>>also a positional issue where you suddenly don't like the position after you see >>>>>deep enough, even though it isn't losing or winning material. >>>>> >>>>>The move may or may not be good... Crafty sees it as materially equal, as a >>>>>piece and 2 pawns for the rook. The position is unbalanced, which it >>>>>particularly seems to favor against humans... >>>> >>>>1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ Kf7 6.Nxb7 Rc7 >>>>is winning material for black(black has rook for knight and a pawn and not rook >>>>for knight and 2 pawns) >>>> >>>>Crafty17.10 admits that the final position in this line is clearly better for >>>>black and not close to equality. >>>> >>>>I do not see a logical way for white to avoid this line except 2.Bh6 and >>>>I guess that crafty does not want to play this move in the main line. >>>> >>>>I read that modified Crafty found 2.Bh6 after a long time but I do not trust >>>>this version and I guess that it can find also wrong sacrifices. >>>> >>>>Uri >>> >>> >>>Here is what I see, search-wise: >>> >>> 7 5.11 -0.19 1. Rxe6 fxe6 2. Nxe6 Qa5 3. Bxd7+ Kxd7 >>> 4. Nec5+ Ke8 5. Qd7+ Kf7 6. Nxb7 >>> 7-> 5.11 -0.19 1. Rxe6 fxe6 2. Nxe6 Qa5 3. Bxd7+ Kxd7 >>> 4. Nec5+ Ke8 5. Qd7+ Kf7 6. Nxb7 >>> 8 6.07 -0.19 1. Rxe6 fxe6 2. Nxe6 Qa5 3. Bxd7+ Kxd7 >>> 4. Nec5+ Ke8 5. Qd7+ Kf7 6. Nxb7 Rc7 >>> 8-> 7.47 -0.19 1. Rxe6 fxe6 2. Nxe6 Qa5 3. Bxd7+ Kxd7 >>> 4. Nec5+ Ke8 5. Qd7+ Kf7 6. Nxb7 Rc7 >>> 9 13.55 ++ 1. Rxe6!! >>> 9 34.80 0.26 1. Rxe6 Rc7 2. Re4 O-O 3. Bh6 Re8 4. >>> Rd1 a6 5. Nc6 Bxc6 6. Bxc6 Rxc6 7. >>> Rxd5 >>> 9-> 35.80 0.26 1. Rxe6 Rc7 2. Re4 O-O 3. Bh6 Re8 4. >>> Rd1 a6 5. Nc6 Bxc6 6. Bxc6 Rxc6 7. >>> Rxd5 >>> 10 46.26 0.35 1. Rxe6 Rc7 2. Re2 O-O 3. Bh6 Re8 4. >>> Rc1 Rxc1+ 5. Bxc1 f5 6. Nc6 Qc7 7. >>> Nxe7+ Rxe7 8. Rxe7 Nxe7 >>> 10-> 48.89 0.35 1. Rxe6 Rc7 2. Re2 O-O 3. Bh6 Re8 4. >>> Rc1 Rxc1+ 5. Bxc1 f5 6. Nc6 Qc7 7. >>> Nxe7+ Rxe7 8. Rxe7 Nxe7 >>> 11 1:15 0.31 1. Rxe6 Rc7 2. Re2 a6 3. Bxd7+ Qxd7 >>> 4. Qxd7+ Kxd7 5. Bh6 Rd8 6. Rd1 Kc8 >>> 7. Re5 Kb8 8. f3 >>> 11-> 1:49 0.31 1. Rxe6 Rc7 2. Re2 a6 3. Bxd7+ Qxd7 >>> 4. Qxd7+ Kxd7 5. Bh6 Rd8 6. Rd1 Kc8 >>> 7. Re5 Kb8 8. f3 >>> 12 3:20 0.15 1. Rxe6 Rc7 2. Re1 a6 3. Bd3 O-O 4. >>> Bh6 Re8 5. Rad1 Bd6 6. Qf3 Rxe1+ 7. >>> Rxe1 Qf6 >>> 12-> 4:28 0.15 1. Rxe6 Rc7 2. Re1 a6 3. Bd3 O-O 4. >>> Bh6 Re8 5. Rad1 Bd6 6. Qf3 Rxe1+ 7. >>> Rxe1 Qf6 >>> 13 8:11 0.17 1. Rxe6 Rc7 2. Re1 O-O 3. Bh6 Re8 4. >>> Rad1 Bf6 5. Nb3 Rxe1+ 6. Rxe1 Bc6 7. >>> Bxc6 Rxc6 8. Qf3 Ne5 >>> 13-> 10:45 0.17 1. Rxe6 Rc7 2. Re1 O-O 3. Bh6 Re8 4. >>> Rad1 Bf6 5. Nb3 Rxe1+ 6. Rxe1 Bc6 7. >>> Bxc6 Rxc6 8. Qf3 Ne5 >>> >>> >>>In your analysis, doesn't Rxe6 Nxe6 pick up two pawns, then the bishop on b7? >>>Which is even in material. >> >>White is a pawn down in the original position and earning 2 pawns and a bishop >>for a rook should not be enough for equality. > >That isn't the point for this _move_. IE the move Re6 (or Ne6) doesn't lose >material itself, because in the case of Re6 it is two pawns and bishop for >the rook. Whether there are other better moves is another thing altogether. >But a program can like Re6 simply on positional grounds, ignoring the fact >he is down a pawn before he plays the move... The evaluation of Rxe6 is positive. The reason that crafty gives positive evaluation to Rxe6 is tactical and not positional. crafty can see that Rxe6 win material because it cannot see deep enough It wins a pawn in the main line when black replies Rc7 and it wins even 2 pawns in the line 1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ Kf7 6.Nxb7 Rc7 7.Nd6+ Kg7 8.Ne8+ Rxe8 9.Qxe8 Nf6 10.Qb8 if you do not extend this line. If you give crafty the position after 1.Rxe6 fxe6 then it is going to see a positive evaluation of more than 1 pawn if you do not give it a long time. > > > >> >>I was wrong when I said that Rxe6 fxe6 is winning material for black but >>trading pieces help the side with the material advantage. >> >>The problem of chess programs is that they believe that Rxe6 fxe6 is good for >>white(this is the reason for the Rc7 in your main line). >> >> >>They like the line >> >>1.Rxe6 fxe6 2.Nxe6 Qa5 3.Bxd7+ Kxd7 4.Nec5+ Ke8 5.Qd7+ Kf7 6.Nxb7 Rc7 7.Nd6+ Kg7 >>8.Ne8+ Rxe8 9.Qxe8 Nf6 10.Qb8 only because of the fact that white is a pawn up >>at the end of the line without understanding that black can win material after >>that. >> >>Uri > > >Note that Crafty didn't think fxe6 would be played. The reason that it dis not think that fxe6 is a good move is probably because of the line that I posted. 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.