Themabewertung:
  • 0 Bewertung(en) - 0 im Durchschnitt
  • 1
  • 2
  • 3
  • 4
  • 5
Does this work as a puzzle? Graeco-Latin squares
#1
I'm terrible at explaining things as I tend to waffle, so this is going to be difficult, apologies in advance. Please see attachments.

I've got a generator that creates graeco-latin squares (or rather, as the depth is greater than 2, the correct term would be mutually orthogonal Latin squares MOLS)

Using the image on the left which is a 5x5 MOLS square of depth 4, I translated it into the grid in the middle. Each 2x2 box is a representation of each cell of the square, with 1 being cyan, 2 is purple, 3 is red, 4 is green, 5 is orange. The topleft 2x2 of the grid is the topleft cell of the image, so 1 5 4 4 is cyan orange green green. Compare that to the image on the left to see how it's been translated. The next 2x2 is 2 3 2 3 which is purple red purple red, then the next is 3 1 1 1 which is red cyan cyan cyan

What makes a MOLS square is that every pair of orthogonal grids is fully unique. What this means for the grid in the middle is that you can pick any pair (of the 6 pair combinations) in each 2x2 box and it'll be unique compared to the likewise pairs in any other 2x2 box. Also the topleft digits in each 2x2 box together form a latin square (e.g. digit 1 appears once in the topleft box in every row and column of 2x2 boxes). Same for topright, bottomleft, bottomright.

Another way of explaining it, is every 1 in the top left of a 2x2 will have digits 1 to 5 to the right of it exactly once throughout the grid. Every 3 in the top left of a 2x2 will have digits 1 to 5 beneath it exactly once throughout the grid. I've highlighted those examples in yellow and green, but that applies to all 12 likewises pairs across all 2x2 boxes (all four digits in a 2x2 box has three other digits to pair with, hence 12 directional pairs)

Knowing how the grids came to be (if you understand so far, well done!), if you were given the grid on the right *on it's own* without the other parts of the image for solution/context, just the rules, would you be able to fill in the missing cells? Is there enough information there to solve it?


Angehängte Dateien Thumbnail(s)
       
Zitieren
#2
I've answered my own question! I created one in the attachment with the rules boiled down to:
1) In each 5x5 square, each row and column must contain digits 1 to 5.
2) If you were to map all instances of a single digit from one square onto the other squares, the maps must also contain digits 1 to 5.
3) Each square cannot be solved individually in isolation without context from the others.

https://imgur.com/a/qIMqH2a

This shows the solving steps of a 5x5 MOLSudoku. I'm giving it that name coz I can't find any similar puzzles online. Yeah there may be similar puzzles at depth 2 but I took it to depth 4 and reduced to a set of 32 clues, where each individual square cannot be solved independently of the others. In this series of images:

Step 1: Start with the empty puzzle.
Step 2: Fill in as many singles and candidates as you can (rows and columns in each 5x5 must contain digits 1 to 5).
Step 3: Notice that all the 3s are filled in top left, so map them to the others.
Step 4: The maps on the others have to contain digits 1 to 5, so fill in those digits.
Step 5: Disregarding the top half for a moment, the left can be filled in but the right still has multiple solutions.
Step 6: Highlight the 1s on the left, arbitrarily, and map onto the right.
Step 7: Fill in those digits as the map has to contain digits 1 to 5.
Step 8: This square can now be completed.
Step 9: Considering the top half again, the right can be fully completed but the left is still ambiguous.
Step 10: Highlight the 2s on the right, arbitrarily, and map onto the left.
Step 11: Fill in those digits as the map has to contain digits 1 to 5.
Step 12: Complete the rest of the puzzle!

Also the generator I made (with the help of Claude, yes, I'm not an expert by far), can reduce the puzzle down to a critical set of just 4 clues per square, 16 in total, still with one unique solution. I couldn't figure out the multiple layers of logic needed to add many digits so I found those way, way too difficult. I could be proven wrong by a top puzzle solver, but for now I'll stick with 32 total clues.


Angehängte Dateien Thumbnail(s)
   
Zitieren


Möglicherweise verwandte Themen…
Thema Verfasser Antworten Ansichten Letzter Beitrag
  Advice on writing puzzle instructions Egubachu 3 6.993 14.12.2024, 21:43
Letzter Beitrag: Snowflake93
  Use of a "skip cell" constraint in puzzle setting psams 0 5.657 16.07.2024, 19:59
Letzter Beitrag: psams
Information Help with puzzle needed Joseph 3 2.895 01.03.2024, 11:07
Letzter Beitrag: Joseph
  New puzzle creator Twan2797 5 7.012 31.01.2023, 23:22
Letzter Beitrag: Twan2797
  Ice Slide Puzzle Fletcher 10 7.925 26.01.2022, 13:53
Letzter Beitrag: Raumplaner

Gehe zu:


Benutzer, die gerade dieses Thema anschauen: 1 Gast/Gäste