Log in

No account? Create an account
14 January 2010 @ 11:00 pm
Friday Puzzle #32 - 2010 Nurikabe  
Most of my time this new year has been spent writing competition-level sudoku puzzles since this is what I enjoy and am often asked to do. In addition to this weekend's SudokuCup, I'm currently busy getting puzzles ready for the Czech and Italian Sudoku Championships and for Silicon Valley Puzzle Day. With so much of my puzzling time spent on these tasks, and not a lot of time free to conceive of new "Frankenstein-esque" puzzle combinations (as GLMathGrant calls them), I've simplified my Friday Puzzle writing this year into classic styles and just executing nicely themed puzzles. It forms what I will now call the 2010 Project, where I write a 2010-themed puzzle in as many Nikoli styles as I can without playing with the rules (so no 2010 Wacky Heyawake, for example).

One of my big goals/resolutions for this year, well represented by the entries in the 2010 project, is to find some venue (or more likely create my own) to start publishing these types of elegant puzzles more routinely in the US. I'm tried of the label "Japanese Puzzle" being thrown around with inferior, computer-generated puzzles when these types are not constructed with an artful touch by hand as they are meant to be. From Sudoku on up the logic ladder, I'm going to create a forum for the puzzles I would want to be solving.

This week's entry to the 2010 Project is a 20x20 Nurikabe. It uses some less common steps (but then what regular Nurikabe puzzles have so many double-digit islands?). It might qualify more as a Botsu Bako puzzle for that reason alone. Still, enjoy!

Rules: See here.

(Choose view image to get a larger form)

zundevilzundevil on January 15th, 2010 06:26 am (UTC)
Well, that was a workout!

I eventually got to the point where it looked like *a* solution was attainable and I went running after it. And it worked, so hooray. But as the author -- how did *you* make sure you only ended up with one solution? I imagine it follows a well-defined logical path, but man -- the 20 on the top and bottom both progressed somewhat by "feel" instead of pure logic.

Good show. I did the LITS and Slitherlink tonite as well, and look forward to more 2010 puzzles in the coming weeks.
motrismotris on January 15th, 2010 06:30 am (UTC)
I guess I've practiced "squish" logic a lot; there are logical steps there to finish up the puzzle if you look for them.

It still ends up being a much easier solve to use feel though at the end, which is one reason I had my favorite tester go over it before I posted to make sure I didn't screw it up.
MellowMelonMellowMelon [wordpress.com] on January 15th, 2010 07:09 am (UTC)
Nurikabe today? One of us is apparently psychic.

Count me in with the group that had to feel out parts of the solution. In my case it was getting started with the top left. I think with some more time I could have rigorously shown it was the only way, but my intuition refused to believe it could have resolved otherwise, so I went ahead and put it down.

It was the bottom right/center had me stumped for the longest. Eventually I did find the relevant squish logic, as you call it.

I was most impressed with the amount of "global" logic needed to put this one away, since from a constructor's standpoint I've found it awfully difficult to include such steps in a puzzle.

I'm wondering when you're going to make one of these puzzles a size of 20 by 10. :P
motrismotris on January 15th, 2010 07:15 am (UTC)
The upper-left was specifically pinned to just that one instance and that wasn't too hard to do (but its one of those design steps where its very hard for a solver to see). The middle-right was the really hard area to resolve and had several versions and island movements to get just right.

Also, I suppose 20x10 will make sense for some future puzzle, but I bet 30x67 appears sooner.

Edited at 2010-01-15 07:18 am (UTC)
(Anonymous) on January 15th, 2010 01:38 pm (UTC)
If it's any consolation, the mid-right was probably equally as difficult for the solver to prove logically as it was for you to constrain uniquely! I'm not sure about the label "squish" logic though, I'd say it has an awful lot in common with some numberlink tricks - I remember having a play with them in a couple of my puzzles.

I'm interested to see how you'd tackle a 2010 themed sudoku. Although if it's a 20x20 version I wouldn't attempt it!

motrismotris on January 15th, 2010 04:30 pm (UTC)
Well, "squish" here to me means what I suppose it does in Numberlink - these two tracks surround some third track (an island or the ocean) that must also escape around a corner. This constrains both the land and the sea in certain squares as a whole block of things must escape around a choke point. It turns out you aren't drawing contiguous groups, but you can go through many individual cells in many different places, proving that if a certain cell isn't what it ends up being, you will have squished the necessary island or the connectivity of the ocean out of existence.

Amazingly, I'd say I'm fairly comfortable showing things like this puzzle have the single solution, but not at all comfortable (yet) doing it on Numberlink.

I've already tackled a 2009 themed sudoku (although it violated Nikoli's symmetry rules in the strictest sense and wouldn't qualify here) but this gives a feel for what it would be. Here it would have to almost be a 16x16 or 25x25 with 2010 just in the given patterns, repeated twice for rotational symmetry to obey Nikoli rules, and neither of these interests me as much in constructing so I won't be doing sudoku, even for the sake of completion of "all possible types".

Edited at 2010-01-15 04:45 pm (UTC)
(Anonymous) on January 18th, 2010 03:41 pm (UTC)
Very enjoyable indeed.

One of your puzzles' best attributes are the use of "global constraints" as you have called them before in a prior comment from a prior puzzle.

I think that I am the turtle here, as I only solve based upon certainty in the logic as an exclusive solution. You folks keep your "squishing". I'll keep my slogging.

Robert Hutchinsonertchin on January 20th, 2010 05:00 am (UTC)
Coming to this post-Mystery-Hunt.

I had one false start that required scrapping, but did mostly okay with the squish logic. I had to actually fill in a few what-if squares rather than imagine their consequences, but I did a good job of spotting the ones that wouldn't work, and quickly proving same.
(Anonymous) on January 21st, 2010 08:26 pm (UTC)
Great puzzle. I didn't find myself being squishy at all, actually, but there were many times that I *wanted* to. At one point in the bottom-right I used the fact that there was a unique solution to knock out two cases (possibly the first time I've done so in nurikabe), but otherwise it felt like a traditional solve.
motrismotris on January 21st, 2010 08:28 pm (UTC)
Well, using uniqueness does not prove 1 answer here, and the "squishy" logic - i really should have used a better name now that everyone is following up with the term - is needed to actually show there is just one. Glad you liked the puzzle.
(Anonymous) on February 22nd, 2017 04:05 pm (UTC)
Fun puzzle !
..36:24..Had great fun solving this .Lovely puzzle !