Triangles:
Triangle Arrangement:
It looks like I made a few mistakes in the table. Here is the new version:
The 'solutions' I had for the 4 case actually wouldn't work. Oops.
For the 5 case, I started out with this pattern in all 3 corners: [{O}{OXX}] (oriented so that if they whole 5-triangle was rotated it would appear to be the same.) The pattern on each of the three sides was OXOXXOXXO, and the center was simply filled symmetrically–when I say 'symmetrically' here, I'm referring to a pattern that, when rotated 120 or 240 degrees, appears the same.
While trying to solve other cases, I've noticed that may kind of recurse. For example, take the 8-triangle. If we sort of cut off the outermost row of triangles, we're left with a 5-triangle, which we know is unsolvable. Similarly, if we start with a 5-triangle and get rid of the outermost row, we're left with a 2-triangle, which we already knew was unsolvable. I haven't really done anything with this yet, but it might help...
I've tried working with the 8 case, but it doesn't work with the same method as the 5-triangle (putting the [{O}{OXX}] pattern in the three corners and then trying to fill the rest.) I think it might be better if I divide the 8 into 4 4-triangles, and make the 3 corner triangles rotations of each other such that the center one can be symmetrical... but I'm not sure that will work. Another thing to note, is that the 5-triangle doesn't divide nicely into 4 same-size triangles; so that might have something to do with solvability.
Also, with the 5 case, with the 3 corner patterns and edge pattern fixed, the middle could be filled in any way that was symmetrical, meaning there were a 3 solutions–the center being [{X}{XXX}],[{X}{XOX}], or [{O}{OOO}]. This would seem to indicate that, at least for the center, the ideas of increasing numbers towards the center and having no number over 3 are incorrect.
I also tried forcing solutions that were all 2s or all 1s, and it doesn't seem to work. The problem with the 2s is that the corners all have to be Xs, which makes filling the entire grid in an asymmetrical manner a bit difficult. With the 1s, 0s or 2s sneak in really easily...
Other:
I'll start looking into percentage of solvable/unsolvable in respect to the square case.
Also, another thing I thought of was, what if the cells weren't perfectly aligned (like a typical brick-like pattern)? It might be kind of difficult to notate and not particularly interesting.

Also, my textbook has arrived.
ReplyDelete1) Yay, we'll get you homework problems soon.
ReplyDelete2) Just to make sure we're doing triangles the same way, each X would be counted in 13 triangles if it's far enough from an edge? This makes the shape it influences a non-regular hexagon side lengths (1,2,1,2,1,2). These hexagon's don't tile the plane, like the squares defined in the square grids. Not sure if that's going to make them more solvable...?
3) I need to go back and read your posts in detail, it's great that you're progressing with our limited feedback!
So I started doing it, and I think you didn't count all 13, did you just count overlapping edges? I think your way is a lot easier to play with. And nothing inferior about it. We might want to consider an alternate of the squares that doesn't count diagonals as well. I'm not sure if that would have the same or different fallibility.
DeleteI've checked your 5 case, and agree that it's correct for the edge way of counting, good job.
Looking at the 2 case, it doesn't work following the rules of the 5 case, I bet you changed definitions in the middle, it works with the corners counting...
ReplyDeleteMaybe separate those cases and make separate tables... or pick your favorite for now.
Using the no corners method you can make all 1's in all the even cases and only the 1x1 odd case. But they are symmetrical, so it doesn't help with proving anything.
With the corners it works for the 1 and 5 cases, but maybe no others... Again this isn't really seeming helpful, but I wanted to draw some puzzles and count stuff too.