Stockfish finds some old vey hard positions

[peter]

Very fine indeed. If there isn't the white Bishop sacrificed sooner, Cyclic Zugzwang arises after 10 moves!
Where is this one from again?

Code: Select all

4B3/1Pb4p/8/4p1k1/3p1p2/5K2/2P5/8 b - - 0 1

Analysis by Deep Shredder 13 x64:
...
1...Kf6 
  -/+  (-0.73 ++)   Depth: 41/55   00:00:08  272MN
1...e4+ 
  -/+  (-0.73 ++)   Depth: 41/55   00:00:09  309MN
...
1...e4+ 2.Kxe4 Kg4 3.Kxd4 f3 4.Ke4 Bb8 5.Bh5+ Kxh5 6.Kxf3 Kh4 7.Kg2 Kg4 8.c3 h5 9.Kg1 h4 10.Kg2 
  -/+  (-1.60 ++)   Depth: 41/55   00:00:13  463MN
...
1...e4+ 2.Kxe4 Kg4 3.Kxd4 f3 4.Ke4 Bb8 5.Bh5+ Kxh5 6.Kxf3 Kh4 7.Kg2 Kg4 8.c3 Bd6 9.c4 Bc7 10.c5 Bb8 11.Kf2 Bg3+ 12.Ke3 h5 13.c6 Bc7 
  -+  (-5.67 --)   Depth: 44/75   00:02:57  10139MN

[Paloma]

Where is this one from again?

From the Post above by Michael B , you didn't read it?
http://www.talkchess.com/forum/viewtopi ... 91&t=54289

scroll down to the last pgn
[White "Stockfish 130817 64 POPC"]
[Black "Aristides v2.0d 64 POPCN"]

[peter]

Where is this one from again?

From the Post above by Michael B , you didn't read it?

Indeed I didn't well enough, Herbert.

Yet it has something study- like, hasn't it?

Black is able just to force with Bishop- Triangles the White King to leave the same- coulored squares again and again alternating to advancing one Queen- side pawn after another being in Cyclic Zugzwang.

Where is Arpad Rusz and where is Mihai Neghina?

[Arpad Rusz]

Yet it has something study- like, hasn't it?

Black is able just to force with Bishop- Triangles the White King to leave the same- coulored squares again and again alternating to advancing one Queen- side pawn after another being in Cyclic Zugzwang.

Where is Arpad Rusz and where is Mihai Neghina?

I'm here!

Yes, the position before 67...e7! (with reversed colours) could be considered as a study. Unfortunately, there are several duals (even if they are all minor ones). But the key move is unique so this is a nice test position and a good example of chained Cyclic Zugzwangs.

[pgn][Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/2p5/5k2/3P1P2/4P1K1/8/1pB4P/4b3 w - - 0 1"]
[PlyCount "53"]

1. e5+ $1 Kxe5 2. Kg5 Kxd5 3. f6 Ke5 $1 4. Kg6 Bh4 5. f7 Be7 6. h4 Bf8 7. h5
Ke6 8. Bb1 Ke7 9. Bf5 c6 10. Bb1 Ke6 11. Bc2 Ke7 12. Bf5 c5 13. Bb1 Ke6 14. Bc2
Ke7 15. Bf5 c4 16. Bb1 Ke6 17. Bc2 Ke7 18. Bf5 c3 19. Bb1 Ke6 20. Bc2 Ke7 21.
Bf5 c2 22. Bxc2 Ke6 23. Bb1 Ke7 24. Bf5 b1=Q 25. Bxb1 Ke6 26. Ba2+ Ke7 27. h6 1-0[/pgn]

[peter]

But the key move is unique so this is a nice test position and a good example of chained Cyclic Zugzwangs.

Knew you would like it too, Arpad!

[Vinvin]

Again, an update 1.7 year later.
Now we can say that SF mastered the 2 first positions !!!
Search is set to 300 Seconds per move.

First position (Karpov's Ng2) found 13 times on 14 tries, 53 seconds average !
Second position (Shirov's Bh3) found 14 times on 14 tries, 24 seconds average !

1) Ne3-g2 Ne3-g2 * 0 Seconds
1) Ne3-g2 Ne3-g2 * 14 Seconds
1) Ne3-g2 Ne3-g2 * 178 Seconds
1) Ne3-g2 Ne3-g2 * 18 Seconds
1) Ne3-g2 Ne3-g2 * 184 Seconds
1) Ne3-g2 Ne3-g2 * 31 Seconds
1) Ne3-g2 Ne3-g2 * 42 Seconds
1) Ne3-g2 Ne3-g2 * 6 Seconds
1) Ne3-g2 Ne3-g2 * 6 Seconds
1) Ne3-g2 Ne3-g2 * 7 Seconds
1) Ne3-g2 Ne3-g2 * 9 Seconds
1) Ne3-g2 Ne3-g2 * 95 Seconds
1) Ne3-g2 Ne3-g2 * 96 Seconds
1) Ne3-g2 g3xh4
-------------------------------------------------
2) .. Bf5-h3 Bf5-h3 * 10 Seconds
2) .. Bf5-h3 Bf5-h3 * 14 Seconds
2) .. Bf5-h3 Bf5-h3 * 16 Seconds
2) .. Bf5-h3 Bf5-h3 * 20 Seconds
2) .. Bf5-h3 Bf5-h3 * 22 Seconds
2) .. Bf5-h3 Bf5-h3 * 23 Seconds
2) .. Bf5-h3 Bf5-h3 * 25 Seconds
2) .. Bf5-h3 Bf5-h3 * 33 Seconds
2) .. Bf5-h3 Bf5-h3 * 47 Seconds
2) .. Bf5-h3 Bf5-h3 * 47 Seconds
2) .. Bf5-h3 Bf5-h3 * 56 Seconds
2) .. Bf5-h3 Bf5-h3 * 7 Seconds
2) .. Bf5-h3 Bf5-h3 * 7 Seconds
2) .. Bf5-h3 Bf5-h3 * 8 Seconds

Update 1.5 year later.
Stockfish_160626_x64_modern_fast , Intel 4930K, 6*4 GHz, 1h per position
Analyzing engine: Stockfish_160626_x64_modern_fast

1) .. Bf5-h3 Bf5-h3 * 650 Seconds
5) .. Bf5-h3 Bf5-h3 * 327 Seconds
9) .. Bf5-h3 Bf5-h3 * 184 Seconds
13) .. Bf5-h3 Bf5-h3 * 2522 Seconds

2) Ne3-g2 g3xh4 - not found
6) Ne3-g2 g3xh4 - not found
10) Ne3-g2 g3xh4 - not found
14) Ne3-g2 g3xh4 - not found

3) Ng4-f6 Ng4xe3 - not found
7) Ng4-f6 Ng4xe3 - not found
11) Ng4-f6 d7-d8Q - not found

Intel 4930K, 6 cores 3.4 ghz

Shirov's Bh3!!! after 18min45
Karpov - Kasparov : Ng2!!! after 3h40m
Georgian peasant : Nf6+!!! after 1h18m

[Jouni]

SF is slowly but steadily improved and is now better problem solver than Houdini. Houdini (default) can be better now only in some weird made-up endgame positions.