⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕
-
-
-
Filed under: let's solve it one number at a time
-
-
You can only post every 10 seconds. Boo.
-
Edit: forgot to grey previous number
As much as I'd like to go on, I have work to do. :|
-
Filed under: What, that's still 1 number added
-
Are there special rules for the mega-sudoku, or is it just 13 sudokus at once? If anything, overlapping squares makes it easier, since it gives you more information to fill them in.
-
@Kian said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
is it just 13 sudokus at once
This. But the overlapping squares don't have any clues.
-
I was going to post just the 5s in the bottom/right segment, but I figured that would be confusing as to how I got there, so I filled in the 1/3s that forced the 5s, and then finished off the center block since it was down to one number
-
Filed under: I might be too bored
-
-
@devjoe You have got to be kidding me.
-
@izzion My contribution in blue, top right corner.
-
And because I couldn't resist doing more than one digit...
-
@PleegWat When I test-solved this, I wrote a custom sudoku solver that allowed me to apply the incoming constraints from neighboring grids in each corner, and which generated a set of solutions which could appear at each position in the grid (the 8 distinct grids that appear on different rows, and special cases at the edges). I then reduced each solution to its four 3x3 corners, for matching purposes.
A second program compared solutions of one grid with solutions of neighboring grids, eliminating ones which don't match any neighbor solution, taking into account that middle grids could have edge or non-edge neighbors. This program iterated the process since the eliminations could cause further eliminations. This got the list down to about 2 to 10 solutions at each position in the grid.
One of the grids at the lower right had only two solutions, and my manual inspection one of those did not work at the bottom (it worked with a solution that could have appeared at that spot in a higher row), so there was a unique solution for the pair of grids. This caused the next pair of grids to the left of those to also have a unique solution, and I wrote a third program to iterate this, across the bottom and then left to right above that, snaking through the entire section of 8x45.5 sudoku grids, and determining the bottom constraints that applied to the next instance of the grids across the bottom.
This allowed me to see it led to a progression of values, and I could then, by plugging in various bottom values, figure out what it was actually doing, and find the answer. You can see some of the details of what it does in the posted solution document. It took too long, and the test-solving feedback led to the distinct red and green coloring of the givens along the bottom edge as a way of eliminating some time I wasted not noticing something important there, but it was still not a puzzle most teams could realistically be expected to solve within the time constraints of Mystery Hunt.
-
@coderpatsy I expected to find more, but work gets in the way…
-
I've been nerdsniped…
Three weeknights, maybe 5 hours starting from scratch.
-
@DCoder I'm starting to worry you spent that time writing a macro to solve it
-
-
@kazitor said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
I'm starting to worry you spent that time writing a macro to solve it
I'm not worried. I just think it's missing the point.
-
@kazitor said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
@DCoder I'm starting to worry you spent that time writing a macro to solve it
No macros. No tricks.
Just my numpad and these wits.
-
@PleegWat Thanks, but I think MIT's Regex Crossword is more interesting.
That's right, I touch regex voluntarily. I'm a madman.
-
@DCoder said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
I touch regex voluntarily
-
@DCoder I seem to recall a topic on that one a few years ago. Probably here, nowhere else I frequent is that nuts.
-
@PleegWat said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
else I frequent is that nuts
Hmmmm.... is that intelligible enough for the
COCK, er,QOOC thread...
-
@TimeBandit regex are good for you.
-
@DCoder Given how many different superfluous ways of fucking up the solution to the same problem modern software provides, I'm surprised that there aren't any real regex competitors out there.
-
@pie_flavor said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
I'm surprised that there aren't any real regex competitors out there
You can write all the matchers and so on with string manipulation functions. It's just that doing so is even more annoying than writing regular expressions (except for a few trivial cases such as whole string equality, or basic glob matching).
Speaking of which, file globs are exactly an alternative to REs. They're vastly more limited, but good enough to cover a ton of common cases.
-
@pie_flavor said in ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕ ⎕:
I'm surprised that there aren't any real regex competitors out there.
We use a homegrown alternative in our ruling engine. It is much more limited, but since it is part of a rules engine which has a goto construct it ends up being more powerful in practice.