|
| Page: [Previous] [Next] 1 2 3 |
| (1) Posted by Miguel Ambrona [Monday, Mar 1, 2021 15:00] |
h#49 legal position, non-unique (after Karl Fabel, Schachmatt 1948)
Hello!
I have modified the original h#48 by Karl Fabel, to make it one move longer:
New h#49:
 (= 13+10 )
Original h#48:
 (= 11+10 )
I have also verified that the position is legal.
Let me know what you think!
|
|
| (2) Posted by Olaf Jenkner [Monday, Mar 1, 2021 21:38] |
According to Gustav there is no shorter solution.
|
|
| (3) Posted by James Malcom [Monday, Mar 1, 2021 22:53] |
Welcome to the Matplus forums, Miguel! A very nice extension to a former 73-year-old record.
|
|
| (4) Posted by Miguel Ambrona [Tuesday, Mar 2, 2021 13:41] |
Thanks for your answers!
I wonder who Gustav is.
|
|
| (5) Posted by James Malcom [Tuesday, Mar 2, 2021 14:06] |
Gustav is the name of a chess solving program created by Olaf, mostly for selfmates, but also other things.
|
|
| (6) Posted by Miguel Ambrona [Tuesday, Mar 2, 2021 15:44] |
Very interesting. Is Gustav publicly available? I have searched a bit but couldn't find it.
How long does it take Gustav to solve the h#49, running on a modern processor?
|
|
| (7) Posted by Olaf Jenkner [Tuesday, Mar 2, 2021 19:41] |
On my laptop it ran 35 seconds.
You can send me an email: ojexyzgustav.in-chemnitz.de
Replace xyz by the well known character.
|
|
| (8) Posted by Viktoras Paliulionis [Tuesday, Mar 2, 2021 19:45] |
Gustav is not free. The free solving program Popeye (with Olive GUI) solved this problem in 14 seconds.
|
|
| (9) Posted by Harry Fougiaxis [Wednesday, Mar 3, 2021 08:59] |
@ Ambrona: Have a look at the page https://www.wfcc.ch/software/
|
|
| (10) Posted by Miguel Ambrona [Wednesday, Mar 3, 2021 12:31] |
Thank you so much for all the info!
I feel overwhelmed by the amount of material that can be found in the wfcc page.
I successfully installed Popeye and solved the h#49 with it (it took 16 seconds on my laptop).
I also tried the web gui of Jacobi, really nice!
I made a tool myself, although the purpose of my tool is deciding unwinnability (in order to classify the result of a game after a player times out).
Check it out here:
https://elrubiongamma.ddns.net/chess-helpmate-analyzer/about.html
(You can interactively use the tool in the "Analyzer" section.)
|
|
| (11) Posted by Sarah Hornecker [Wednesday, Mar 3, 2021 14:44] |
There are some interesting cases in your file, such as this.
8/8/6p1/5pkp/8/7P/3r2r1/7K w - - 0 45 UsN4NLQ4
 (= 2+6 )
White to move.
https://lichess.org/UsN4NLQ4#88
In this bullet game, Black - with his last move 44.-Re2xPg2?? - enforced the stalemate, but Lichess did not see the stalemate (45.h4+ is forced and any reply stalemates) and declared Black the winner.
A similar case, again White to move:
8/p4pk1/3b4/5Pp1/6q1/8/8/7K w - - 0 1
 (= 2+6 )
https://lichess.org/BFfZguIQ#84
Black's previous move 42.-Be7-d6?? blunders away half a point, enforcing the stalemate (43.f6+ and any move stalemates), but Lichess declared Black the winner when White overstepped the time.
A similar number of false non-draws (ca. 0.005 to 0.01 percent) would also be expected for any other online chess platform.
|
|
| (12) Posted by Miguel Ambrona [Wednesday, Mar 3, 2021 15:33] |
Exactly! There are many interesting positions in the analyzed subset of games from Lichess (you can easily see all the final position of the unfairly classified games here: https://elrubiongamma.ddns.net/d3-chess/games.html).
This is my favourite one (https://lichess.org/PIL4PUtT#87):
8/8/3Q4/2P5/1PB3k1/P7/5PpK/R4R2 b - - 1 44
 (= 9+2 )
It's Black to move and they promoted to a queen. After that, White ran out of time.
Any underpromotion would have better!
It is also quite interesting to see real games that end up like this (https://lichess.org/jnsJhG8T):
8/1k6/1p1pB3/pPpP4/P1Pp1p1p/3PbP1P/6K1/8 w - - 48 85
 (= 9+9 )
|
|
| (13) Posted by Viktoras Paliulionis [Wednesday, Mar 3, 2021 22:09] |
Miguel, your tool is interesting, but its name is misleading, because it can neither find the shortest solution nor analyze the solution. Its name almost coincides with the tool "Helpmate Analyzer" developed by me especially for helpmate composers (http://helpman.komtera.lt).
The Helpmate Analyzer not only solves helpmates (using Popeye), but also analyzes their thematic content, detects themes, and so on. It currently analyzes helpmates up to 32 moves only, as the longest known correct helpmate without cooks is h#28.
I think your tool is useful for chess players. As I know, previously it was named Dead Draw Detector. I think that name is more accurate and also doesn’t duplicate the name of the tool "Helpmate Analyzer".
|
|
| (14) Posted by Miguel Ambrona [Thursday, Mar 4, 2021 14:11] |
Hi Viktoras,
My tool can find the shortest helpmate, it is mentioned in the installation instructions (https://github.com/miguel-ambrona/D3-Chess). But in order to decide whether the position can be won it is more efficient to not aim for the shortest.
I though a lot about the name. Especially because the previous name "Dead Draw Detector", as you say, was criticized: the tool does more than deciding whether a position is a dead draw, but tells you whether a specific player can win.
I think the name "Chess Helpmate Analyzer" is quite accurate, since the tool basically analyzes whether a helpmate exists. However, given the collision with your tool's name, I will change mine. Do you have any suggestions?
Btw, nice tool! I wonder what the reason for the 32 moves depth limit is.
|
|
| (15) Posted by Viktoras Paliulionis [Thursday, Mar 4, 2021 20:48] |
I have tried only online version of your tool, and it cannot solve helpmates.
By the way, how long does it take for your tool to solve the following typical helpmates:
8/1p3n1r/1p2k3/p2p4/3r1n2/q7/P4K2/8 (h#9)
3n4/2k5/1p6/1b6/p2p2p1/Pp1p2p1/1P1P3p/R1B1K3 (h#5)
I don’t know what name would suit your tool, but I don’t think the word "helpmate" is appropriate in it. By definition, "helpmate" is a type of chess problems, but not the checkmate achieved by cooperative play. Various types of problems with a cooperative play have the common name "helpplay problems", so perhaps the term "helpplay" could be used in the name of your tool.
Now about the limit of 32 moves. The main purpose of Helpmate Analyzer is not to solve a problem, but to analyze its solution. The analysis includes the identification of the helpmate themes, the identification of the motives of each move, the calculation of various aesthetic criteria. This requires specific algorithms and data structures. The limit of 32 moves (64 half-moves) is because it is convenient to use 64bit numbers as bitsets in some algorithms. Since typical helpmates are up to 10 moves, 32 moves should be fully sufficient in most cases.
|
|
| (16) Posted by Andrew Buchanan [Friday, Mar 5, 2021 03:58] |
“Helpmate” in English also means assistant or helper. So the name “helpmate helpmate” would be cool once understood. Maybe abbreviated to HH to avoid confusing the uninitiated :-)
|
|
| (17) Posted by Viktoras Paliulionis [Friday, Mar 5, 2021 09:40] |
Since the number of moves does not matter, this is similar to solving an endgame study, therefore, by analogy, this could be called a help-study or help-win study.
|
|
| (18) Posted by Joost de Heer [Friday, Mar 5, 2021 10:12] |
QUOTE
Since the number of moves does not matter, this is similar to solving an endgame study, therefore, by analogy, this could be called a help-study or help-win study.
No, the 'dead' check checks whether a mate is theoretically possible (i.e. whether the position is dead or not), not whether it can be forced. It probably won't check for optimal length.
|
|
| (19) Posted by Frank Richter [Friday, Mar 5, 2021 11:02] |
What about the name "chess-(un)winnability-analyzer"?
|
|
| (20) Posted by Miguel Ambrona [Friday, Mar 5, 2021 15:03] |
@Viktoras Paliulionis.
Right, I do not provide online querying to the -min procedure, I don't want to run costly operations on my server. You have to install the tool if you want to try that feature.
* 8/1p3n1r/1p2k3/p2p4/3r1n2/q7/P4K2/8 (h#9)
This is quite tough, it was solved in 7 minutes.
But it is actually a h#8, right? Not 9.
I guess the official solution is to capture everything on the diag a2-f7 and promote on f8 (the bK being on h8), although there are many other solutions, e.g., with promotion on e8 and the bK on c8.
* 3n4/2k5/1p6/1b6/p2p2p1/Pp1p2p1/1P1P3p/R1B1K3 (h#5)
This was solved in 5 seconds.
Thanks for the explanation on the depth 32, it makes sense.
@Andrew Buchanan
"Helpmate helpmate" is cool! Thanks for your help, mate!
But I guess I want to avoid the word "helpmate" in the name.
@Frank Richter
"Chess Unwinnability Analyzer" is very good. It was actually my second candidate. I will consider it!
|
|
Read more... |
Page: [Previous] [Next] 1 2 3
MatPlus.Net  Forum Helpmates h#49 legal position, non-unique (after Karl Fabel, Schachmatt 1948) |
ISC 2024
