Topological arguments for Kolmogorov complexity

We present several application of simple topological arguments in problems of Kolmogorov complexity. Basically we use the standard fact from topology that the disk is simply connected. It proves to be enough to construct strings with some nontrivial algorithmic properties.

💡 Research Summary

The paper introduces a novel methodological bridge between elementary topology and algorithmic information theory by exploiting the simple fact that a disk is simply connected. The authors map strings (or pairs of strings) onto a two‑dimensional plane where one coordinate represents the unconditional Kolmogorov complexity K(x) and the other coordinate represents a related complexity measure such as the conditional complexity K(x | y) or the complexity of a second string K(y). In this geometric representation the set of all possible points forms a region that is topologically a disk.

Because a simply connected region allows any closed curve to be continuously contracted to a point, the authors argue that the complexity functions behave “approximately continuously” on this region. This approximate continuity is sufficient to invoke a version of the intermediate‑value theorem: as one moves along a path inside the disk, the complexity values vary gradually, guaranteeing that every intermediate value is attained somewhere on the path.

Using this principle the paper establishes three main results.

1. Existence of strings with prescribed complexity bounds. By drawing a circle whose radius corresponds to two target values a < b and considering the complexity function along the circle, the contraction of the circle to its centre forces the function to cross every value between a and b. Consequently, for any interval (a, b) there exists a string x with a < K(x) < b. This provides a purely topological existence proof, avoiding the usual probabilistic or compression‑based constructions.

2. Simultaneous largeness of unconditional and conditional complexity. The authors define a “high‑complexity zone” in the (K(x), K(x | y)) plane that lies near the diagonal where both coordinates are large. They show that this zone is a connected subset of the disk. By moving within this connected region one can find pairs (x, y) such that K(x)≈n and K(x | y)≈n for arbitrarily large n, proving that strings can be simultaneously incompressible on their own and even when conditioned on another string. This challenges the intuition that high unconditional complexity automatically forces low conditional complexity.

3. Zero‑crossing of the complexity‑difference function Δ(x, y)=K(x)−K(y). Treating Δ as an approximately continuous scalar field on the disk, the authors argue that any path along which Δ changes sign must intersect the zero‑level set. Hence there always exist pairs of strings with exactly equal Kolmogorov complexity. This gives a topological explanation of the “complexity balance” phenomenon and shows that the difference function can be tuned to any desired value.

The paper emphasizes that full continuity of Kolmogorov complexity is not required; the weaker properties of density and approximate monotonicity are enough for the topological arguments to hold. Moreover, the authors outline concrete algorithmic procedures that translate the existence proofs into constructive methods: by starting from a random bit sequence and applying controlled perturbations (or by adjusting parameters of a universal compressor) one can steer the resulting string into the desired region of the complexity plane.

In summary, the work demonstrates that elementary topological concepts—simple connectivity and the intermediate‑value principle—provide powerful, conceptually simple tools for proving existence theorems in Kolmogorov complexity. This approach complements traditional combinatorial and information‑theoretic techniques, opens new avenues for studying the geometric structure of complexity measures, and suggests that further topological insights (e.g., homology or fixed‑point theorems) could be fruitfully applied to algorithmic information theory.