Last night I constructed two Nikoli-style Yajilin puzzles. The 17×17 was constructed first, while the 10×10 was added later. This pair emulates willow's style, who will always post puzzles in pairs (one smaller, one larger). I don't expect to continue following this pattern, but it is an interesting idea.
Smaller puzzle interface (puzz.link)
Larger puzzle interface (puzz.link)
I wrote this puzzle almost two months ago and completely forgot about it! This genre is quirky but pretty neat.
Rules: Shade some dominoes of cells. No two dominos may touch orthogonally, but each domino must touch exactly two others by the corners such that all dominoes form a single loop. Cells separated by a square must both be shaded or both be unshaded. Cells separated by an X must not (i.e. X marks overlap exactly one shaded cell).
Another Numberlink, this time monster sized!
I originally used pzprrt to attempt to prove uniqueness as always, but after 4+ hours of running it had yet to find a second solution. (My original attempt was non-unique, and pzprrt took 2 hours to find a second solution.) After posting the puzzle on PuzzleSquareJP anyway, someone linked me to this program. It's an incredible solver, somehow claiming uniqueness in only an hour! Given how old the program is I'm surprised it's not more well-known outside Japan. Maybe that's because nobody in the west constructs Numberlink puzzles at the rate Nikoli does.
Pzprrt is probably more convenient in most cases because its puzz.link implementation makes up for the time it would take to input the solution into the above external program. But it's nice to know that constructors have been using programs for 15 years to verify uniqueness; it makes the constant output from Nikoli make more sense.
It's been a while since I posted here. Have a Numberlink! This was also posted to PuzzleSquareJP.
