See P1398939
Henrik Juel: Black captured b7xLc6 and axbxc; White captured [Lf8], e2xd3, and fxexdxc6xb7-b8=D.
Ke8 to b7, Th8 to b8, Kb7 to g8, Tb8 to f8, Da4 to b8 for unpromotion, -22.c6xDb7 Da8 -23.d5xLc6 Dd8 -24.d4 Lb7 -25.e3xTd4 Lc8 -26.f2xSe3 c6 -27.h5 b7xLc6, Lc6 to f1, f2xSe3 etc.
Black may not castle, because he already did.
It is not possible to avoid the early castling:
-11... Ka7 -12.Db4 Ka6 -13.Db8 Ka7 -14.D=b7 Kb8 -15.c6xDb7 Kc8 -16.d5xLc6 Kd8 -17.d4 Ke8 -18.e3xTd4 Da8 -19.f2xSe3 Dd8 -20.h5 Lb7 -21.h4 Lc8 and two white pawn retractions are missing (2012-07-23)
Yoav Ben-Zvi: The solution does not require a full analysis since if Black King and Rook never moved then Black is almost immediately in retro-stalemate. An alternative stipulation is "First move of Black King?". (2014-06-03)
A.Buchanan: Maybe there is intentional irony? (2014-06-03)
Henrik Juel: Nowadays composers are not afraid to use the real stipulation in this type of resolution retro, like 'Release the position'.
Formerly a formal stipulation was popular, like 'Mate in 1', with the real stipulation understood but not mentioned.
Here, the stipulation question may well be ironic, but you will get no solution points, if you just answer 'No'; you still need to explain how the position arose (2014-06-03)
A.Buchanan: In matters of art, there is no single correct answer. Certainly, fashion has changed. It is no longer felt that stipulations can only be forward: - a retro stipulation can stand alone. However, personally, I like to see forward and retro play combined even if one aspect dominates. Although for me an ideal stipulation may hence express just a forward goal, it's a different rationale from the old days when only forward goals counted. Speaking for myself, I find the stipulation "Release the position" prosaic and unappetizing, and I usually move on without bothering to solve. I am "not afraid" to say this, even though Henrik may perhaps be scornful. (2022-02-07, hidden) more ...
comment

Moldenhauer: Computerprüfung: C+ Stelvio 2.1 ca. 30 Minuten. 3S/12P 64 GB Hash.
Der Cook ist keine Cook weil die Endstellung nicht mit dem Diagramm
zusammenstimmt. Letzter Zug Lf3?
Nur die weiße kleine Rochade fand wirklich statt.
Keine Lösung: BP 19.5. (2024-03-24)
A.Buchanan: Please forgive me for being stubborn, but I don't think just checking BP20.0 & BP19.5 is sufficient for C+. The test must include BP19.0 & BP18.5, and these should be very quick to test compared to the first two. They can be done first, which would allow for some short cooks to be identified more quickly. Maybe Stelvio could offer an option for this "fourfold" test, to save on typing, which is the main inconvenience? (2024-03-25)
Moldenhauer: Hi Andrew, diese Frage hatten wir schon einmal. Damals antwortete Reto darauf.
In der Dokumentation zu Stelvio 2.1, Seite 6 unten und Seite 7 oben, steht die Erklärung. (2024-03-25)
A.Buchanan: Thanks Moldenhauer. Yes, Reto responded today to say: "Actually 20 and 19.5 are sufficient, Stelvio always checks for any solutions of same parity, so 20 includes 19, 18, ... and 19.5 includes 18.5, 17.5... I can see your point with adding an option for automatic double checking, I've added it to my list."
That's great, so in my opinion the PDB C+ report needs to indicate that Stelvio has also tested for the shorter solutions of the same parity, as this is tool-dependent behaviour which future readers might not know. Ideally this statement would appear at the end of the Stelvio output, as a kind of "certificate" that could be copied to PDB.
A problem with this whole issue is that there are different choices being made by different people. It complicates things unnecessarily. I think C+ for a PG should mean that the three shorter lengths have also been explored. This should be doable without getting into the fraught question of whether a shorter solution cooks a PG, although in my opinion it does, whatever the parity: a PG is just a diagram. Conceptually to me, PG in x.y is just SPG, with a courtesy hint to the solver of what length they are looking for. (2024-03-26)
comment

paul: Checked with Euclide for the first 19 moves (~45 min) (2014-07-04)
Moldenhauer: Computerprüfung: C+ Stelvio 1.11 00:02:47 Minuten. (hh:mm:ss)
Keine Lösung: BP 23.0, BP 23.5. (2023-04-01, hidden) more ...
comment

Henrik Juel: The paradox is explained by the need for Black to play De1xLc1
Natch 3.1 made no progress after an hour (2018-12-14)
Moldenhauer: Computerprüfung: C+ Stelvio 1.0 14:25:47 Stunden.
Keine Lösung: BP 14.0, BP 14.5. (2023-02-20)
TBr: In solution read 11.Taf1 (2008-09-20, hidden) more ...
comment

Henrik Juel: typos in solution, should read: 5.Kg2 Kd6 6.Kf3 Sh4+ 7.Ke3
Fake castling, the hard way in 12 moves
First 21.5 moves found OK by Natch 3.1 (2018-12-13)
Moldenhauer: Computerprüfung: C+ Stelvio 1.0 09:09:13 Stunden. (hh:mm:ss)
Keine Lösung: BP 21.0, BP 21.5. (2023-03-12, hidden) more ...
comment

A.Buchanan: Not just hidden castling, but the full Lois theme with sKh8 & sTe8 (2022-02-08)
Joost de Heer: 1. f4 Sf6 2. f5 Se4 3. f6 Sd2 4. Kf2 Sb3 5. Qd2 b6 6. Qh6 gh6 7. Bg5 Bg7 8. e3 OO 9. Bd3 Kh8 10. Bg6 fg6 11. f7 Rg8 12. f8=Q Sc1 13. Qf3 Re8 14. Qa8 Kg8 15. Sf3 Kf7 16. Rd1 Rh8 17. Qd5 Ke8 18. Qg8 Bf8 19. Rd5

Theme: Phoenix(Q), switchback(b), Lois Theme(kr), uncastling (2009-12-20, hidden) more ...
comment

Henrik Juel: Natch 3.1 did little in half an hour
Stopped at pos. 1 (2018-12-27)
paul: 1.h4 g5 2.hxg5 c5 3.Rh6 c4 4.Rb6 axb6 5.Sc3 Ra3 6.Sb5 Re3 7.fxe3 c3 8.dxc3 h5 9.Qd5 Rh6 10.Bd2 Rf6 11.O-O-O Rxf1 12.Be1 Rf4 13.Qf3 f5 14.Rd6 Rd4 15.Rf6 exf6 16.Sa3 Qe7 17.Sb1 Qa3 18.Bd2 Qa8 19.a4 Ba3 20.Kd1 d6 21.Ke1 Kd7 22.Bc1. Hidden white castling. (2020-12-01)
Moldenhauer: Computerprüfung: C+ Stelvio 1.11 00:25:05 Minuten. (hh:mm:ss)
Keine Lösung: BP 20.5, BP 21.0. (2023-03-12)
comment

Henrik Juel: Natch 3.1 made little progress in more than half an hour
It was stopped at pos. 272 (2018-12-25)
Moldenhauer: Computerprüfung: Stelvio 1.11 00:44:45 Minuten.(hh:mm:ss)
Keine Lösung: BP 20.0, BP 20.5. (2023-03-12, hidden)
Moldenhauer: Bei Computerprüfung C+ vergessen. (2023-03-12, hidden) more ...
comment

Objective: Objective: Staugaard (Pam-Krabbé) rook is captured.
Henrik Juel: solution
1.e4 d5 2.e5 Dd6 3.exd6 Kd7 4.dxe7 Kc6 5.e8=T Kb6 6.0-0-0-0-0-0 Lg4 7.Kf4 Lxe2 8.Kg3 Lc4
Hidden castling (2021-01-21)
A.Buchanan: Thematic Ceriani-Frolkin seems a great way to conceal P-K castling! But normally wK returns to e1 in hidden castling. Absence of P-K in engine makes it harder to verify soundness of implausible moves like that. (2021-01-22)
comment

cheeky retro-active version of P0001136
A.Buchanan: Looking at the original, I am wondering why wBh cannot be on h5. It clearly cannot retract to h3, but how many times is it needed to retract if White retracts from the critical position 1Q3rk1/2pppppp/8/2p4P/2P5/PPRP4/RBpP1PP1/N1K5 (2022-02-07)
A.Buchanan: Maybe the distinguished composer felt that the more spare White moves has, the more creditable is the fact that Black had no way to avoid prior castling. The problem is still sound either way. Of course, with the h#1 version, wBh6 is mandatory. wD location was pretty much arbitrary in the original, but now there is only one wD location for soundness: g3. (e5 & g4 are unsound). (2022-02-07)
A.Buchanan: The solution given in Keym’s wonderful “Eigenartige Schachprobleme” uses an unnecessary move by wPh. (2022-02-23)
A.Buchanan: In offering this version, I do not think that pure retros are in anyway inferior to forward problems, which was perhaps the motivation for forward stipulations in the days of yore. Rather, the aesthetic tension between past and future never tires for me. It is not an issue to me whether the balance of interest in a retro-active problem be equally divided between the two directions. Rather, it’s essential that the position have two readings: that the pieces have two purposes. This formal requirement of double meaning which is ubiquitous and quotidian in a cryptic crossword should also find a place near to the hearth in chess problems. So Morse might welcome Dexter. (2022-02-23)
A.Buchanan: I also like it when the locations of all the units are uniquely determined. This humble version finally nails down wQ & wPh (2023-08-15, hidden) more ...
comment

A.Buchanan: Terrific composition, by a new composer (2022-05-04)
Henrik Juel: Welcome back, Andrew
Yes, a very impressive proof game:
The white rooks switch places, [Pg7] goes slowly to create a Phenix, and the white king backtracks after castling
The motivations for this fine content are fairly simple, so I can only hope that it has not been done before (2022-05-04)
Henrik Juel: C++ Euclide 1.01 (2022-05-04, hidden)
Henrik Juel: solution
1. Bb2-b4 Bg7-g6 2. Lc1-b2 Bg6-g5 3. Lb2xh8 Bg5-g4
4. Sb1-c3 Bg4-g3 5. Dd1-b1 Bg3xh2 6. Db1-b3 Bh2xg1=T
7. 0-0-0 Tg1xf1 8. Th1-h6 Tf1-h1 9. Th6-a6 Th1-h5
10. Td1-h1 Th5-a5 11. Kc1-d1 Ta5xa2 12. Kd1-e1 Ta2-b2
13. Ta6-a1 (2022-05-04, hidden) more ...
comment

Henrik Juel: 1.fxg3ep 2.0-0 3.Txf5 4.Kf8 5.Ke8 6.Tf8 7.Th8 Lh5#
Another 'uncastling' (2022-05-27)
comment

Thomas Brand wrote: "By the way, the position is also correct with sTd1 and wLc1 still present. Here, however, I completely agree with the author that this is one of the rare cases where the position with additional (even non-thematic) captures is preferable, since it hides the issue even better: why should the rook not have come via a1?"
Henrik Juel: White castles (to let wD out), 'uncastles' (to let sT in), and 'castles' again (to reach diagram) (2022-06-11)
Henrik Juel: Thanks for correcting the diagram, James (2022-06-12)
James Malcom: Ti's was Olaf, not me. I merely corrected the comma in the stipulation, which is something PDB needs sorted out--commas v. periods in BP/KBP problems. Consensus seems to learn towards periods, but many past entries contain commas. (2022-06-13)
A.Buchanan: You're welcome, Henrik. I didn't have the FEN, so edited the original diagram position manually - not the best procedure, but there was a solution to check against. Normally FEN to pieces is automated by a spreadsheet.
Continental European usage is comma rather than period, see: https://brilliantmaps.com/decimals/ (For readability of large numbers, the opposite convention is used: e.g. 1.234.567 vs 1,234,567. So combining them it would be 1,234,567.89 vs 1.234.567,89! See https://www.grammar-monster.com/lessons/commas_with_numbers.htm)
At some future point we can maybe have full internationalization of stipulations, but in the meantime, if I input e.g. a German or French problem, I respect their usage.
If I had to pick one, it would be comma, because period is used to indicate multiple solutions for helpmates e.g. 2.1... This matters because one often wants to search the stipulation as a string.
On the subject of the helpmate code: for various coherent reasons, I prefer no semicolons, ",5" over "0." and ellipsis after the first repeating "1.". It would be good to get consensus on this before implementing it. (2022-06-13)
Henrik Juel: diagram error black Kh6, not P (2022-06-11, hidden)
Henrik Juel: C+ Euclide 1.01 (2022-06-11, hidden) more ...
comment