Author: Peter McKenzie
Date: 19:09:07 05/12/00
Go up one level in this thread
On May 12, 2000 at 16:21:01, blass uri wrote: >[D]1r3nk1/3rb1p1/p2p1pP1/1p2p3/Pnq1P1R1/2N1B3/1PPR3P/2NQ3K w - - 0 1 > >This position is from the ssdf games(Junior5-Junior6). >The evaluation was close to equality and suddenly Junior5 failed high and found >more than 1 pawn advantage for white. > >I post this position to demonstrate the fact that tactics is not only >sacrifices. > >Unfortunately tactical test positions are usually sacrifices. > >I am also not sure if there is only one good move for white. >White played in the game Rg2. >This is a good move with the idea to give mate by Rh4 and Qh5 but I am not sure >if this is the only good move and it is possible that axb5 and after it Rg2 also >give the same result(I did not check it) > >The point is not to find Rg2 but to find a significant change in the evaluation. > >How much do programs need to fail high? Interesting position. Here is the output for my program LambChop (version 9.7) running on my Pentium 133mhz: 1 52 0 49 a4b5 a6b5 2 52 0 265 a4b5 a6b5 3 60 0 867 a4b5 a6b5 c1d3 b4d3 d2d3 4 41 100 4528 a4b5 a6b5 c1d3 f8e6 d3b4 c4b4 5 53 100 9425 a4b5 a6b5 c1d3 f8e6 g4h4 b4d3 d2d3 6 58 300 24294 a4b5 a6b5 d1f3 b8a8 f3h3 a8a1 7 44 1400 140110 a4b5 a6b5 c1d3 f8e6 g4g2 b4d3 c2d3 c4c6 8 49 3900 376243 a4b5 a6b5 c1d3 e7d8 d3b4 c4b4 d1b1 f8e6 b1a2 9 61 9600 944074 a4b5 a6b5 c1d3 b8a8 d3b4 c4b4 d1c1 f6f5 c3d5 b4d2 c1d2 f5g 4 10 85 60300 5920397 d2g2 d6d5 g4h4 d5e4 d1h5 c4f1 e3g1 f8g6 h5h7 g8f7 h7g6 f7e 6 c1e2 e4e3 a4b5 a6b5 10 85 61600 6050237 d2g2 d6d5 g4h4 d5e4 d1h5 c4f1 e3g1 f8g6 h5h7 g8f7 h7g6 f7e 6 c1e2 e4e3 a4b5 a6b5 11 85 159600 15735517 d2g2 d6d5 g4h4 d5e4 d1h5 c4f1 e3g1 f8g6 The 5 columns show iteration, score (100ths of pawns), time (100ths of seconds), nodes and PV. So it switched from axb5 to Rdg2 during iteration 10, after 603 seconds (about 10minutes). Looking at the PV, it seems it has the correct 'idea' but I can't say for sure if it would stay with Rdg2 at iteration 12. I plan to leave it running on the position overnight, hopefully it will finish iteration 12 and maybe 13. > >I think it is a good idea to generate a tactical test suite based on positions >from practical games and not based on finding sacrifices. > >In order to do it we need to analyze many games and find cases when there is a >significant change in the evaluation. > > >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.