[pnm]

.. working on improving my Tectonic (/ Suguru) generator:

https://www.instagram.com/p/CjDgT-isIKo/



edit: and please retweet if possible: https://twitter.com/calcudokudotorg/sta ... 8662556672

[pnm]

It is interesting to keep removing clues and see if the puzzle is still doable:

[pnm]

I was working on the code that computes a valid set of numbers, given a Tectonic / Suguru grid,
and kept running into a "blocked" configuration, where putting the final number would produce an invalid set.

After trying to improve the code for a while, I decided to take a closer look at the grid:



10 points if you can spot the problem

[pnm]

Here's one where in one step I used the fact that the puzzle must have a unique solution (!)

[pnm]

...

10 points if you can spot the problem

A similar problem with this one:

[rafaelhoukes]

That is indeed a funny problem. Perhaps you can prevent it by not allowing the L- or U-shape at all, since they will always cause this problem.

[pnm]

That is indeed a funny problem. Perhaps you can prevent it by not allowing the L- or U-shape at all, since they will always cause this problem.

Yes, that's one of the first things I added yesterday

There's also a quick check that fills each cage with numbers,
and checks if any neighbouring cell has become "impossible" because of it.

It's interesting because it makes you wonder if these cage layouts have certain properties.

Looking at other such puzzles online, there do seem to be a lot of "expansive" cages
(ones that cover more rows and columns). I also wonder if having more "boxy" cages
means you're forcing the numbers to alternate in a more regular pattern (if the puzzle
is possible at all).