Author: leonid
Date: 07:46:03 09/03/00
Go up one level in this thread
On September 02, 2000 at 18:51:22, Eelco de Groot wrote: >A difficult position it turned out to be for the programs also, Leonid! >Not that the programs I tried cannot find a mate-solution... The programs I >tried are selective, a brute force approach could always give the shortest mate >but would probably also take longer unless it is specifically designed to find >mates. > >I hesitate to place this because it turns out more like some bug-report or else >I am doing several things wrong. I'm hoping this does not sound too negative: >when Rebel Century 2.0 for ECTool announces a mate the numbering of the moves >seems to be a little displaced. Not a giant bug. In Crafty's analysis mode I >can't get pawns to promote but maybe there I'm doing something wrong entering >the move. > >The Q5T personality for Rebel Century 2.0 in ECTool needs 1 minute 36 to >announce a mate. I think ChessMaster, aka The King, can beat this. Q5T is >similar to Q3, a selective personality and much too aggressive to do well in >games. But good in many tactical positions. This was on my Celeron 500MHz, 28Mb >transposition tables. Unfortunately the engine stops after announcing the mate >and only gives 4 half-moves of a mate in eight, you would like to see the full >mate sequence. > >Rebel Engine for ECTool. (c) Ed Schröder > >Engine version : REBEL CENTURY 2 >Rebel personality : q5t.eng >Opening book file : REBEL.MVS >Hash table size : 28 MB >Analysis mode : Analyzing next move >Refresh interval : 500 ms > >Game begin > >00:00 02.00 29.13 1.hxg8=Q+ Qf8 2.Nxe6+ Kc8 3.gxf8=Q+ Qxf8 4.Qxc7+ >00:00 02.06 29.21 1.Nxe6+ Kc8 2.hxg8=Q+ Qf8 3.gxh8=Q >00:01 03.01 29.54 1.hxg8=Q+ >00:01 03.01 31.82 1.hxg8=Q+ Qf8 2.Qxh8 Qh3+ 3.Qexh3 >00:10 04.00 32.69 1.hxg8=Q+ Qf8 2.Qxh8 Qh3+ 3.Qexh3 Qxh3+ 4.Qhxh3 >00:21 04.04 32.69 1.fxg8=Q+ >00:23 04.04 34.46 1.fxg8=Q+ Qee8 2.Nxc6+ Qdxc6 3.Qa8+ Qxa8 4.Qhh4+ >00:24 04.05 34.46 1.f8=Q+ >00:41 04.05 37.23 1.f8=Q+ Qde8 2.Nxe6+ Kc8 3.fxe8=Q+ Qcd8 4.exd8=Q+ >01:30 04.45 37.23 1.Nxc6+ >01:36 04.45 Mate in 1.8 moves 2.Nxc6+ Qdxc6 3.f8=Q+ Qee8 4.Qxc7+ Kxc7 > >Q3 personality from Eric Campos, winner of the Rebel contest for tactical >positions is fastest so far! Needs 41 seconds to announce a mate. >It gives a mate in 9 and stops looking further, many programs do this, I believe >the shortest mate is a mate in seven Leonid, with several first moves possible? >Different second move for Black here and in the given PV it seems to say that >Nxc6+ is the second move which I think is not correct, it's White's first move. > >Rebel Engine for ECTool. (c) Ed Schröder > >Engine version : REBEL CENTURY 2 >Rebel personality : q3.eng >Opening book file : REBEL.MVS >Hash table size : 28 MB >Analysis mode : Analyzing next move >Refresh interval : 500 ms > >Game begin > >00:00 01.00 19.68 1.fxg8=Q+ Qde8 2.Qxe6 Qxh2+ 3.Qxh2 Qcxe6 4.Nxe6+ >00:00 02.00 28.71 1.hxg8=Q+ Qf8 2.Nxe6+ Kc8 3.gxf8=Q+ Qxf8 4.Qxc7+ >00:01 03.00 31.13 1.hxg8=Q+ Qf8 2.Qxh8 Qh3+ 3.Qexh3 >00:10 04.00 31.96 1.hxg8=Q+ Qf8 2.Qxh8 Qh3+ 3.Qexh3 Qxh3+ 4.Qhxh3 >00:11 04.01 31.96 1.f8=Q+ >00:28 04.01 36.71 1.f8=Q+ Qde8 2.Nxe6+ Kc8 3.fxe8=Q+ Qcd8 4.exd8=Q+ >00:33 04.03 36.71 1.Nxc6+ >00:34 04.03 36.71 1.Nxc6+ >00:41 04.03 Mate in 1.9 moves 2.Nxc6+ Qdxc6 3.f8=Q+ Kd7 4.Qxc7+ Qxc7 5.Qa4+ > >Finally a possible bug in Crafty? At least I seem to be unable to promote a pawn >in "analyze" mode? Maybe I entered the moves in the wrong way.. I also tried a >simple position like 8/2Pk4/8/8/8/8/8/4K3 w - - and when I try to promote the >c-pawn Crafty 17-11 says it's an illegal move. > >Crafty finds a mate in seven in the many queens position and gives the full mate >sequence. Again on the Celeron 500, time is 2 min 48 seconds. > >White(1): analyze > clearing hash tables > time surplus 0.00 time limit 30.00 (3:00) > nss depth time score variation (1) > 4-> 1.69 32.25 1. hxg8=Q+ Qf8 2. Qhxh8 Qh3+ 3. Qhxh3 > Qxh3+ 4. Qgxh3 > 5 1.80 ++ 1. hxg8=Q+!! > 5 2.47 35.78 1. hxg8=Q+ Qf8 2. Qhxh8 Rh5+ 3. Qh4+ > Rxh4+ 4. Qxh4+ Qde7 5. Nxe6+ Kc8 6. > Qxe7 Qxg2+ 7. Qexg2 > 5-> 18.82 35.78 1. hxg8=Q+ Qf8 2. Qhxh8 Rh5+ 3. Qh4+ > Rxh4+ 4. Qxh4+ Qde7 5. Nxe6+ Kc8 6. > Qxe7 Qxg2+ 7. Qexg2 > 6 19.12 ++ 1. hxg8=Q+!! > 6 26.09 36.72 1. hxg8=Q+ Qf8 2. Qhxh8 Rh5+ 3. Qh4+ > Rxh4+ 4. Qxh4+ Kc8 5. Qa8+ Qxa8 6. > Qxa8+ Qb8 7. Qxf8+ Qd8 8. Qhxd8# > 6 38.17 ++ 1. Qxc7+!! > 6 2:17 39.36 1. Qxc7+ Kxc7 2. Nxe6+ Kc8 3. Nb6+ > Kb8 4. Qeg3+ Qe5 5. Qgxe5+ Qdc7 6. > Qxc7+ Qexc7 7. hxg8=Q+ Rd8 8. Nxc7 > Qxc7 9. f8=Q Rxf8 10. Qgxf8+ > 6 2:17 ++ 1. f8=Q+!! > 6 2:48 Mat07 1. f8=Q+ Qde8 2. Nxc6+ Qexc6 3. Qxe8+ > Qcxe8 4. Qxd5+ Q8d7 5. hxg8=Q+ Qxg8 > 6. Qaa8+ Qc8 7. Qxe7# > 6-> 4:02 Mat07 1. f8=Q+ Qde8 2. Nxc6+ Qexc6 3. Qxe8+ > > >Eelco > >The two positions: >[D]3k2nq/2qqqP1P/2q1q3/3r4/2NN4/4Q3/Q3QQQQ/7K w You are right! This position is inevitable mate in 7 with multiple solutions. I even not found all of them. You surpised me with Rebel, since yeasterday I tried on Rebel 10 and he found the mate. I not tried to see how he will make that mate up to the end. Today when I tried to do the same, I was not able to find mate on Rebel with yesterday time. Why? I have no idea. Computer AMD 400Mhz. Rebel 10. Hash table 28M. Brute force. 6 moves search for mate. Waited for 30 min. I stopped waiting. 7 moves search for mate. Waited for 30 min. I stopped waiting. Yesterday Rebel found mate by brute force in around 11 min. The same computer. Mate Solver in LLchess. No hash tables. Brute force. 5 moves - 41 sec. No mate found. 6 moves - 11 min. 58 sec. No mate found. 7 moves - 1 min 56 sec. Mate. White pawn is promoted into queen: f7-f8. next mate in 7 - 2 min 19 sec. White pawn promted into rook: f7-f8. Waited for the next mate for 12 minites and stopped. After branching factor (between move 5 and 6) search can go for around 4 hours to see everything. Selective search for 7 moves. One don't solve. Three others solve between 1 and 0.5 sec. Mate is done by white queen from a2-a8. Leonid. >[D]8/2Pk4/8/8/8/8/8/4K3 w - -
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.