Other than ease of solution, a future-proof test suite

[Dann Corbit]

Created with the where clause:
where dm between 8 and 12 and acs > 0 and acd = 999 and dbo.ChessmanCount(Epd) >= 16 and (not (bm like '% %'))
This means every position is a mate in between 8 and twelve moves
The positions were not solved instantaneously by Chest319
The positions were proven (no special pruning so they are actually provably optimal and correct)
There are at least 16 chessmen on the board, so that they won't be trivially defeated by EGTB files
There is a single shortest mate (though other moves might mate, but it would take more moves)
Here is the link:


The files inside look like this:
10/09/2024 06:51 AM 2,935,920 provmats-ana.epd
10/09/2024 06:51 AM 772,064 provmats.epd

The archive looks like this:
10/09/2024 06:52 AM 879,085 provmats.7z

There are 12855 unique positions in the list.
Since the positions are uniquely solved with a single move for minimum mate, they can never be improved by additional search.
However, due to the exponential advances in both hardware and software, the test sets will eventually become trivial.

Notes:
the analysis file (provmats-ana.epd) gives the source when I have it.
many of the positions will be found in existing problem tests, but most are not.
many of the positions will also be from actual games.

[Dann Corbit]

A similar set with mates between 5 and 7 moves.

45,662 positions. Some are permutations or reflections due to experiments I have run in the past.

[Dann Corbit]

Note that the analysis file and the bare problem file have their names flipped.
Sorry, I do stuff like that a lot (dyslexic)

[Joerg Oster]

Created with the where clause:
where dm between 8 and 12 and acs > 0 and acd = 999 and dbo.ChessmanCount(Epd) >= 16 and (not (bm like '% %'))
This means every position is a mate in between 8 and twelve moves
The positions were not solved instantaneously by Chest319
The positions were proven (no special pruning so they are actually provably optimal and correct)
There are at least 16 chessmen on the board, so that they won't be trivially defeated by EGTB files
There is a single shortest mate (though other moves might mate, but it would take more moves)
Here is the link:


The files inside look like this:
10/09/2024 06:51 AM 2,935,920 provmats-ana.epd
10/09/2024 06:51 AM 772,064 provmats.epd

The archive looks like this:
10/09/2024 06:52 AM 879,085 provmats.7z

There are 12855 unique positions in the list.
Since the positions are uniquely solved with a single move for minimum mate, they can never be improved by additional search.
However, due to the exponential advances in both hardware and software, the test sets will eventually become trivial.

Notes:
the analysis file (provmats-ana.epd) gives the source when I have it.
many of the positions will be found in existing problem tests, but most are not.
many of the positions will also be from actual games.

A big thank you, Dann.

[peter]

Great, Dann, thanks a lot, regards

[Jouni]

Thanks. So these are checked with Chest full search?

[peter]

Checked on my own so far only those, that Huntsman doesn't solve within 5" at 30 threads, and only out of the latest 506 (#11+#12).

Here's nr. 12367:

[d]1r4k1/5pp1/2Rp4/3Pp3/pB2P2P/2P1P3/5K2/1q6 b - - 0 1

Analysis by The Huntsman 1:

1. -+ (-#11): 1...Dd3 2.h5 a3 3.Tc7 Ta8 4.Ta7 Txa7 5.c4 a2 6.Lc3 a1D 7.Lxa1 Ta2+ 8.Lb2 Txb2+ 9.Kf3 Df1+ 10.Kg3 Tg2+ 11.Kh4 Dh1#
2. -+ (-#11): 1...Dc2+ 2.Kf3 f5 3.Ta6 Tf8 4.Ta8 Txa8 5.Kg3 a3 6.c4 a2 7.Lc3 a1D 8.Lxa1 Txa1 9.h5 Dxe4 10.Kf2 Ta2+ 11.Kg3 Dg4#
3. -+ (-#11): 1...a3 2.Ta6 a2 3.Ke2 a1D 4.Txa1 Db2+ 5.Kf3 Dxa1 6.Ke2 Ta8 7.c4 Ta2+ 8.Ld2 Dc1 9.Kf3 Dxd2 10.Kg4 Dg2+ 11.Kf5 Tf2#
4. -+ (-#12): 1...Ta8 2.La3 Da2+ 3.Kf3 Dxa3 4.Tb6 Dc1 5.c4 Dh1+ 6.Ke2 a3 7.Tb1 Dxb1 8.Kf3 Tb8 9.h5 Dd1+ 10.Kg2 Tb2+ 11.Kh3 Dg1 12.h6 Th2#

Nr. 12398:

[d]3r3k/pp1P1Rp1/8/3n2Pp/2p4P/2P3K1/2q2B2/8 b - - 0 1

Analysis by The Huntsman 1:

1. -+ (-#11): 1...Kg8 2.Tf3 Sxc3 3.Lc5 Se4+ 4.Kh3 De2 5.Tf4 Sxc5 6.g6 Txd7 7.Tf2 Td3+ 8.Kg2 Dg4+ 9.Kh1 Dh3+ 10.Kg1 Td1+ 11.Tf1 Txf1#
2. -+ (-#11): 1...Txd7!? 2.Tf8+ Kh7 3.Tf3 Sxc3 4.Kh2 Td2 5.Kg3 De4 6.g6+ Kg8 7.Tf8+ Kxf8 8.Lc5+ Kg8 9.Lf2 Dg4+ 10.Kh2 Txf2+ 11.Kh1 Dd1#
3. -+ (-#11): 1...Sxc3 2.Le1 Dd3+ 3.Kf2 Kg8 4.g6 Sd1+ 5.Kg2 Dxg6+ 6.Kh2 Dxf7 7.Ld2 Df2+ 8.Kh3 Txd7 9.Le3 Df3+ 10.Kh2 Sxe3 11.Kg1 Dg2#
4. -+ (-#12): 1...Dg6 2.Tf3 Dd6+ 3.Kg2 Dxd7 4.Ld4 Dg4+ 5.Kf2 Sf4 6.Te3 Dg2+ 7.Ke1 Sd3+ 8.Kd1 Df1+ 9.Kd2 Dc1+ 10.Ke2 De1+ 11.Kf3 Df2+ 12.Ke4 Df4#

So, given solution 1...a3 at the one and 1...Rxd7 of the other one aren't single best moves, just edited those as for DTM equal solutions as second and third one bm(s), regards

[peter]

Next ones, nr. 12419:

[d]4r3/p5k1/8/5pPp/P1Pp3P/1P1P1RK1/6R1/3q4 b - - 0 1

Analysis by The Huntsman 1:

1. -+ (-#11): 1...f4+ 2.Txf4 Te3+ 3.Kh2 Dxd3 4.b4 Th3+ 5.Kg1 De3+ 6.Tff2 d3 7.Th2 Txh2 8.Kxh2 Dxf2+ 9.Kh3 d2 10.a5 d1D 11.g6 Ddf1#
2. -+ (-#11): 1...Dh1 2.Tgf2 Dg1+ 3.Kh3 Te3 4.Txe3 dxe3 5.Tg2 Df1 6.Kh2 f4 7.b4 e2 8.Tg1 Df2+ 9.Kh1 Dxh4+ 10.Kg2 Dg3+ 11.Kh1 Dh3#
3. -+ (-#11): 1...Te3 2.Txe3 dxe3 3.Kf4 Dxd3 4.b4 De4+ 5.Kg3 f4+ 6.Kh3 e2 7.Tg1 De3+ 8.Kh2 Df2+ 9.Kh1 Dxh4+ 10.Kg2 Dg3+ 11.Kh1 Dh3#
4. -+ (-#12): 1...Te6 2.c5 f4+ 3.Txf4 Te3+ 4.Kh2 Dxd3 5.Kg1 Th3 6.Tff2 De3 7.g6 d3 8.Tg5 Tf3 9.Kh2 Txf2+ 10.Tg2 Df4+ 11.Kh1 Df3 12.Kg1 Dxg2#

12432:

[d]5rk1/6p1/3p4/3P3q/1P2p3/p5Pp/PbR1Q3/7K b - - 0 1

Analysis by The Huntsman 1:

1. -+ (-#11): 1...Dg5 2.Kh2 Le5 3.De1 e3 4.Dg1 Dg4 5.b5 Tf3 6.Tc8+ Dxc8 7.b6 Df5 8.Kh1 Txg3 9.Dd1 De4+ 10.Df3 Dxf3+ 11.Kh2 Dg2#
2. -+ (-#11): 1...Dg6 2.b5 e3 3.b6 Le5 4.Tc1 De4+ 5.Kh2 Tf2+ 6.Dxf2 exf2 7.Kxh3 Kh7 8.Kh2 f1D 9.Txf1 Dh4+ 10.Kg2 Dxg3+ 11.Kh1 Dh2#
3. -+ (-#12): 1...Df5 2.Kh2 e3 3.b5 Dg6 4.b6 Le5 5.b7 Lxg3+ 6.Kxh3 Le5 7.Dg4 Dxc2 8.De6+ Kh7 9.Dxe5 dxe5 10.d6 Tf3+ 11.Kg4 Df5+ 12.Kh4 Th3#

12440:

[d]6k1/p1p5/1p1p4/2PPp3/1P3q2/5R2/P3BKP1/7r b - - 0 1

Analysis by The Huntsman 1:

1. -+ (-#11): 1...Dh4+ 2.Tg3+ Kf8 3.cxd6 cxd6 4.Lf1 e4 5.Le2 Df4+ 6.Lf3 e3+ 7.Ke2 Dc4+ 8.Kxe3 Te1+ 9.Kd2 Dc1+ 10.Kd3 Te3+ 11.Kd4 Dc3#
2. -+ (-#11): 1...Dd4+ 2.Te3 e4 3.cxd6 cxd6 4.a3 Df6+ 5.Tf3 exf3 6.Lxf3 Dd4+ 7.Ke2 Dc3 8.g4 Th2+ 9.Lg2 Txg2+ 10.Kf1 Df3+ 11.Ke1 De2#
3. -+ (-#15): 1...Dd2 2.Tg3+ Kf8 3.Tg4 De1+ 4.Ke3 Tf1 5.Kd3 Db1+ 6.Ke3 Dc1+ 7.Kd3 Da3+ 8.Kd2 Tb1 9.Ld1 Dc1+ 10.Kd3 Dxd1+ 11.Kc4 De2+ 12.Kc3 Dxg4 13.Kc2 Dd1+ 14.Kc3 e4 15.cxd6 Dd3#

Will stop that here now and return not until I'm through with the #11+#12- list, that will take some time anyhow.

[peter]

Edit- time over, just this one also already now to show, DTM isn't always correct neither, here is 12747 with a given #12- line, there's a shorter one with 2...Qxe1 instead of 2...Bxa2 (!?) of analyse- line of epd:

[d]6k1/6p1/2q2p1p/p1P1pP2/Prb3P1/3R3K/QP5P/4B3 b - - 0 1

Analysis by The Huntsman 1:

1. -+ (-#10): 1...De4 2.Tg3 Dxe1 3.b3 Lf1+ 4.Dg2 Lxg2+ 5.Kxg2 De2+ 6.Kh3 h5 7.gxh5 Df1+ 8.Tg2 Dxf5+ 9.Kg3 Tg4+ 10.Kh3 Dxh5#
2. -+ (-#11): 1...Lxa2 2.Lxb4 De4 3.Tg3 Lc4 4.Kh4 Df4 5.Tg2 axb4 6.c6 Kh7 7.Tc2 Ld3 8.Tg2 Lxf5 9.Kh3 Df3+ 10.Kh4 g5+ 11.Kh5 Dh3#
3. -+ (-#13): 1...Dh1 2.Lxb4 Lxa2 3.b3 Lb1 4.Tg3 h5 5.Ld2 hxg4+ 6.Kxg4 Dd1+ 7.Kh3 Dxd2 8.Tf3 Le4 9.Kg3 De1+ 10.Tf2 De3+ 11.Kh4 Dxf2+ 12.Kg4 Lxf5+ 13.Kh5 Dxh2#

[peter]

here is 12747 with a given #12- line, there's a shorter one

Position is stored as #11, not as #12 as mistyped above, everything else is correct.