Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: hard test position for a computer I think

Author: Stan Arts

Date: 03:48:56 10/18/03

Go up one level in this thread


>[D]r2q1k1r/p1n1pp1p/1pp3p1/2p1b3/4N1b1/3P1N2/PPP2PPP/R1BQR1K1 w - - 0 13
>
>This was a position I got (unfoutunately I was black). White finished in style
>with Nxe5 Bxd1 Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7, and black is totally
>lost. Can any computers get this (I don't know if it's hard or a piece of cake),
>but while my program has no problem seeing all the pretty tactics. It evaluates
>the position at the end as clearly good for black (black can take on c2 and then
>d3, and be two pawns up)
>
>have fun.

Hello

Neurosis 1.4 (next version) on slow PIII-550 mHz, 32 MB hash
Thinks

Legal moves: 35
Ply Score Time     Nodes      Best move and expected line
------------------------------------------------------------
2   -83   6        1081       h3 Bh5
3   -74   6        1885       h3 Bxf3 Qxf3
4   -80   17       9431       h3 Bxf3 Qxf3 Kg8
5   -73   94       85697      h3 Bxf3 Qxf3 Qd4 Kh1
6   -74   264      185582     h3 Bxf3 Qxf3 Qd4 Bh6+ Kg8 Ra1-b1
7   -71   643      402790     h3 Bxf3 Qxf3 Bg7 c3 Kg8 Bf4
8   -74   2620     1594910    h3 Bxf3 Qxf3 Bg7 c3 Kg8 Bg5 Nd5
9   -73   8272     5489353    h3 Bxf3 Qxf3 Bg7 Ng5 Qd5 Qe2 e5 Bd2
10  -77   32741    21301007   h3 Bxf3 Qxf3 Bg7 Rb1 Kg8 Bg5 f5 Nd2 Nd5
11  -74   111296   68393957   h3 Bxf3 Qxf3 Bg7 Ng5 Qd5 Re4 Rc8 Qe2 Bf6 c4
Total nodes:245686375 n/sec:67220  (Q-nodes:29278159 max depth:23)

Then, after forcing Nxe5, and Bxd1, it sees the combination you described
but thinks the same as your program, that black will be up atleast 2 pawns

Legal moves: 39
Ply Score Time     Nodes      Best move and expected line
------------------------------------------------------------
2   -367  0        871        Bh6+ Kg8 Ra1xd1 Qd5 Nxg6 hxg6 Nxc5
3   -367  0        990        Bh6+ Kg8 Ra1xd1 Qd5 Nxg6 hxg6 Nxc5
4   -368  11       7536       Bh6+ Kg8 Ra1xd1 Qd4
5   -309  55       34003      Bh6+ < + > Ke8 Nxc6 Nd5
5   -286  55       38144      Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7
6   -286  165      100590     Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7
7   -286  412      249360     Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7
8   -286  1440     995841     Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7
9   -286  4219     2841899    Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7
10  -289  27771    20586791   Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7 Nd5
11  -293  95813    65893354   Bh6+ Kg8 Nxc6 Qd7 Nf6+ exf6 Ne7+ Qxe7 Rxe7 Nd5 Rd7
Total nodes:84670941 n/sec:70859  (Q-nodes:6996659 max depth:22)

But I've also run Delfi 4.3 with 24 MB hash.

3  -34  0     961      e4xc5 e5xh2 g1xh2 b6xc5
4  -74  2     3001     e4xc5 e5xh2 g1xh2 d8d6 h2g1 d6xc5
4  -55  5     7338     c1h6 f8g8 a1b1 d8d5 c2c4
5  -57  15    25033    c1h6 e5g7 h6xg7 f8xg7 d1e2 g4xf3 e2xf3 c7e6
5  -55  21    36092    e4xc5 e5xh2 g1xh2 d8d6 h2g1 d6xc5 c1h6 f8g8
5  -52  26    47422    h2h3 g4xf3 d1xf3 d8d4 c1h6 f8e8
6  -49  39    72214    h2h3 g4xf3 d1xf3 d8d4 c1h6 f8e8 c2c3
7  -48  92    180005   h2h3 g4xf3 d1xf3 d8d5 c1h6 f8e8 c2c4 d5d4
8  -42  222   470528   h2h3 g4xf3 d1xf3 d8d5 c1h6 e5g7 c2c4 d5h5 f3f4 h5xh6
f4xc7 g7xb2 c7xc6 b2xa1 c6xa8
9  -38  426   960000   h2h3 g4xf3 d1xf3 e5g7 c2c4 d8d7 a1b1 c7e6 c1g5
10 -47  903   2123241  h2h3 g4xf3 d1xf3 e5g7 c2c3 d8d5 c1f4 c7e6 c3c4
11 -50  3655  8497325  h2h3 g4xf3 d1xf3 e5g7 c2c3 d8d5 f3g3 c7e6 e4g5 e6xg5
c1xg5 e7e5
12 -52  10518 23875408 h2h3 g4xf3 d1xf3 e5g7 e4c3 d8d7 c1g5 c7e6 g5e3 a8d8 a1d1
g7e5
13 -53  33411 76843683 h2h3 g4xf3 d1xf3 e5g7 e4g5 d8d5 f3g3 e7e5 g5e4 c7e6 c1g5
d5d4 c2c3

And after forcing Nxe5 and Bxd1, it doesn't think it's winning, but does
quickly give a better score then Neurosis and thinks black will be up about 1
pawn, and searching longer the score seems to be getting better (for white)
but not positive yet.

3  -350 1     1310     c1h6 f8g8 e1xd1 d8d5
4  -326 2     4453     c1h6 f8g8 a1xd1 d8d5 e4c3
5  -285 5     9206     c1h6 f8g8 e5xc6 d8d7 e4f6 e7xf6 c6e7 d7xe7 e1xe7
6  -287 9     22152    c1h6 f8g8 e5xc6 d8d7 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
7  -318 29    67860    c1h6 f8g8 e5xc6 d8d7 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7
7  -306 79    164148   e4g5 d1xc2 g5xf7 d8d5 f7xh8 f8g7 h8f7 c2xd3
8  -350 151   331703   e4g5 f7f6 g5f7 d8e8 f7xh8 d1xc2 e1e2 f8g8 e2xc2 g8xh8
8  -121 173   408060   c1h6! f8g8 e5xc6 d8d7 e4f6 e7xf6 c6e7 d7xe7 e1xe7 d1xc2
e7xc7 c2xd3 a1e1
9  -140 237   595476   c1h6 f8g8 e5xc6 d8e8 e4f6 e7xf6 c6e7 e8xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8
10 -140 387   1115473  c1h6 f8g8 e5xc6 d8e8 e4f6 e7xf6 c6e7 e8xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8
10 -125 436   1251765  e5xc6..
10 -112 500   1460768  e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d5d7
11 -112 671   1992374  e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d5d7
12 -119 1180  3574049  e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d5d7 e8e2 d7xa7 c2xd3
13 -105 2996  9198897  e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d5d7 a7a6 a1c1
14 -110 7389  21589536 e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d5d7 a7a6 d3d4 c2f5
15 -94  19864 57147022 e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d5d7 a7a6 d7d6 f6f5 d6xb6 c2xd3
16 -89  49743 140976041 e5xc6 d8d7 c1h6 f8g8 e4f6 e7xf6 c6e7 d7xe7 e1xe7 c7d5
e7d7 d1xc2 d7xd5 a8e8 d3d4 c5xd4 d5xd4 c2f5 a1d1 e8e2

Greetings
Stan



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.