The computer unsolvable position…

[Damir]

The computer unsolvable position

http://www.solidfiles.com/v/LKj2gd84zxgkv

Recent discoveries show that certain chess positions can be immediately
resolved (within only a few seconds) by any human brain (because of
intuition or pattern recognition), while the latest strongest computer based
chess software are totally unable to solve the position, even when we allow
them countless hours of calculation time.
Any software thinks the following position is won for black.
Yet, any chess player knows that it is not won for black, almost instantly, by
using pattern recognition and intuition.
The key of this position is that white simply has to first move his knight c3
away (to d5 or e4 or d1) and then move his king around (and nothing else!)
and black will NEVER be able to win.
The c6 pawn is blocking the complete black possibilities to liberate enough
material to allow mating white’s lonely king! Also, of course cxb5 and bxa4
instantly win for black, we see and know this instantly.

Maybe someday, faster, stronger quantum computers will perhaps be able to
solve this kind of unsolvable type of position. Or maybe not. And then, other
types of unsolvable solutions will be found by humans.
The key is that the human creativity and intuition will alway slead the
movement of artificial intelligence. These human abilities will always be the
key of the uniqueness of mankind.
Its power for creativity, its consciousness, and its intuition. No scientist can
define or explain exactly how and why the human consciousness is working.
A human brain is capable of solving some tasks in a second, whereas any
computer would take hours of computation.

[Dicaste]

Fen would be better instead of image.

[Ajedrecista]

Hello:

Fen would be better instead of image.

I agree with you:

[d]8/p7/kpP5/qrp1b3/rpP2b2/pPN3b1/P3K3/8 w - - 0 1

The source that can be read in the image is 'The Hippo System' by Eric Briffoz (2018).

Regards from Spain.

Ajedrecista.

[Ras]

The key of this position is that white simply has to first move his knight c3 away

Not even that. Just not doing any capture is already fine.

[yurikvelo]

Depend what to claim as "solved' or 'unsolved'.

Classical explanation for 'solved' is providing best moves for either side.

Within milliseconds engine provide draw sequence for white and never suggest lose move for defending side.
Any non-draw moves for white are 2000 cp lower than draw moves instantly. After 10 seconds any non-draw moves are resolved with DTM metric.

Modern explanation for term "solved position' is DTM/DTZ-adjudication. We expect either DTM metrics if position is not draw, or 0.00 value if position is draw. If engine provide best moves, but do not provide DTM or 0.00 - nowdays we call it "non-solved"

Not even that. Just not doing any capture is already fine.

Code: Select all

-15,52	1.Nd5 Bg5 2.Kf3 Bb8 3.Kg4 B3h4
-15,89	1.Nd1 Bd4 2.Kf3 Bb8 3.Kg4 Bbc7
-15,60	1.Nb1 Bc3 2.Kd3 Bg5 3.Ke4 Bg7
-15,60	1.Ne4 Bh2 2.Nf2 Bc3 3.Nh1 Bd4
-M13	1.Nxb5 Qxb5 2.cxb5+ Kxb5 3.c7 Bxc7
-M13	1.cxb5+ Qxb5+ 2.Nxb5 Kxb5 3.c7 Bxc7
-M13	1.c7 Bxc7 2.Nd5 Kb7 3.Kf3 Qa6
-M12	1.Nxa4 Qxa4 2.bxa4 b3 3.cxb5+ Ka5
-M12	1.bxa4 Bxc3 2.axb5+ Qxb5 3.cxb5+ Kxb5
-M09	1.Kf3 bxc3 2.Kg4 c2 3.bxa4 Qd2
-M09	1.Kd3 bxc3 2.Ke4 c2 3.bxa4 Qe1+
-M09	1.Kf1 bxc3 2.c7 Bxc7 3.Ke2 c2

[Ras]

-M09 1.Kf3 bxc3 2.Kg4 c2 3.bxa4 Qd2

Ooops, I only looked at bishop captures. Sure, the pawn capture breaks free.

[Jouni]

Unsolvable are you joking????

Analysis by Crystal 140121:

1. = (0.00): 1.Nd5 Bc1 2.Kd3 Bf6 3.Ke2
2. = (0.00): 1.Ne4 Bh2 2.Nf2 Bhg3
3. = (0.00): 1.Nb1 Bb8 2.Kf3 Bbe5
4. = (0.00): 1.Nd1 Bg7 2.Kd3 Bc7 3.Ke2
5. -+ (-#14): 1.c7 Kb7 2.Nd5 Qa6 3.Kf3 Rba5 4.Kg2 Bg5 5.c8Q+ Kxc8 6.Kh3 Qb7 7.Kg4 Bc1 8.Kf3 Qf7+ 9.Ke4 Kd8 10.bxa4 Qf2 11.Nxb4 cxb4 12.c5 Rxc5 13.a5 Rc4+ 14.Kd3 Qf1#
6. -+ (-#14): 1.Nxb5 Qxb5 2.cxb5+ Kxb5 3.c7 Bxc7 4.Kd3 Ra6 5.Ke4 c4 6.bxc4+ Kxc4 7.Kf5 b3 8.axb3+ Kxb3 9.Kg6 b5+ 10.Kf7 a2 11.Ke7 a1Q 12.Kd7 Qh8 13.Ke7 Qg7+ 14.Ke8 Re6#
7. -+ (-#14): 1.cxb5+ Qxb5+ 2.Nxb5 Kxb5 3.c7 Bxc7 4.Kd3 Ra6 5.Ke4 c4 6.bxc4+ Kxc4 7.Kf5 b3 8.axb3+ Kxb3 9.Kg6 b5+ 10.Kf7 a2 11.Ke7 a1Q 12.Kd7 Qh8 13.Ke7 Qg7+ 14.Ke8 Re6#
8. -+ (-#12): 1.bxa4 Bxc3 2.cxb5+ Qxb5+ 3.axb5+ Kxb5 4.Kd3 b3 5.c7 Bxc7 6.axb3 a2 7.Ke3 a1Q 8.Ke2 Qe1+ 9.Kf3 Qf1+ 10.Ke3 c4 11.bxc4+ Kxc4 12.Ke4 Qd3#
9. -+ (-#11): 1.Nxa4 Qxa4 2.bxa4 Ra5 3.Kd3 b3 4.axb3 a2 5.Ke4 a1Q 6.Kf5 Qb1+ 7.Kg4 Qe4 8.b4 cxb4 9.c7 Bxc7 10.c5 Be1 11.c6 Qg2#
10. -+ (-#9): 1.Kd3 bxc3 2.Ke4 c2 3.bxa4 Qxa4 4.c7 Qxc4+ 5.Kf5 Qf7+ 6.Ke4 Rb4+ 7.Kf3 Bh6+ 8.Kg2 Qf2+ 9.Kh3 Rh4#
11. -+ (-#9): 1.Kf3 bxc3 2.Ke4 c2 3.bxa4 Qxa4 4.c7 Qxc4+ 5.Kf5 Qf7+ 6.Ke4 Rb4+ 7.Kf3 Bh6+ 8.Kg2 Qf2+ 9.Kh3 Rh4#
12. -+ (-#9): 1.Kf1 bxc3 2.Kg2 c2 3.Kh3 c1Q 4.c7 Kb7 5.Kg4 Qf1 6.bxa4 Kxc7 7.Kf5 Bc1+ 8.Ke6 Qf6+ 9.Kd5 Qc6#
13. -+ (-#8): 1.Kd1 bxc3 2.Ke2 c2 3.Kf3 c1Q 4.Kg4 Bh6 5.Kf5 Qf4+ 6.Ke6 Qf6+ 7.Kd7 Qd2+ 8.Ke8 Qe6#

[Chessqueen]

I never heard of this chess Engine to solve chess problems before ==> The last version (4.59) of the chess problem solving program Popeye is now available! ==>
http://kobulchess.com/en/news/7-2011-07 ... --459.html

[Vinvin]

This unsolvable position is solved in 0.207 seconds by Crystal3.1 (1 core, old i7).

Code: Select all

Crystal3.1-sse4-popcnt:
NNUE evaluation using nn-62ef826d1a6d.nnue enabled
...
 10/21	00:00	 121 402	1 364 067	-9,09	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.c7 Bxc7 5.cxb5+ Kxb5 6.Nxc7+ Kc6
 11/16	00:00	 265 084	1 280 599	 0,00	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.Kd1 Bf4 5.Nxf4
 12/10	00:00	 279 204	1 286 654	 0,00	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.Kd1 Bf4 5.Nxf4
 13/10	00:00	 292 841	1 278 781	 0,00	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.Kd1 Bf4 5.Nxf4
 14/10	00:00	 318 807	1 285 512	 0,00	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.Kd1 Bf4 5.Nxf4
 15/10	00:00	 666 409	1 139 160	 0,00	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.Kd1 Bf4 5.Nxf4
 16/22	00:00	 732 183	1 160 353	 0,00	1.Nd5 Be1 2.Kxe1 Bd2+ 3.Kxd2 Bh2 4.Kd1 Bg3 5.Ke2 Bh2

2021-04-24 00:40:14,310<--1:info depth 10 seldepth 21 multipv 1 score cp -909 nodes 121402 nps 1364067 tbhits 0 time 89 pv c3d5 g3e1 e2e1 f4d2 e1d2 e5h2 c6c7 h2c7 c4b5 a6b5 d5c7 b5c6
2021-04-24 00:40:14,425<--1:info depth 11 seldepth 16 multipv 1 score cp 0 nodes 265084 nps 1280599 tbhits 0 time 207 pv c3d5 g3e1 e2e1 f4d2 e1d2 e5h2 d2d1 h2f4 d5f4
2021-04-24 00:40:14,435<--1:info depth 12 seldepth 10 multipv 1 score cp 0 nodes 279204 nps 1286654 tbhits 0 time 217 pv c3d5 g3e1 e2e1 f4d2 e1d2 e5h2 d2d1 h2f4 d5f4

...

[d]8/p7/kpP5/qrp1b3/rpP2b2/pPN3b1/P3K3/8 w - - 0 1

The source that can be read in the image is 'The Hippo System' by Eric Briffoz (2018).

Regards from Spain.

Ajedrecista.

[MikeB]

BD , which is based on Crystal, using two threads .... -> 0.04 seconds ..., the single thread was also faster, but only due to hardware, the single thread node count was higher by a few thousand over Crystal, BD also uses a different net here


28 0.00 137.7M 0:20.08 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
27 0.00 102.9M 0:14.98 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
26 0.00 48.4M 0:07.16 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
25 0.00 31.0M 0:04.57 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
24 0.00 22.6M 0:03.34 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
23 0.00 19.4M 0:02.90 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
22 0.00 13.1M 0:02.00 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
21 0.00 12.0M 0:01.84 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
20 0.00 4.46M 0:00.64 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
19 0.00 3.63M 0:00.52 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
18 0.00 2.08M 0:00.30 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
17 0.00 1.93M 0:00.28 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
16 0.00 1.45M 0:00.21 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
15 0.00 1.02M 0:00.15 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
14 0.00 721103 0:00.11 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
13 0.00 597613 0:00.09 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
12 0.00 381879 0:00.06 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
11 0.00 245631 0:00.04 Nd1 Bb8 Kd3 Be1 Ke4 Bh2 Kd3
10 -9.05 51881 0:00.01 c7 Bxc7 cxb5+ Qxb5+ Nxb5 Kxb5 bxa4+ Kxa4 Kd3 Kb5 Ke4 a5 Kf5 c4 Ke4
9 -8.43 32053 0:00.01 c7 Bxc7 bxa4 bxc3 axb5+ Qxb5 cxb5+ Kxb5 Kd3 Kb4
8 -8.79 20494 0:00.01 Nxb5 Qxb5 cxb5+ Kxb5 bxa4+ Kxa4 Kd3 Kb5 Kc2 c4 Kd1 Kxc6 Kc2
7 -8.63 6250 0:00.00 Nxb5 Qxb5 cxb5+ Kxb5 bxa4+ Kxa4 Kd3 Kb5 Kc2 Kxc6
6 -8.52 4809 0:00.00 Nxb5 Qxb5 cxb5+ Kxb5 bxa4+ Kxa4 Kd3 a5 Kc4
5 -8.46 4037 0:00.00 Nxb5 Qxb5 cxb5+ Kxb5 bxa4+ Kxa4 Kd3 a5
4 -8.82 542 0:00.00 cxb5+ Qxb5+ Nxb5 Ra5 c7 Bxc7 Nxc7+ Bxc7
3 -8.72 260 0:00.00 cxb5+ Qxb5+ Nxb5 Ra5
2 -8.85 158 0:00.00 cxb5+ Qxb5+ Nxb5
1 NNUE evaluation using eval.bin enabled
1 -8.85 85 0:00.00 cxb5+ Qxb5+ Nxb5