Wherever the observer focuses, whatever the ripple: the lower side of the logarithmic law of the resolution gap for every window of the class
An observer looks at a periodic ripple through a lens of finite resolution σ and chooses freely where and how sharply to focus. Paper I (doi:10.5281/zenodo.24371172) proved that no choice of focus beats the logarithmic law of the resolution gap, g√ℓ ≥ 1 − ε̄ with ℓ = ln(1/σ), for windows whose ripple is weak, η = sup|R| ≤ 1/ℓ. This paper removes that assumption: the ripple may be as strong as the window class allows, η ≤ 1/4. Main results:- Theorem 1 (every focus, every ripple of the class). For every admissible window, every state and every ℓ ≥ 55.26, g√ℓ ≥ (1 − ε̄)(1 − η)/(1 + η); for ℓ ≥ 960.8 the bound carries the factor √(1 + 10/ℓ). The coefficient (1 − η)/(1 + η) = λ/Υ is the ratio of the two factors of the ripple sandwich of paper H, and it is at least 3/5.- At the resolution the coefficient stays one: (a) + (a′) ≥ 0.49 whenever the ripple reaches the focus, and ≥ 0.1995 otherwise; both are above the Γ_ℓ ≤ 0.1462 that the law asks for (proof by hand).- The true quantity does not show the factor: on windows with η = 1/4 the smallest ((a) + (a′))√ℓ found at ℓ = 55.26 is 3.40, and at those states the factors of the sandwich move it by a few per cent (numerical). The paper conjectures that the coefficient one holds for every ripple of the class. Method: every remainder of the local expansion is a product of six non-negative, increasing, convex functions of η, so it lies under its chord in the gain factor λ = 1/(1 + η); with the target scaled by λ/Υ = 2λ − 1, interval-arithmetic proofs at η = 0 and η = 1/4 cover every ripple in between. Every lemma behind the proof was read again for its use of η; one step of paper I that held only for a weak ripple is repaired by a larger constant, with no new lemma. The computer-assisted parts use decimal interval arithmetic with outward rounding, run with inexact operations trapped. Every claim was first attacked by adversarial searches, and every check reports the deliberately wrong versions it catches. Every statement carries its status: proof, computer-assisted proof, derivation, numerical, or conjecture. The paper also states what remains outside: the class constant ε̄, the upper side, and the general sandwich factors of paper H. Code, stored results and a verification script: https://github.com/neginyan/resolution-gap-every-ripple (archived on Zenodo: doi:10.5281/zenodo.24418549). Related: paper I, doi:10.5281/zenodo.24371172; paper H v1.3, doi:10.5281/zenodo.24369012; paper F, doi:10.5281/zenodo.23856796. Papers F, H and I are not changed by this work.
Authors
- Taishi Namba (ORCID: https://orcid.org/0009-0002-7098-970X)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24418628
- Primary Topic
- Mathematical Analysis and Transform Methods
- Type
- preprint