Nonexistence of universal Fourier summation formulas for the degree below the Hölder threshold $α=1/3$

Abstract

Let $(σ_{n,\varepsilon})_{n\in\mathbb{Z},\,0<\varepsilon<1}$ be a summation process in the sense of Brezis, that is: \[ \forall \varepsilon\in(0,1),\quad \sup_{n\in\mathbb{Z}}|nσ_{n,\varepsilon}|<\infty \qquad \forall n\in\mathbb{Z},\quad \lim_{\varepsilon\rightarrow0^+}σ_{n,\varepsilon}=1. \] We prove that for every $0<α<1/3$ there exists $f\in C^{0,α}(S^1;S^1)$ such that \[ \sum_{n\in\mathbb{Z}}σ_{n,\varepsilon}n|\widehat f(n)|^2 \] does not converge to $\mathrm{deg}\,f$ as $\varepsilon\rightarrow 0^+$. This gives a negative answer to Open Problem 5.6 from Brezis's list of favourite open problems for all $p>3$ and $0<α<1/3$, leaving the endpoint $C^{0,1/3}$ unresolved. We also observe that, under their literal formulations, Open Problems 5.7 and 5.8 have immediate positive and negative answers, respectively. The proof of the main result combines degree-zero quotients of Blaschke factors with a Baire category argument.

Disclosure

“s partially supported by National Sci- ence Centre grant 2022/01/1/ST1/00021 and by IMAI Centre (University of Warsaw). The author thanks his advisor Katarzyna Mazowiecka for drawing his atten- tion to this problem. Use of generative AI. ChatGPT (OpenAI) was used to generate prelimi- nary drafts of the proofs in this manuscript. The author subsequently verified each argument in detail, revised the proofs where necessary, checked the cited sources, and determined the final formulat”

PDF page 3

Classification

Proof ideas or individual proof-step assistance

Multiplier

8

Verified

2026-08-23T19:06:58.509826Z

Structural counts