Huh, that’s a really cool theorem. Someone should make a distillation of this post with cool fractal visualizations (see e.g. Chaos Games for some idea of the relation).
Sure. The image is kind of a diffuse cloud of points. Its actually dense on the plane. But I can still kind of plot it by gradually fading points out. The transforms I used are shifting right by 0.1 and rotating clockwise by 1 radian. Ie this image is K, st shift_right(K) ∪ Rotate_clockwise(K)=K
Huh, that’s a really cool theorem. Someone should make a distillation of this post with cool fractal visualizations (see e.g. Chaos Games for some idea of the relation).
Sure. The image is kind of a diffuse cloud of points. Its actually dense on the plane. But I can still kind of plot it by gradually fading points out. The transforms I used are shifting right by 0.1 and rotating clockwise by 1 radian. Ie this image is K, st shift_right(K) ∪ Rotate_clockwise(K)=K
That does not look like it will be mapped into a subset of itself if I shift it right by 0.1.
Here is a link (as lesswrong doesn’t seem to support animated gifs in comments)
https://i.imgur.com/CHXYOZf.gif
Note that the points are gradually fading out, if you shift a bright point right by 0.1, you find a slightly fainter point.