Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Obryant 2022 size finite sidon sets
theorem_1: O'Bryant's explicit form of the Erdős–Turán bound: for every k >= 1 the least diameter s_k of a k-element Sidon set satisfies s_k >= k^2 - 2k^(3/2) + k + sqrt(k) - 1, restated as Theorem 3 together with R_2(n) < n^(1/2) + n^(1/4) + 1/2 for every n >= 1.
theorem_2: O'Bryant's main theorem: a k-element Sidon set has diameter at least k^2 - 1.99405 k^(3/2) once k is sufficiently large, and the largest Sidon set in {1,...,n} has fewer than n^(1/2) + 0.99703 n^(1/4) elements once n is sufficiently large.
theorem_4: O'Bryant's identity for a finite Sidon set A and a positive integer T: the diameter of A equals |A|^2 T^2 divided by T(T + |A| - 1) - (2S(A,T) + V(A,T)), minus T, where S counts the missing small differences and V is the variance of the window counts.
Kevin O'Bryant, On the size of finite Sidon sets. arXiv:2207.07800 (2022); the version read is v2 (arXiv stamp 25 July 2022; the title page is dated July 27, 2022). The arXiv record names arXiv's non-exclusive distribution license (arXiv:2207.07800), every other right reserved.
The paper studies s_k, the least diameter of a k-element Sidon set, and R_2(n), the largest size of a Sidon set inside {1,...,n}. Theorem 1 (p. 2), restated with an R_2 bound as Theorem 3 (p. 3), gives the explicit bound s_k >= k^2 - 2k^{3/2} + k + sqrt(k) - 1 for every k >= 1, by a simplified form of the Erdos-Turan windowing argument, and R_2(n) < n^{1/2} + n^{1/4} + 1/2 for every n >= 1. Keeping the equalities in that argument gives Theorem 4 (p. 5), an identity expressing the diameter of a Sidon set through the variance of its window counts and its missing small differences. Theorem 2 (p. 2) is the main improvement: if k = |A| is sufficiently large, a Sidon set A has diameter at least k^2 - 1.99405 k^{3/2}, against the 1.996 of Balogh-Furedi-Roy, and R_2(n) < n^{1/2} + 0.99703 n^{1/4} for sufficiently large n, against their 0.998. The method refines the Erdos-Turan argument alone, without Lindstrom's, and the paper describes it as logically simpler than Balogh-Furedi-Roy's but computationally more involved. The author tracks the secondary constant through b_k, defined by s_k = k^2 - b_k k^{3/2}, and b_infinity = limsup b_k (p. 2). The paper does not address whether the n^{1/4} term can be removed.
Read status: claims checked for the results linked below, statements read clause by clause on the printed pages of arXiv v2; no proof is checked step by step.
Source: https://arxiv.org/abs/2207.07800.
Bears on.
- #30: Theorem 2 gives h(N) < N^{1/2} + 0.99703 N^{1/4} for sufficiently large N, and Theorem 3 gives h(N) < N^{1/2} + N^{1/4} + 1/2 for every N >= 1; both keep an N^{1/4} term and do not answer the question.
Results.
- Theorem 1 (p. 2): For k >= 1, s_k >= k^2 - 2k^{3/2} + k + sqrt(k) - 1; restated as Theorem 3 (p. 3), which adds R_2(n) < n^{1/2} + n^{1/4} + 1/2 for n >= 1.
- Theorem 2 (p. 2): For a finite Sidon set A with k = |A| sufficiently large, diam(A) >= k^2 - 1.99405 k^{3/2}; for n sufficiently large, R_2(n) < n^{1/2} + 0.99703 n^{1/4}.
- Theorem 4 (p. 5): For a finite Sidon set A and a positive integer T, diam(A) = |A|^2 T^2 / (T(T + |A| - 1) - (2S(A,T) + V(A,T))) - T, with S the weighted count of missing differences below T and V the variance of the window counts.
- Inequality (1) (p. 1): the Erdos-Turan windowing inequality n - 1 >= R_2(n)^2 T/(T + R_2(n) - 1) - T for all positive integers n, T, which the paper rederives in the proof of Theorem 3 (pp. 3--4).
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.