Fractal sets are generated by simple generating formulas (iterated functions) and therefore have an almost zero algorithmic (Kolmogorov) complexity. Yet when observed as data with no knowledge of the iterated function, for instance, when observing pixel values of any region of a fractal image, the fractal set is very complex. It has rich and complicated patterns that appear at any arbitrary level of magnification. This suggests that fractal sets have a rich information content despite their essentially zero algorithmic complexity. This highlights a significant gap between algorithmic complexity of sets and their information richness. To explain this, we propose an information-based complexity measure of fractal sets. We extend a well-known notion of compression ratio of general binary sequences to two-dimensional sets and apply it to fractal sets. We obtain numerical estimates that indicate that the compression ratio depicts complexity of sets in a manner consistent with their observed complexity. For instance, comparing the compression ratio values of several fractal and non-fractal sets shows quantitatively that the fractal sets are more complex.
Joel Ratsaby (Thu,) studied this question.