Long(ish) mates?

[JVMerlino]

Hi all,

I'm trying to verify certain parts of my search, and am looking for long mating sequences that are relatively easy to find (certainly no more than one quiet move), and don't require EGTBs.

My engine now finds mates like this quite quickly. It's a Mate in 7 with either Rg1 or Ba4. Myrddin finds it at depth 4 in about 0.1 seconds

[d] 7k/3b3p/1p6/3p1P1P/2p3r1/2P1R3/7r/2RK4 b - - 0 40

So I'm now looking for longer stuff to test with. If there's a test suite available, great. If not, then hopefully I can build a decent one from the responses.

Many thanks,
jm

[Eelco de Groot]

Hello John,

Norbert Raimund Leisner kindly sent me a PM last week with a link that should suit your needs, a compilation of chess problem solving software and links to sites with all sorts of chess puzzles:

http://computer-chess.org/doku.php?id=c ... s_problems

If you want the really hard ones, I believe ChestUCI from Franz Huber has a database that is very extensive and you can get no better source to compare materesults with than that of ChestUCI!

Regards, Eelco

[JVMerlino]

This is marvelous. Thanks very much, Eelco!

[Tord Romstad]

Hi all,

I'm trying to verify certain parts of my search, and am looking for long mating sequences that are relatively easy to find (certainly no more than one quiet move), and don't require EGTBs.

Here's an old favorite of mine, from a blitz game my program played on the ICC many years ago:

[d]3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - -

The position is a mate in 16, with a single quiet move (13. Nxh3) towards the end. The first move is probably easy to find, announcing mate will probably take some time for most programs.

The complete (and very pretty) main line is 1. Rh8+ Bxh8 2. Rxh8+ Kg7 3. Bd4+ f6 4. gxf6+ exf6 5. Bxf6+ Kf7 6. Ng5+ Kxf6 7. e5+ dxe5 8. Nce4+ Kg7 9. Rh7+ Kf8 10. Qh3 Qxh3 11. Rf7+ Kg8 12. Nf6+ Kh8 13. Nxh3 Rd7 14. Rxd7 Re7 15. Rxe7, followed by 16. Rh7#.

Stockfish's search explodes extremely violently in this position, so much that I think there might be some sort of bug. I need to look into this...

[stegemma]

Very nice position. I think that the tree explosion could be normal because the black pawns on queen side can easly promote in different variants. Freccia slow down very soon, at ply 12 (and find only Bd4, at that ply). While i'm writing. Freccia is still searching...

[Vinvin]

There's a lot of mate test suite on Dann Corbit's FTP !

http://cap.connx.com/EPD/

longmates.epd.bz2
easy-mates.epd.bz2

and more ...

[Dirt]

Very nice position. I think that the tree explosion could be normal because the black pawns on queen side can easly promote in different variants.

Maybe not so normal. Toga II 1.4 beta5c took about a minute to find the mate, Stockfish 1.2 still hasn't found it in over two hours.

[JVMerlino]

Great position, Tord!

Myrddin still needs some time just to find Rh8 -- 35 seconds on a 3.0GHz CPU. It finds it at depth 10. I don't know if that is reasonable or not for an engine that is around 2000 ELO.

But either way, thanks again!

jm