Kornrade Test

[peter]

Hi Vincent!

Always 8 key move found.

Key move only doesn't give full solution in some or even most of the pieces.
Especially the cyclic zugzwang- ones and the fortresses are often "solved" for wrong reasons, as long as the output- lines together with correct evals don't fit.
This isn't a test suite to be used for simple bm 1-0- measurement.

[Dann Corbit]

I have some engines taking a peek at these positions over the weekend.
We'll see how well they did on Monday.

Engines used on moderately strong machines for one hour are: Stockfish (bleeding edge development version), Matefinder, McCain, Sting10 (crashed 2/3 of the way through), Houdini 6 Tactical 1.

Solved by move choice (05 has a bad score, but it stays put for many iterations):

Code: Select all

5k2/2p4p/6P1/1p3n2/3q4/PB4BN/KP5P/8 w - - acd 73; bm Ng5; c3 "Ng5"; ce 432; id "* Smothering II +-"; pm Ng5; pv Ng5; 
3k4/2pq3p/pp5R/4P3/P6P/2PN4/1PN3K1/8 w - - acd 64; bm Nd4; c1 "Hard"; c3 "Nd4"; ce 723; id "01 Constrictor  +-"; pm Nd4; pv Nd4; 
6Nk/pp2Np1p/2p2Pp1/2R2bP1/7K/P7/1q1n1Q2/5n2 w - - acd 69; bm Rxf5; c1 "diff=9"; c3 "Rxf5"; ce 1155; id "02 Mirror       +-"; pm Rxf5; pv Rxf5 gxf5; 
5B1q/3N3p/p3p2p/7k/B7/5PP1/K6P/8 w - - acd 56; bm Ne5; c3 "Ne5"; ce 32582; id "03 Walk         +-"; pm Ne5; pv Ne5 Qxf8 f4 Qd8 Bc6 Qd2+ Kb3 Qd1+ Kc3 Qe1+ Kc4 Qc1+ Kb4 Qb2+ Kc5 Qa3+ Kb6 Qd6 Kb7 Qe7+ Kc8 Qf8+ Kd7 Qg7+ Kxe6 Qg8+ Kf5 Qc8+ Ke4 Qd8 Ke3 Qb6+ Kf3 Qb3+ Kg2 Qc2+ Kh3 Qf5+ g4+ Qxg4+ Nxg4 Kg6 Be4+ Kf7 Bxh7 h5 Kh4 Ke7 Ne5 Kd8 Bd3 a5 Nc6+ Kc7 Nxa5 Kc8 Kxh5 Kd8 f5 Kd7 h3 Ke7 h4 Kf6 Nb3 Ke7 f6+ Kf7 Na1 Kg8 f7+ Kf8; 
8/b4kp1/2p1p3/5pB1/p1PP4/1p6/3K1P2/3N4 w - - acd 75; bm Kc3; c3 "Kc3"; ce -396; id "05 Stubbornness =="; pm Kc3; pv Kc3 Bxd4+ Kxd4 b2 Nxb2 a3 Nd3 a2 Ne5+ Ke8 f4 a1=Q+ Kc5 Qh1 Kd6 c5 Kxe6 Qd1 Bh4 Qb3 Bg5 Qa3 Kxf5 Qh3+ Ke4 Qe6 Ke3 Qf5 Kf3 Qb1 Kg4 Qd1+ Kg3 g6 Kh4 Qd6 Kg3 Qd4 Kf3 Qd1+ Ke3 Qh1 Ke2 Qg2+ Ke3 Qh3+ Kf2 Qh2+ Kf3 Qc2 Ke3 Kf8 Kf3 Qd1+ Kg3 Qd4 Kg4 Kg7 Kh3 Qe4 Kg3 Qe3+ Kg4 Qf2 Kh3 Qc2 Kg3 Qe4 Kg4 Qg2+ Kh4 Kf8 Bf6 Qh1+ Kg3 Qe1+ Kg4 Qe4 Kg3 Qe3+ Kg4 Qb3 Bh4 Ke8 Bg5 Qe3 Bh4 Qg1+ Kf3 Qd1+ Kg2 Qe2+ Kg3 Qc2 Bg5 Qe4 Kg4; 
q7/b1p5/kp1p4/p2PN3/PPP5/1K6/8/5B2 w - - acd 92; bm c5+; c1 "Hard"; c3 "c5+"; ce 1892; id "08 Slowness     +-"; pm c5+; pv c5+ Kb7 Ba6+ Kxa6 b5+ Kb7 c6+ Kc8 Nf7 Kb8 Kc4 Kc8 Kc3 Bb8 Kc2 Qa7 Kb3 Qa8 Kc4 Ba7; 
1N4K1/1pp1p3/7k/2r3pP/p2pp1P1/P6N/8/2B5 w - - acd 47; bm Nc6; c0 "http://www.chessvideos.tv/forum/viewtopic.php?t=10770"; c3 "Nc6"; ce 0; id "09 Standstill   =="; pm Nc6; pv Nc6 bxc6 Nxg5 Rxg5+ Kh8 e3 Bxe3 d3 Bf4 c5 Bc1 e5 Bf4 c4 Bc1 c3 Be3 e4 Bd2 cxd2; 
4B3/k1K1pp2/p3bq2/2p1p3/1p2P1P1/4P2Q/nR6/8 w - - acd 53; bm g5; c3 "g5"; ce 0; id "10 Shelter      =="; pm g5; pv g5 Qxg5 Qxe6 fxe6 Rxa2 Qg8 Ra4 Qxe8 Rxa6+ Kxa6; 
4N2k/4PB1p/7P/B1p5/2P5/3q2Pp/1P3P1K/2r5 w - - acd 62; bm Bc3+; c3 "Bc3+"; ce 696; id "11 Chase        +-"; pm Bc3+; pv Bc3+; 
2b5/1pr4p/3bp1pk/1p6/1PpN2PP/K1P1n3/P3N1R1/3R4 w - - acd 58; bm Nxb5; c3 "Nxb5"; ce 0; id "12 Desperation  =="; pm Nxb5; pv Nxb5 Nxd1 Nxd6 e5 Nxc8 Rxc8 Ka4 Ra8+ Kb5 Rxa2 Rg1 Nf2 Ng3 Nxg4 Kb6 Nf6 Kxb7 Rc2 Kc6 Rxc3 b5 Rb3 b6 c3 Rc1 Ng8 b7 Ne7+ Kc7 Nd5+ Kd6 Nf4 Kc7; 
3B2r1/n3Pp2/P7/5k1p/q3NP2/7K/5P2/4NB2 w - - acd 56; bm Nf6; c3 "Nf6"; ce 299; id "14 Fork-field   +-"; pm Nf6; pv Nf6 Rh8; 
8/q1p1N1p1/3p1p1p/pk5P/1p2BB2/8/PK3P2/8 w - - acd 86; bm Be3; c3 "Be3"; ce 170; id "15 Harassment   +-"; pm Be3; pv Be3; 
6kq/2p4p/6P1/5nN1/K7/2P5/7B/3B4 w - - acd 63; bm Bb3+; c3 "Bb3+"; ce 1814; id "16 Smothering   +-"; pm Bb3+; pv Bb3+; 
8/1B3b1p/7P/p2p4/P2pp1kp/Q7/1N1Pq2P/7K w - - acd 61; bm Qh3+; c3 "Qh3+"; ce 0; id "17 Fishing Rod  =="; pm Qh3+; pv Qh3+ Kxh3 Bc8+ Qg4 Nd1 e3 d3 Bg6 Bf5 Bh5 Be6 Be8 Bd7 Qxd7 Nf2+ exf2; 
8/pr1Pq3/br1kP3/1R2RQ2/5p2/1nPK4/5P2/8 w - - acd 67; bm Qxf4; c3 "Qxf4"; ce 0; id "19 Evil Eye     =="; pm Qxf4; pv Qxf4 Rxb5 Rxb5+ Kxe6 Qc4+ Kxd7 Rxb7+ Bxb7 Qxb3 Qe4+ Kd2 Qf4+ Ke1 Bc6 Qb4 Qc1+ Ke2 Qa1 Qb3 Qa5 Qf7+ Kc8 Qf8+ Kc7 Qf4+ Kb7 Qb4+ Qxb4 cxb4; 
4b3/5p2/2p4p/P1p1PN2/2pn2P1/P2k4/Pr1npKN1/B7 w - - acd 62; bm Ne1+; c3 "Ne1+"; ce 0; id "20 Roundabout   +-"; pm Ne1+; pv Ne1+ Kc3 Ng3 Bd7 a6 Bc8 a7 Bb7 Kg2 Ba8 Kf2; 

Not solved:
7q/b1p5/1p1Npkb1/pPP2ppP/P1P5/3B2P1/5P1R/K3R3 w - - acd 54; bm Nb7; c0 "http://www.yacpdb.org/#340530"; c1 "Hard"; c2 "Neghina, Mihai; Euxinus Pontus (13/E4) 2010-07"; c3 "hxg6"; ce 0; id "04 Detour +-"; pm Nb7; pv Nb7 Ke7+ Ka2 Qc3 Rd1 Qb4 Bc2 Qxc4+ Kb1 Qb4+ Ka2;
2k4N/Q1np4/2p2Bpp/1p1P4/pPP1p2P/P7/7q/1K6 w - - acd 55; bm Bb2; c3 "Nf7"; ce 0; id "06 Guard Change +-"; pm Bb2; pv Bb2 Qh3 dxc6 Qd3+ Ka2 Qxc4+ Ka1 Qf1+ Ka2;
8/4Pr1N/8/6PN/8/3K1ppp/4prkb/4Bbnn w - - acd 74; bm e8=Q; c3 "Nf8"; ce 0; id "13 Shield +-"; pm e8=Q; pv e8=Q Rd7+ Qxd7;
6k1/1qr1p2p/ppN5/3p1N2/6p1/1P1P2Pp/PP2PP1P/2RK4 w - - acd 51; bm Rc3; c3 "Ncxe7+"; ce -564; id "18 Stronghold =="; pm Rc3; pv Rc3 h5;

[peter]

Hi Dann!

Not solved:
7q/b1p5/1p1Npkb1/pPP2ppP/P1P5/3B2P1/5P1R/K3R3 w - - acd 54; bm Nb7; c0 "http://www.yacpdb.org/#340530"; c1 "Hard"; c2 "Neghina, Mihai; Euxinus Pontus (13/E4) 2010-07"; c3 "hxg6"; ce 0; id "04 Detour +-"; pm Nb7; pv Nb7 Ke7+ Ka2 Qc3 Rd1 Qb4 Bc2 Qxc4+ Kb1 Qb4+ Ka2;
2k4N/Q1np4/2p2Bpp/1p1P4/pPP1p2P/P7/7q/1K6 w - - acd 55; bm Bb2; c3 "Nf7"; ce 0; id "06 Guard Change +-"; pm Bb2; pv Bb2 Qh3 dxc6 Qd3+ Ka2 Qxc4+ Ka1 Qf1+ Ka2;
8/4Pr1N/8/6PN/8/3K1ppp/4prkb/4Bbnn w - - acd 74; bm e8=Q; c3 "Nf8"; ce 0; id "13 Shield +-"; pm e8=Q; pv e8=Q Rd7+ Qxd7;
6k1/1qr1p2p/ppN5/3p1N2/6p1/1P1P2Pp/PP2PP1P/2RK4 w - - acd 51; bm Rc3; c3 "Ncxe7+"; ce -564; id "18 Stronghold =="; pm Rc3; pv Rc3 h5;

As I told Vincent already, first move only doesn't prove most of the positions to be solved.
For sure not at the cyclic zugzwang positions and not at the fortresses.
It's not a best move- choice test- set.

[Dann Corbit]

Hi Dann!

Not solved:
7q/b1p5/1p1Npkb1/pPP2ppP/P1P5/3B2P1/5P1R/K3R3 w - - acd 54; bm Nb7; c0 "http://www.yacpdb.org/#340530"; c1 "Hard"; c2 "Neghina, Mihai; Euxinus Pontus (13/E4) 2010-07"; c3 "hxg6"; ce 0; id "04 Detour +-"; pm Nb7; pv Nb7 Ke7+ Ka2 Qc3 Rd1 Qb4 Bc2 Qxc4+ Kb1 Qb4+ Ka2;
2k4N/Q1np4/2p2Bpp/1p1P4/pPP1p2P/P7/7q/1K6 w - - acd 55; bm Bb2; c3 "Nf7"; ce 0; id "06 Guard Change +-"; pm Bb2; pv Bb2 Qh3 dxc6 Qd3+ Ka2 Qxc4+ Ka1 Qf1+ Ka2;
8/4Pr1N/8/6PN/8/3K1ppp/4prkb/4Bbnn w - - acd 74; bm e8=Q; c3 "Nf8"; ce 0; id "13 Shield +-"; pm e8=Q; pv e8=Q Rd7+ Qxd7;
6k1/1qr1p2p/ppN5/3p1N2/6p1/1P1P2Pp/PP2PP1P/2RK4 w - - acd 51; bm Rc3; c3 "Ncxe7+"; ce -564; id "18 Stronghold =="; pm Rc3; pv Rc3 h5;

As I told Vincent already, first move only doesn't prove most of the positions to be solved.
For sure not at the cyclic zugzwang positions and not at the fortresses.
It's not a best move- choice test- set.

All of the scores except for one agree for the "solved" set (assuming the == and += notations are correct).
E.g., all the ones that state an advantage show and advantage and all the draws have a score of zero except for one. And the draw with a negative score held the same score for many plies so it's drawn.

The unsolved set are all wrong answers.

[Dann Corbit]

Hi Dann!

Not solved:
7q/b1p5/1p1Npkb1/pPP2ppP/P1P5/3B2P1/5P1R/K3R3 w - - acd 54; bm Nb7; c0 "http://www.yacpdb.org/#340530"; c1 "Hard"; c2 "Neghina, Mihai; Euxinus Pontus (13/E4) 2010-07"; c3 "hxg6"; ce 0; id "04 Detour +-"; pm Nb7; pv Nb7 Ke7+ Ka2 Qc3 Rd1 Qb4 Bc2 Qxc4+ Kb1 Qb4+ Ka2;
2k4N/Q1np4/2p2Bpp/1p1P4/pPP1p2P/P7/7q/1K6 w - - acd 55; bm Bb2; c3 "Nf7"; ce 0; id "06 Guard Change +-"; pm Bb2; pv Bb2 Qh3 dxc6 Qd3+ Ka2 Qxc4+ Ka1 Qf1+ Ka2;
8/4Pr1N/8/6PN/8/3K1ppp/4prkb/4Bbnn w - - acd 74; bm e8=Q; c3 "Nf8"; ce 0; id "13 Shield +-"; pm e8=Q; pv e8=Q Rd7+ Qxd7;
6k1/1qr1p2p/ppN5/3p1N2/6p1/1P1P2Pp/PP2PP1P/2RK4 w - - acd 51; bm Rc3; c3 "Ncxe7+"; ce -564; id "18 Stronghold =="; pm Rc3; pv Rc3 h5;

As I told Vincent already, first move only doesn't prove most of the positions to be solved.
For sure not at the cyclic zugzwang positions and not at the fortresses.
It's not a best move- choice test- set.

All of the scores except for one agree for the "solved" set (assuming the == and += notations are correct).
E.g., all the ones that state an advantage show and advantage and all the draws have a score of zero except for one. And the draw with a negative score held the same score for many plies so it's drawn.

The unsolved set are all wrong answers.

I was mistaken.
This position:
[d]4b3/5p2/2p4p/P1p1PN2/2pn2P1/P2k4/Pr1npKN1/B7 w - -
shows a score of zero and it should show an advantage.

[Vinvin]

http://home.scarlet.be/vincentlejeune/chichitza.zip

pagina bestaat niet

Correct link : http://home.scarlet.be/vincentlejeune/c ... chitza.zip

[peter]

All of the scores except for one agree for the "solved" set (assuming the == and += notations are correct).
E.g., all the ones that state an advantage show and advantage and all the draws have a score of zero except for one. And the draw with a negative score held the same score for many plies so it's drawn.

Stubbornness (5) isn't solved to me, if the negative score is a clearly losing one (-3+) for about 100 moves till 50 moves boundary comes into engines' horizon finally, after black pawn move at about 50 starts the counter all over again.
Since when is a losing eval to be judged as correct, if the result is draw?

Just because it's held to be a clearly losing one for 100 moves? Of course you can adjudicate that as you like, a GUI wouldn't, at least not as drawn.

Did all the engines you tried, solve all the positions you counted as solved within an hour, or was it sometimes a single one engine only?

[peter]

All of the scores except for one agree for the "solved" set (assuming the == and += notations are correct).

P.S.
I guess, with += you mean +-, there isn't a single += eval of one of the studies to be counted as solved of course, if it's "White to move and win".
If it wasn't a typo only, have one more look at Mihai's commentary symbols at the positions in the set again.
http://talkchess.com/forum3/viewtopic.p ... 30#p773630

[Dann Corbit]

All of the scores except for one agree for the "solved" set (assuming the == and += notations are correct).
E.g., all the ones that state an advantage show and advantage and all the draws have a score of zero except for one. And the draw with a negative score held the same score for many plies so it's drawn.

Stubbornness (5) isn't solved to me, if the negative score is a clearly losing one (-3+) for about 100 moves till 50 moves boundary comes into engines' horizon finally, after black pawn move at about 50 starts the counter all over again.
Since when is a losing eval to be judged as correct, if the result is draw?

Just because it's held to be a clearly losing one for 100 moves? Of course you can adjudicate that as you like, a GUI wouldn't, at least not as drawn.

Did all the engines you tried, solve all the positions you counted as solved within an hour, or was it sometimes a single one engine only?

Mix and match. Some positions were solved by all and some were solved by only one engine.

[peter]

Mix and match. Some positions were solved by all and some were solved by only one engine.

So I agree with your results, Dann, with only one exception of Stubbornnes (5) which you doubted by yourself too.

The ones that weren't solved by your engines, are the same ones, that aren't solved by default settings on my machine, having seen, you corrected Roundabout (20) in the extra- posting too.
http://talkchess.com/forum3/viewtopic.p ... 74#p774174
A very nice cyclic zugzwang- piece, no chance to get more then the first 5 moves of solution into output- line nor the winning eval. Too much demanded for Lazy King's- eval together with the many Zugzwang- moves.

And of course you meant +-, not +=, as I saw in your solved-list already too.