The New Sudoku Players' Forum

[jco]

Hello,

I have two related questions that I've being willing to ask for a while. Perhaps all I need is a reference if the answer is yes.

When person A brings a (Sudoku) puzzle for analysis (in some sense. It can be asking for help or otherwise) how can an experienced player find out about the puzzle source if, suppose, person A doesn't provide (or lost) that information. How to find that out where it came from? Suppose it is not a new puzzle: it has been published somewhere and if given the information of the source one can access it). I mentioned one scenario, but of course, one can think of many similar situations. For any of them, I 'd like to know how to identify the source of the puzzle. Is this possible? I tried a couple times and it seems that using interned search engines does not work.
This issue called my attention because in more than 3 occasions someone brought a puzzle without mentioning its source and some experienced player immediately identified its source. However, I also saw a couple times an experienced player asking for the source of the puzzle.
If the answer in the general context is no, there is nothing, I still would like to know (second question) if there is a way to identify a puzzle already discussed in this Forum.
In the more general setting, it would be very nice if there was a standard (clever) way to register all published (in the above sense, or in other, perhaps more realistic) puzzles (like the ISBN in published books).

Regards,
jco

[Pupp]

I don't know if it's that important. There are millions of puzzles available, and people often take these data dumps of collections and analyze them in bulk, using programs, so they can find smaller lists of puzzles with specific attributes.

Then there is the fact that many Sudoku apps can just spit out random, instantly created puzzles on the fly. So it wouldn't be very useful to anybody if they happen to post a puzzle by say, for example, the program Hodoku that the program just created out of some algorithm. In fact, I'd say a great many of the crappier Sudoku apps (as oppose to the good apps), have no master list of puzzles at all, and just rely on instantly created random puzzles. So that puzzle wouldn't really be attached to a particular list of puzzles, since it was just an algorithm that spit out the puzzle.

The vast majority of puzzles were created by computer algorithms, then eventually got grouped into some list that eventually was put on the internet. If you looking for hand crafted puzzles, Cracking the Cryptic has several apps with hand crafted puzzles. There are not really a ton of puzzles on the apps per say compared to other Sudoku apps, but all the puzzles were picked from hand crafted puzzles. I'm not sure who crafts their puzzles, they've probably mentioned it on one of their videos, but they pump out an amazing number of Sudoku videos on Youtube.


Of course, over the decades of people playing Sudoku, there are a great many hand crafted puzzles, but I don't know offhand if there is any sort of data dump of them on the internet. I'd love to know if there was a large collection of hand crafted puzzles on the internet.

If this is any help, all my puzzles come from Sudoku 10000 Plus. Nobody's ever asked me for an exact puzzle number.

[coloin]

I can see why one would have difficulty ...

There are several billion grid solutions and each grid solution has hundreds of billions of minimal puzzles !!

To add to this each puzzle has 9! [362880] ways to exchange the numbers and potentially 6^8 x 2 [3359232] ways to row swap/rotate /reflect/shute swap etc - and be left with the same puzzle - all termed isomorphs..

Each puzzle can be represented by a 81 character text string.
eg

Code: Select all

..12..3...9..6..4.8.......7........2.5.....7.4........2.......3.6..4..5...3..58..


Code: Select all

+---+---+---+
|..1|2..|3..|
|.9.|.6.|.4.|
|8..|...|..7|
+---+---+---+
|...|...|..2|
|.5.|...|.7.|
|4..|...|...|
+---+---+---+
|2..|...|..3|
|.6.|.4.|.5.|
|..3|..5|8..|
+---+---+---+


here are 2 ways 2 standardize a puzzle, these are isomorphs of the above puzzle

Code: Select all

+---+---+---+                                             
|...|...|..1|                                             
|...|..2|...|                                             
|...|.3.|.4.|                                             
+---+---+---+                                             
|...|..1|..5|                                             
|..2|.6.|.3.|                                             
|.3.|5..|7..|                                             
+---+---+---+                                             
|...|..7|..4|                                             
|..6|.8.|.2.|                                             
|1..|9..|5..|                                             
+---+---+---+   min lex puzzle                         
                                                           
+---+---+---+                                             
|1..|..6|7..|                                             
|.5.|...|.3.|                                             
|..9|.2.|..4|                                             
+---+---+---+                                             
|...|...|.6.|                                             
|.4.|...|...|                                             
|..8|...|..3|                                             
+---+---+---+                                             
|..2|.4.|..8|                                             
|7..|8..|5..|                                             
|.6.|...|.7.|                                             
+---+---+---+  puzzle as it appears in minlex solution grid


A computor program can easily identify the minimum value for this text string out of all the isomorphic strings .... this is one way to compare if 2 puzzles are the same - or different.

Most puzzles are produced with the aid of a computor - either by taking a full grid solution - and removing clues, or taking a valid puzzle and removing /adding clues.
This process can make many thousands of puzzles per minute, and the book keeping of these puzzles soon becomes problematic !!!
For clarity only specific hard puzzles have been given specific names.
It has been possible to specify each of the 5 billion solution grids but its too much of a task to even think about their individual puzzles.

[enxio27]

Unless the puzzle is presented as an image (rarely here unless the poster is presenting a puzzle he himself created), I don't think there's any way at all to identify its source. Even with an image it would be difficult unless there is some identifying information in the image itself, or another user is familiar enough with the source to be able to identify the image by appearance alone. If a puzzle is presented as a text grid or in 81-character format, there is NO way to identify its source unless the source presented the puzzle in the exact same text format (such as a Sudoku Web site that presents a puzzle in both image and 81-character formats). In that case, one MIGHT be able to conduct a successful search on the text.

I think it has already been established (long ago) that puzzle data and mere grid patterns cannot be copyrighted; only the presentation (image, etc.) can be copyrighted. I believe we should give credit to puzzle creators whenever possible, but sometimes it's simply not possible.

[jco]

Hello,

Thank you all for your thoughts on this. I should have made the question in more practical terms to get the answers I searched for. I will be concrete. Does some server store information (like SE rating AND the 81 string of numbers) to identify puzzles that are well-known due to some aspect (historical, difficulty level, nice mathematical properties of the pattern, etc) ?
I know this exists in chess, at least for games between players of very high rating.
It is easier for (classical) chess because the starting position is always the same, so when you make moves in some strong chess code with huge database, a very large set of games would be out at every move (the number of possible games played in such position gets quickly very small). In fact, this is the way they know when someone just played a 'novelty' (new move in a known position).
In Sudoku, I remember looking recently at a thread in the Forum that helped me to see that it is a much harder problem.
I see that copyright is not na issue, but to avoid going through something already well-known, it seems necessary at least to be able to quickly identify a small subset of well-known puzzles.
In practical terms, the next three puzzles are quickly identifiable, right?

Hidden Text: Show

Code: Select all

.---------.---------.---------.
| 2  .  . | .  .  . | .  .  1 |
| .  4  . | 7  .  . | .  3  . |
| .  .  9 | .  .  . | 5  .  . |
:---------+---------+---------:
| .  6  . | 5  8  . | .  .  . |
| .  .  . | 3  6  4 | .  .  . |
| .  .  . | .  7  2 | .  8  . |
:---------+---------+---------:
| 5  .  . | .  .  . | 2  .  . |
| .  7  . | .  .  3 | .  4  . |
| .  .  1 | .  .  . | .  .  9 |
'---------'---------'---------'
 

Code: Select all

.---------.---------.---------.
| .  .  . | .  .  . | .  .  3 |
| .  .  1 | .  .  5 | 6  .  . |
| .  9  . | .  4  . | .  7  . |
:---------+---------+---------:
| .  .  . | .  .  9 | .  5  . |
| 7  .  . | .  5  . | .  .  8 |
| .  5  . | 4  .  2 | .  .  . |
:---------+---------+---------:
| .  8  . | .  2  . | .  9  . |
| .  .  3 | 5  .  . | 1  .  . |
| 6  .  . | .  .  . | .  .  . |
'---------'---------'---------'

Code: Select all

+-------+-------+-------+
| . 9 . | . . 1 | 8 6 . |
| . . 8 | . 3 . | 1 . . |
| . . . | 9 . . | . . 3 |
+-------+-------+-------+
| . 3 1 | . 9 . | . . 6 |
| 2 . . | . . 7 | . . . |
| . . . | . . . | 5 . . |
+-------+-------+-------+
| 3 . . | . . . | . 4 . |
| . 6 2 | . 1 . | . . 8 |
| . . . | 2 . . | . . . |
+-------+-------+-------+

It seems to me that there should be a service where one could pay a small fee to register a puzzle in a server (database) that would became well-known for such service (again, thinking on ISBN analogy, it is like getting one for a new book). This is more or less what I had in mind.

Regards,
jco.
[Edit: small correction]

[mith]

For the hardest* puzzles, this thread contains links to a few different lists/databases of puzzles, most notably champagne's ph_1910 database of over 3 million puzzles. It includes the puzzle grid, SE ratings, number of clues, creator, and additional identifiers.

For my own purposes, I have been using a copy of this with added information for minlex form (as coloin discussed), other ratings (q1/q2 from gsf's solver), and some other information relevant to my puzzle generation scripts. Having the minlex form (or some similar standard) is useful for determine if a puzzle is "new" (in the sense of not being able to rearrange it to match one of the puzzles already in the database).

There are other databases out there for specific puzzle types (for example, the database of all known essentially different 17 clue puzzles).

If someone wanted to put in the effort, I'm sure such a database could be created for any given publication - it would just involve a lot of manual entry work for print-only puzzles, and would be prone to error. And certainly, with enough resources, someone could make such a database more accessible online.

*hardest by SE, which is of course only one way to rate puzzles.

[jco]

Hello, mith

Thank you for your comments and link.

Regards,
jco.

[Pupp]

At best, there are puzzles that are so famous they've been given actual names. I don't know where to find of list of named puzzles, but it would be interesting to download those.

[coloin]

but it would be interesting to download those.

see above links

Meanwhile ...these 3 puzzles have not been puplished anywhere in the world [or universe ] ... edit sorry for the spelling mistake !!!!!!!
feel free to name them !

Code: Select all

.........197..4.2.4.5..897.384......9..............345.4.5..1.9.123..456......2.. ED=7.1/1.2/1.2
.........964..3.1.1.5..483.478......6..............768.4.8..2.3.123..456......1.. ED=7.2/1.2/1.2
.........574..8.6.1.6..587.841......6..............731.6.9..3.7.123..456......2.. ED=7.3/1.2/1.2


Code: Select all

+---+---+---+
|...|...|...|
|197|..4|.2.|
|4.5|..8|97.|
+---+---+---+
|384|...|...|
|9..|...|...|
|...|...|345|
+---+---+---+
|.4.|5..|1.9|
|.12|3..|456|
|...|...|2..|
+---+---+---+ 


[mith]

The first one is now named Fred.

[enxio27]

I name the second one Bob.

[coloin]

Code: Select all

.........197..4.2.4.5..897.384......9..............345.4.5..1.9.123..456......2.. ED=7.1/1.2/1.2 Fred
.........964..3.1.1.5..483.478......6..............768.4.8..2.3.123..456......1.. ED=7.2/1.2/1.2 Bob
.........574..8.6.1.6..587.841......6..............731.6.9..3.7.123..456......2.. ED=7.3/1.2/1.2 Rita


of course now when we want to discuss puzzles - there is no confusion.

I popped Rita into SE .... and if you dont want to use uniqeness [I don't} its a bit harder ... just an observation...

[jco]

Hello,

I had a nice time with Rita this morning, testing initially the new (to me) features of SE for writing the my solution path. No uniqueness involved. Thank you for the puzzle!

Regards,
jco
[Edited: to avoid duplicity left solution path only in the linked reference.]