interesting endgame

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

jdart
Posts: 4420
Joined: Fri Mar 10, 2006 5:23 am
Location: http://www.arasanchess.org

interesting endgame

Post by jdart »

This position arose in a game Spiderchess-Quark, part of a tournament by George Speight
(http://www.talkchess.com/forum/viewtopi ... 58&t=16672).

[d] 7R/3Kb3/4Pk2/7p/5ppP/8/8/8 b - -

here Quark played .. f3 and won. I did some analysis and for a while I thought .. g3 is also winning, but after long analysis Arasan thinks it is a draw. It is not simple, though. At first engines seem to predict a repetition by .. g3 Rg8 Bc5 Rg5 Be7 Rg8, but after Rg5, we are at this position:

[d] 8/3K4/4Pk2/2b3Rp/5p1P/6p1/8/8 b - -

and here Black can play Bb4 with a winning advantage.

The draw line I found after ..g3 requires Rh6+, not Rg8 as a followup. This is hard for many engines to see. They seem to get "stuck" on the repetition line and not see the Bb4 line.

--Jon
Will Singleton
Posts: 128
Joined: Thu Mar 09, 2006 5:14 pm
Location: Los Angeles, CA

Re: interesting endgame

Post by Will Singleton »

My engine suffers from the same myopia. Can't see f3, and after g3, thinks it's a draw. I don't think it's an artifact of any repetition bugs, but rather the inability to assess the two passers as unstoppable. In other words, if you have the proper eval for those pawns, then the fact that the rook can check the king a few times doesn't matter.

Will
jdart
Posts: 4420
Joined: Fri Mar 10, 2006 5:23 am
Location: http://www.arasanchess.org

Re: interesting endgame

Post by jdart »

I can find f3 but it takes a while:

Fen: 7R/3Kb3/4Pk2/7p/5ppP/8/8/8 b - - 0 1
Bm: f3

no 1 0 +45 13
no 1 0 -60 30 g3 Rxh5
no 2 0 -73 123 g3 Rh6+ Kg7 Rxh5
no 3 0 -73 314 g3 Rh6+ Kg7 Rxh5
no 3 0 -128 567 g3 Rxh5 g2 Rh6+ Kf5
no 4 0 -85 1544 g3 Rxh5 Kg6 Re5
no 5 10 -60 3335 g3 Rg8 Bc5 Rg5 Be7 Rxh5
no 6 20 -60 11486 g3 Rg8 Bc5 Rg5 Be7 Rxh5
no 6 20 -60 14978 g3 Rg8 Bc5 Rg5 Be7 Rxh5
yes 6 20 -154 17151 f3 Rh7 Bb4 Rf7+ Kg6 e7 Bxe7 Kxe7
no 7 40 -104 30803 g3 Rh6+
no 7 50 -85 35886 g3 Rh6+ Kg7 Rxh5 Kg6 Rh8 Kg7 Rc8
no 8 60 -85 51850 g3 Rh6+ Kg7 Rxh5 Kg6 Rh8 Kg7 Rc8
no 8 80 -140 62697 g3 Rh6+ Kg7 Rxh5 Kg6 Re5 g2 h5+ Kf6 Re1
f3
no 9 110 -90 89001 g3 Rh6+
no 9 150 -87 123012 g3 Rh6+ Kg7 Rxh5 g2 Kxe7 g1=Q Rg5+ Qxg5+
hxg5 Kg6 Kf8 Kxg5
no 10 280 -37 229714 g3 Rh6+
no 10 470 +0 389059 g3 Rg8 Bc5 Rg5 Be7 Rg8
no 11 980 +0 816221 g3 Rg8 Bc5 Rg5 Be7 Rg8
no 12 1830 +0 1581167 g3 Rg8 Bc5 Rg5 Be7 Rg8
no 13 3000 +0 2576469 g3 Rg8 Bc5 Rg5 Be7 Rg8
no 14 5450 +0 4678529 g3 Rg8 Bc5 Rg5 Be7 Rg8
no 15 11640 +0 10090368 g3 Rg8 Bc5 Rg5 Be7 Rg8
no 16 33590 +0 30036770 g3 Rg8 Bc5 Rg5 Be7 Rg8
yes 17 184680 +50 158383029 f3 Rh7
yes 17 320780 +140 276772578 f3 Rh6+ Kg7 Rxh5 Kg6 Ra5 f2 Kxe7 f1=Q
Rg5+ Kh6 Rxg4 Qf3 Rd4 Kg7 Kd7 Qb7+ Kd6
Qb6+ Ke5 Qc5+ Rd5 Qe3+ Kd6 Qf4+ Re5


Finding Rh6+ after g3 is even harder:

Fen: 7R/3Kb3/4Pk2/7p/5p1P/6p1/8/8 w - - 0 1
Bm: Rh6+

no 0 0 +40 20
yes 1 0 +73 33 Rh6+ Kg7
no 2 0 +123 80 Rxh5
no 2 0 +128 200 Rxh5 g2 Rh6+ Kf5
no 3 10 +85 599 Rxh5 Kg6 Re5
no 3 10 +85 1672 Rxh5 Kg6 Re5
no 4 10 +60 1772 Rg8 Bc5 Rg5 Be7 Rxh5
no 5 10 +60 3387 Rg8 Bc5 Rg5 Be7 Rxh5
yes 6 10 +110 7290 Rh6+ Kg7
yes 6 20 +159 9801 Rh6+ Kg7 Rxh5 Kf6 Rh7 g2 Rf7+ Ke5
yes 7 20 +159 14542 Rh6+ Kg7 Rxh5 Kf6 Rh7 g2 Rf7+ Ke5
yes 7 30 +79 20303 Rh6+ Kg7 Rxh5 Kg6 Rh8 Kf6 Rg8
yes 8 40 +129 27784 Rh6+ Kg7
yes 8 50 +140 41480 Rh6+ Kg7 Rxh5 Kg6 Re5 g2 h5+ Kf6 Re1 f3
yes 9 110 +140 91125 Rh6+ Kg7 Rxh5 Kg6 Re5 g2 h5+ Kf6 Re1 f3
yes 9 150 +87 124792 Rh6+ Kg7 Rxh5 g2 Kxe7 g1=Q Rg5+ Qxg5+
hxg5 Kg6 Kf8 Kxg5
yes 10 710 +87 574813 Rh6+ Kg7 Rxh5 g2 Kxe7 g1=Q Rg5+ Qxg5+
hxg5 Kg6 Kf8 Kxg5
yes 10 830 +87 674229 Rh6+ Kg7 Rxh5 g2 Kxe7 g1=Q Rg5+ Qxg5+
hxg5 Kg6 Kf8 Kxg5
no 10 980 +0 823899 Rg8 Bc5 Rg5 Be7 Rg8
no 11 1640 +0 1426668 Rg8 Bc5 Rg5 Be7 Rg8
no 12 2240 +0 1908483 Rg8 Bc5 Rg5 Be7 Rg8
no 13 3820 +0 3265846 Rg8 Bc5 Rg5 Be7 Rg8
no 14 6760 +0 5925037 Rg8 Bc5 Rg5 Be7 Rg8
no 15 21580 +0 18910212 Rg8 Bc5 Rg5 Be7 Rg8
no 16 46970 +0 42089482 Rg8 Bc5 Rg5 Be7 Rg8
no 17 187260 -26 162389863 Rg8 Bb4 Rg5 Ba3 Rg8
no 18 347950 -26 302902440 Rg8 Bb4 Rg5 Ba3 Rg8
no 18 477260 -76 417441271 Rg8 Bb4
yes 18 525760 -84 461443491 Rh6+ Kg7 Rxh5 f3 Kxe7 f2 Rf5 g2 Rf7+ Kg6
Rxf2 g1=Q Rf6+ Kh5 Kf7 Qa7+ e7 Kxh4 Ra6
Qb7 Kf8 Qf3+ Kg7 Qc3+ Kf8 Qh8+ Kf7 Qh7+
Kf8
yes 19 603080 -34 529464896 Rh6+ Kg7
yes 19 759160 +0 666895643 Rh6+ Kg7 Rxh5 f3 Kxe7 f2 Rf5 g2 Rf7+ Kg6
Rxf2 g1=Q Rf6+ Kh7 Rf7+ Kh6 Rf6+ Kh7
bob
Posts: 20943
Joined: Mon Feb 27, 2006 7:30 pm
Location: Birmingham, AL

Re: interesting endgame

Post by bob »

jdart wrote:This position arose in a game Spiderchess-Quark, part of a tournament by George Speight
(http://www.talkchess.com/forum/viewtopi ... 58&t=16672).

[d] 7R/3Kb3/4Pk2/7p/5ppP/8/8/8 b - -

here Quark played .. f3 and won. I did some analysis and for a while I thought .. g3 is also winning, but after long analysis Arasan thinks it is a draw. It is not simple, though. At first engines seem to predict a repetition by .. g3 Rg8 Bc5 Rg5 Be7 Rg8, but after Rg5, we are at this position:

[d] 8/3K4/4Pk2/2b3Rp/5p1P/6p1/8/8 b - -

and here Black can play Bb4 with a winning advantage.

The draw line I found after ..g3 requires Rh6+, not Rg8 as a followup. This is hard for many engines to see. They seem to get "stuck" on the repetition line and not see the Bb4 line.

--Jon
Nice to see crafty get one right every now and then. :)

On my laptop:
16-> 1.28 -0.01 1. ... g3 2. Rg8 Ba3 3. Rg5 Be7 4.
Rg8 (s=3)
17 1.64 -0.01 1. ... g3 2. Rg8 Ba3 3. Rg5 Be7 4.
Rg8 (s=2)
17-> 2.27 -0.01 1. ... g3 2. Rg8 Ba3 3. Rg5 Be7 4.
Rg8 (s=2)
18 3.26 -0.01 1. ... g3 2. Rg8 Ba3 3. Rg5 Be7 4.
Rg8
18 8.26 -1 1. ... f3!!
18 10.55 -1.22 1. ... f3 2. Rh7 Ba3 3. Rf7+ Ke5 4.
e7 Bxe7 5. Kxe7 Ke4 6. Rf6 g3 7. Re6+
Kf4 8. Rf6+ Kg4 9. Rb6 Kxh4 10. Rb4+
Kh3 11. Rb5 h4 12. Rf5
18-> 11.05 -1.22 1. ... f3 2. Rh7 Ba3 3. Rf7+ Ke5 4.
e7 Bxe7 5. Kxe7 Ke4 6. Rf6 g3 7. Re6+
Kf4 8. Rf6+ Kg4 9. Rb6 Kxh4 10. Rb4+
Kh3 11. Rb5 h4 12. Rf5 (s=3)
19 17.51 -1 1. ... f3!!
19 24.16 -1.65 1. ... f3 2. Rh6+ Kg7 3. Rxh5 Kg6 4.
Rh8 f2 5. Kxe7 f1=Q 6. Rg8+ Kh6 7.
Rxg4 Qf3 8. Rd4 Qa3+ 9. Kf6 Qf8+ 10.
Ke5 Qg7+ 11. Kd5 Kh5 12. Rf4 Qb7+ 13.
Ke5 Qc7+ 14. Kf5 Qe7 (s=2)
19-> 28.04 -1.65 1. ... f3 2. Rh6+ Kg7 3. Rxh5 Kg6 4.
Rh8 f2 5. Kxe7 f1=Q 6. Rg8+ Kh6 7.
Rxg4 Qf3 8. Rd4 Qa3+ 9. Kf6 Qf8+ 10.
Ke5 Qg7+ 11. Kd5 Kh5 12. Rf4 Qb7+ 13.
Ke5 Qc7+ 14. Kf5 Qe7
20 38.36 -1 1. ... f3!!
20 1:04 -1.96 1. ... f3 2. Rh6+ Kg7 3. Kxe7 Kxh6
4. Kd6 f2 5. e7 f1=Q 6. e8=Q Qd1+ 7.
Kc5 Qc1+ 8. Kd6 Qd2+ 9. Kc6 Qc2+ 10.
Kd5 Qa2+ 11. Kc5 Qf2+ 12. Kd5 Qf5+
13. Kd4 Qf4+ <HT>
20-> 1:12 -1.96 1. ... f3 2. Rh6+ Kg7 3. Kxe7 Kxh6
4. Kd6 f2 5. e7 f1=Q 6. e8=Q Qd1+ 7.
Kc5 Qc1+ 8. Kd6 Qd2+ 9. Kc6 Qc2+ 10.
Kd5 Qa2+ 11. Kc5 Qf2+ 12. Kd5 Qf5+
13. Kd4 Qf4+ <HT>
21 1:46 -2.06 1. ... f3 2. Rh6+ Kg7 3. Kxe7 Kxh6
4. Kd6 f2 5. e7 f1=Q 6. e8=Q Qf4+ 7.
Kd5 Kg7 8. Qxh5 g3 9. Qe2 Qxh4 10.
Qe5+ Kg6 11. Qe8+ Kf5 12. Qe6+ Kf4
13. Qe4+ Kg5 14. Qe7+ Kg4 15. Qe4+
Kh5 16. Qg2
21-> 2:02 -2.06 1. ... f3 2. Rh6+ Kg7 3. Kxe7 Kxh6
4. Kd6 f2 5. e7 f1=Q 6. e8=Q Qf4+ 7.
Kd5 Kg7 8. Qxh5 g3 9. Qe2 Qxh4 10.
Qe5+ Kg6 11. Qe8+ Kf5 12. Qe6+ Kf4
13. Qe4+ Kg5 14. Qe7+ Kg4 15. Qe4+
Kh5 16. Qg2
time=2:22 mat=0 n=794255869 fh=85% nps=5.6M
ext-> check=89.1M 1rep=2.2M mate=25K pp=0 reduce=438.0M/59.0M
predicted=0 evals=476.7M 50move=0 EGTBprobes=0 hits=0
SMP-> splits=3169 aborts=336 data=7/128 elap=2:22