Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Erdős et al.: On sum sets of Sidon sets, II
theorem_1: States that for a Sidon set A in {1,...,N} and positive integers N and L, every interval (K, K+L] with K an integer holds fewer than L/2 + 7 L^(1/2) N^(1/4) elements of A + A; with L about 200 N^(1/2) this gives Corollary 1, H(N) < 200 N^(1/2) for large N.
theorem_2: Constructs an infinite Sidon set A such that for all n > n_0 some m at most n has m+1, ..., m+h in A + A with h > n^(1/3)/50; hence H(N) > N^(1/3)/50 for N > N_0.
theorem_3: States that for every finite B_2[g] set A of positive integers and every dimension m, any covering of A by T generalized arithmetic progressions of dimension m has T times their total size above |A|^2/(2^(m+1)g); Corollary 1 is the Sidon case g = 1.
theorem_4: States that for all positive integers g, m and q some finite B_2[g] set A of positive integers has |A| = 2gq and D_m(A) at most |A|^2/(2g), so the quadratic order in Theorem 3 is attained for fixed g and m.
The copy read for this card is the publisher's scan. No copyright line is printed in it; the Springer article page shows "© Hebrew University 1995" and names no license (DOI 10.1007/BF02783214), every other right reserved.
P. Erdős, A. Sárközy and V. T. Sós, "On sum sets of Sidon sets, II," Israel Journal of Mathematics, 90(1-3), 221-233, 1995. https://doi.org/10.1007/bf02783214
Overview
The paper studies two features of sum sets of Sidon sets: runs of consecutive sums and coverings by generalized arithmetic progressions. Its definitions of Sidon and sets use unordered representations (Eq. (1.1), p. 221). For Sidon, Theorem 1 gives the uniform interval bound for all and (Eq. (3.2), p. 223; proof pp. 223–226). The section 3 Corollary 1 consequently bounds the longest run of consecutive sums by for large (p. 223). The proof partitions into short intervals and uses the uniqueness of positive differences to control a sum of squared local counts (Eqs. (3.8)–(3.16), pp. 224–225). Conversely, Theorem 2 constructs an infinite Sidon set with for (Eq. (4.1), p. 226; proof pp. 226–228), by greedily adding pairs whose sum fills the next missing point of a prescribed interval. Thus Eq. (3.1) leaves a gap between the established and scales; the authors’ view that the upper scale is closer is explicitly a remark, not a theorem (p. 223).
For a covering of a finite set by progressions of dimension , section 5 defines as the minimum of , where is the product of its side lengths (pp. 222, 228). Theorem 3 proves for finite sets and all (Eq. (5.5), p. 229; proof pp. 230–231); its section 5 Corollary 1 gives (p. 229). The proof counts unordered pairs within each progression, bounds their sums using , and applies Cauchy’s inequality across the cover (Eqs. (5.7)–(5.10), p. 230). Theorem 4 constructs, for all , sets of size with (Eqs. (6.1)–(6.3), p. 231; proof pp. 231–232), showing the quadratic order is attainable for fixed . Section 7 states unresolved questions about Sidon subsets of high dimensional progressions and coverings by short arithmetic progressions (pp. 232–233). The Freiman covering statement in section 2 and the bound for coverings of squares in Eq. (5.3) are cited background, not results proved here.
Read status. Claims checked: Theorems 1–4 and the two Corollaries 1, with the definitions they use, were read clause by clause on the printed pages. The proofs were read for their structure only.
Results. Theorem 1 (p. 223), the interval bound for sums of a Sidon set, with Corollary 1, ; Theorem 2 (p. 226), the infinite Sidon set with long runs of consecutive sums; Theorem 3 (p. 229), the lower bound for progression coverings of sets, with Corollary 1 for Sidon sets and the application to squares; Theorem 4 (p. 231), the construction showing the quadratic order is attained. The questions of Section 7 are recorded in the digest above.
Relation to E864
Bears on. Problem 864: background only. The paper proves nothing about sets with one repeated sum, and the problem page does not cite it. Its results concern the same objects: Theorem 1 bounds the sums of a Sidon set in an interval, and Theorem 3 bounds coverings of sets in .
E864 permits one sum with , in the paper’s representation convention (Eq. (1.1), p. 221). Theorem 1 does not apply to such sets: its proof, Eq. (3.8), uses that each positive difference occurs at most once, and a repeated sum with repeats the difference . Theorem 3 applies with , which may grow with , and then gives no bound of the order E864 asks for.
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.