Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Conlon 2023 homogeneous structures subset sums non averaging
theorem_1_6: The 2023 polynomial improvement of the Erdős–Sárközy upper bound for non-averaging sets, the intermediate step between (n log n)^{1/2} and the sharp n^{1/4+o(1)} of Pham and Zakharov, with the paper's account of the earlier bounds.
David Conlon, Jacob Fox, Huy Tuan Pham, Homogeneous structures in subset sums and non-averaging sets. arXiv preprint (2023). arXiv:2311.01416.
The copy read for this card is arXiv:2311.01416v1 (2 November 2023, 34 pages), whose pagination is used here; no later arXiv version and no journal record were found on 2026-09-18 (arXiv API record; Crossref bibliographic query for the title), so the paper is cited as a preprint. For a set or sequence of integers, is the set of subset sums (p. 1); a generalized arithmetic progression (GAP) is proper if its sums are distinct and homogeneous if divides (p. 2). Theorem 1.4 (p. 2): for each integer there are constants such that, whenever has elements, the subset sums include a proper homogeneous GAP of some dimension with at least elements, the homogeneous form of a theorem of Szemerédi and Vu (Theorem 1.3). Theorem 1.5 (p. 3), the main technical result: for and there are such that if has size with and , then some of size at least lies, with , in a proper GAP of dimension at most , and some of size at most has containing a homogeneous translate of the proper GAP . Theorem 1.6 (p. 5), the application: there is a constant such that a subset of in which no element is the average of two or more other elements has (the abstract's ), the first polynomial improvement of the Erdős--Sárközy bound of 1990. The introduction (pp. 1--4) records the history of the non-averaging function : Straus's , the Erdős--Straus bound through the function (two subsets of whose subset sums share no nonzero element; ), Abbott's and , Bosznay's construction , , giving , Erdős and Sárközy's from the Freiman--Sárközy theorem, and the authors' earlier , sharp for . Section 2 develops the tools (approximation of dense sets by GAPs, stability), Section 3 proves Theorem 1.5, Section 5 proves Theorem 1.4 from a variant of it (Theorem 3.1) and the convex geometry of Section 4, and Section 6 (pp. 31--33) proves Theorem 1.6, whose deduction is outlined on p. 5.
Read status: claims checked, in the text layer, for the definitions, Theorems 1.4, 1.5 and 1.6 and the introduction's account of the earlier bounds (pp. 1--5); the proofs were not read, apart from the opening of Section 6 (pp. 31--32), read on the page images for the result page's proof pointer. Result page: theorem_1_6. The digest that stood here before 2026-09-18 was written from the abstract alone and named only problem 789.
Source: https://arxiv.org/abs/2311.01416. The arXiv record names arXiv's non-exclusive distribution license (arXiv:2311.01416), every other right reserved.
Bears on. #186: Theorem 1.6 (p. 5, text layer), for non-averaging , the upper bound the site's commentary places between Erdős--Sárközy's and Pham--Zakharov's , which supersedes it; the paper's is the problem's , and its introduction (p. 4) is the attestation on record of Straus's, Erdős--Straus's, Abbott's and Erdős--Sárközy's bounds, none of whose papers is held. Bosznay's paper, the paper's reference [6] (Acta Math. Hungar. 53 (1989), 155--157), is filed as bosznay_1989_lower_estimation_non_averaging_sets; its Theorem, for all sufficiently large , and the construction behind the reported here are on printed p. 155 (PDF p. 1), read there on the page image and paged on theorem. #789: the paper's subject, homogeneous progressions in subset sums, is the structure behind the site's cross-reference; it states no bound for that problem's (subsets whose subset sums determine the number of summands), which is a different function from the non-averaging above.
Results to transcribe.
- Theorem 1.4 (p. 2): for elements of the subset sums contain a proper homogeneous -dimensional GAP of size at least for some .
- Theorem 1.5 (p. 3): the structure theorem quoted above, the paper's main technical result.
- Theorem 1.6 (p. 5): for every non-averaging (theorem_1_6).
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.