So I was messing around with Larger than Life majority rules using the Moore neighborhood, and was wondering what the minimally-sized survivors would be for each range. It turns out that the minimal survivors were not converging to a perfect circle as I had first expected (based on my experience with range-1 cellular automata), but were instead converging to some other shape, which resembled a rounded square rotated by 45°. Here is the minimal survivor for the r=20 case: Since this shape wasn't a perfect circle, but was still the target of a relatively simple convergence, I naturally wanted to find out exactly what shape it was, so I came up with a few apparent properties (assuming center at (0, 0) with radius 1):
- boundary passes through (2/3, 2/3)
- at x=1/3, dy/dx=-1/2 at the boundary (y≈.915)
- (shape area)/(bounding box area)≈.738
Or maybe I'm looking at this wrong, as there seem to be slight cusps present at each of the four "corners" of the rounded square. (It's not all that obvious of an effect, but the shape does seem to curve more strongly than expected when it meets the bounding box.) Perhaps each quadrant is actually part of a different larger curve; it could also be that the proper way to analyze this would be to extend three of the corners of the original square to infinity, before the cellular automaton is ever run. I honestly don't know; that's why I'm asking the board, after all!