Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Another example of very deep tactics that Computers cannot solve

Author: Dann Corbit

Date: 18:03:20 12/07/01

Go up one level in this thread


On December 07, 2001 at 12:48:22, ALI MIRAFZALI wrote:

>The following position occured in one of the oldest Correspondence games
>played in America:
>[D] 2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1  b   - 0 1
>This game was played in 1840.
>Here black played 15...Rxd2!! Game finished with 16.Qxd2 Nxe4 17.Qc1 Bf3
>18.g3 h5! 19.Bd5 h4 20.Bxe4 Qxe4 21.Bxb6 Qg4 22.Qe3 axb6 23.R(f)-b1  e4
>24.Qe1 f5! 25.Rb2 f4 26.R(a)-b1  Black announced Mate in Some number of moves.
>Fritz 6 and chessmaster 8000 donot see 15...Rxd2!! No matter how long one runs
>the position.
>Does your program see 15...Rxd2!!!.Althuogh Tactics is the area that Computers
>are naturally strong ;it is interesting that one still runs into tactical
>positions which programs cannot solve.Please indicate your result.

Quite an interesting position...
[D] 2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - -

It appears that Quark is the only engine to carefully consider Rxd2:
position 1
 5    102       5      16684  b6e3 f2e3 g6g5 a1e1 c7c6
 5    102       6      21481  b6e3 f2e3 g6g5 a1e1 c7c6
 6     78      10      37153  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1
 6     78      14      51152  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1
 7     96      20      78321  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1 h6f6
 7     96      32     132732  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1 h6f6
 8     84      61     250523  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1 h6c6 c4d5
 8     84      88     385576  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1 h6c6 c4d5
 9     81     424    1576731  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1 h6g5 a2a4 g5h4
 9     81     597    2364464  b6e3 f2e3 g6h6 f1e1 f6g4 d2f1 h6g5 a2a4 g5h4
10     93    1252    4874575  b6e3 f2e3 g6g5 f1e1 d8d6 a2a4 h8d8 d2f1 d6c6 c4d3
10     93    1411    5598432  b6e3 f2e3 g6g5 f1e1 d8d6 a2a4 h8d8 d2f1 d6c6 c4d3
11    111    2236    8933072  b6e3 f2e3 g6g5 f1e1 d8d6 c4f1 f6g4 d2c4 d6f6 c2a4
a7a6
11    111    2744   11187292  b6e3 f2e3 g6g5 f1e1 d8d6 c4f1 f6g4 d2c4 d6f6 c2a4
a7a6
12    109    5143   21088556  b6e3 f2e3 d8d7 f1f2 h8d8 h2h3 g6g3 d2f1 g3h4 g2g4
h5g6 c4b5
12    109    6579   27904369  b6e3 f2e3 d8d7 f1f2 h8d8 h2h3 g6g3 d2f1 g3h4 g2g4
h5g6 c4b5
13    118   10994   46376265  b6e3 f2e3 d8d7 h2h3 g6g5 a1e1 g5g3 g1h1 h8d8 d2b3
c7c6 a2a4 h7h6
13    119   15729   70062407  d8d2!
13    122   16497   73654902  d8d2 c2d2 f6e4 d2d3 h8d8 d3b1 b6e3 f2e3 d8d2 g2g3
e4c3 b1g6 f7g6
13    122   18540   83286233  d8d2 c2d2 f6e4 d2d3 h8d8 d3b1 b6e3 f2e3 d8d2 g2g3
e4c3 b1g6 f7g6
14    108   25489  118020562  d8d2 c2d2 f6e4 d2d3 h8d8 d3b1 e4c3 b1g6 h5g6 e3b6
a7b6 a1c1 c3d5 f1e1

Although Quark eventually switches back to Bxe3


Beowulf:
Current Position = -0.50
Number of Possible Moves = 40

  6     73      69      59143   b6e3 f2e3 g6g5 g2g4 g5e3 g1h1
  7     88     143     112576   b6e3 f2e3 g6h6 a1e1 f6g4 d2f3 g4e3
  8     88     321     301998   b6e3 f2e3 g6h6 f1e1 h6g5 a2a4 c8b8 g2g3
  9    101     631     702259   b6e3 f2e3 g6g5 f1e1 c8b8 a2a4 d8d7 a1d1
 10     92    2057    2614250   b6e3 f2e3 g6g5 f1e1 d8d7 g1h1 h8d8 d2f3 h5f3
g2f3
 11    121!   4382    6327992   b6e3 f2e3 g6g5 f1e1 d8d7 g1h1 h8d8 a1d1
 11    121    4778    6862582   b6e3 f2e3 g6g5 f1e1 d8d7 g1h1 h5g6 d2f3 d7d1
a1d1 f6e4
 12    131   24383   31707509   b6e3 f2e3 g6g5 f1e1 d8d7 g1h1 h8d8 d2f3 h5f3
g2f3 d7d2 c2a4
 12    131   99900  114152522   b6e3 f2e3 g6g5 f1e1 d8d7 g1h1 h8d8 d2f3 h5f3
g2f3 d7d2 c2a4

Search Time : 999.0 Seconds
Total Nodes Searched   : 114152522  (89.5% Qui)
Total Nodes Evaluated  : 102160016
Cuts       : Delta 22748279  : SEE     22807388  : Mate          0
           : Razor  7443478  : Eval    55013059
Extensions : Check  2313832  : OneRep    83541  : CMThreat  26069
           : Pawn      2344  : Recap    321588  : RevCheck  68378
Hash Stores (Size) : 7870372  (5592373)
Hash Probes (Hits) : 62166883  (6031179)
EGTB Probes (Hits) : 0  (0)
Move Ordering      : 69.4%
Best Move = b6e3
[1]B >

Phalanx:
    -> easy move      (1)  Bb6xe3
  5 ->   0:00.74    98470   0 turns
  6    106   100   140314  Bb6xe3  Pf2xe3  Qg6-g5  Rf1-e1  Nf6-g4  Nd2-f1
                           Rh8-e8
  6 ->   0:01.47   215458   0 turns
  7     79   703   397895  Bb6xe3  Pf2xe3  Nf6-g4  Rf1-e1
  7 ->   0:21.08   889754   0 turns
  8     99  2551  1628993  Bb6xe3  Pf2xe3  Qg6-g5  Rf1-e1  Rh8-e8  Bc4-b5
                           Re8-e6  Nd2-f1  Kc8-b8
  8 ->   0:31.70  2689478   0 turns
  9     84  4098  4424999  Bb6xe3  Pf2xe3  Qg6-g5  Rf1-e1  Rd8-d6  Bc4-b5
                           Pa7-a6  Bb5-c4  Nf6-g4  Nd2-f1
  9 ->   1:01.44  8361639   0 turns
 10    101 10838 16723310  Bb6xe3  Pf2xe3  Qg6-g5  Rf1-e1  Rd8-d6  Bc4-b5
                           Pa7-a6  Bb5-c4  Nf6-g4  Nd2-f1  Rh8-d8
 10 ->   2:32.27 24687615   0 turns
 11    101 25825 44607004  Bb6xe3  Pf2xe3  Rd8-d7  Nd2-f3
 11 ->   6:18.19 66805774   0 turns

Chess Tiger 14:
2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - - ce 92; acd 15; acs
999; pv Bxe3 fxe3 Rhe8 Rae1 Re7 Be2 Bxe2 Rxe2 Red7 Nc4 Qxe4 Qxe4 Nxe4 Nxe5 Rd1
Nxf7 Nxc3;

Gambit Tiger II:
2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - - ce 114; acd 14; acs
999; pv Bxe3 fxe3 Rd7 Nf3 Re8 Qa4 a6 Bb5 c6 Bxc6 bxc6 Qxc6+ Kb8 Qb6+ Rb7 Qd6+
Ka7 Qc5+;

Little Goliath 3:
2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - - ce 124; acd 13; acs
999; pv Bxe3 fxe3 Rd7 Nf3 Bxf3 Rxf3 Qxe4 Qxe4 Nxe4 Rxf7 Rxf7 Bxf7 Nxc3 Be6+ Kd8
Bc4 Ke7;

Crafty:
2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - - ce 117; acd 15; acs
999; pv Bxe3 fxe3 Rd6 Bb3 Be2 Nc4 Bxc4 Bxc4 Nxe4 Bxf7 Qg4 c4 Rd2 Be6+ Qxe6 Qxe4
Kb8 a4;

ExChess:
2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - - ce 134; acd 14; acs
999; pv Bxe3 fxe3 Rd7 Nf3 Bxf3 Rxf3 Qxe4 Qxe4 Nxe4 Rxf7 Rxf7 Bxf7 Nxc3 Be6+ Kb8
Rf1 Rd8;

Pepito:
2kr3r/ppp2ppp/1b3nq1/4p2b/1PB1P3/2P1B3/P1QN1PPP/R4RK1 b - - ce 74; acd 15; acs
999; pv Bxe3 fxe3 Rd6 Bb3 Be2 Nc4 Bxc4 Bxc4 Rhd8 Rad1 Rxd1 Rxd1 Rxd1+ Qxd1 Qxe4
Qe2 Ng4 Bxf7 Nxe3 a4;


Of course, since this was a correspondence game, we should really let the
computers think overnight.

I strongly suspect that both Chess Tiger and Gambit Tiger will find the move
under those circumstances.  Both engines had a very long think on their last
iteration, so the evaluation between the chosen move and the key move was close
already.




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.