Let B be the set of positive integers whose Collatz orbit reaches 1, and τ(n) the total stopping time. On B, τ satisfies a central limit theorem, but it is not locally Gaussian: in ℓ¹, the histogram of τ on B ∩ (1, X) is a Gaussian times a periodic profile Ω(k − log₂X) of period log₂3, built from the law ν of the terminal residue R(n) = Σ log₂(1 + 1/(3u)) over the odd values u of the orbit. Every such profile, on B or on an invariant subset of positive density, is an average of translates of the sawtooth (log 3/2)·2z mod log₂3. Hence the total variation along clocks is at most 1/2 and the ℓ¹ distance to slowly varying sequences at most 2 − √3, with equality exactly when the terminal law modulo log₂3 is a point mass. Given positive predecessor density, the starts whose orbit passes through 2j approach both bounds as j → ∞, without numerical input. On B, a slowly varying approximation exists only if ν is uniform modulo log₂3. A computer-assisted residue bound excludes this: the distance is at least 0.00718, and a positive proportion of central clocks is empty in every window [X, (1+δ)X) with δ < 0.254. That proportion is given exactly by the gaps of the support of ν. Builds on A Gaussian law for Collatz total stopping times (Zenodo 23046896). Companion paper: Brownian motion in Collatz orbits. The single numerical input is documented in the computational supplement.
No takes yet. Share an insight, caveat, or question.
David Leen (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: