Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Statement
Theorem 1 (p. 421). "Let ; be two sequences of integers. Assume that the products are all distinct. Then for some absolute constant
"
The paper introduces it on p. 420 as display (4), "Erdös conjectured and Szemerédi proved that then [Szemerédi (to appear)]", and writes: "First of all we give a simpler proof of (4), which nevertheless uses many of the ideas of the original proof." The same page conjectures (5) and shows by the construction (6) (the 's the primes in , the 's the integers up to with all prime factors at most , ) that is attainable; the paper says that (5), if true, is best possible.
Source. P. Erdős and A. Szemerédi, On multiplicative representations of integers, J. Austral. Math. Soc. Ser. A 21 (1976), no. 4, 418--427; Theorem 1 on printed p. 421 (PDF p. 4 of the scan), proof on pp. 421--423 (PDF pp. 4--6), read on the page images.
Read depth. Claims checked: the statement, display (4), the conjecture (5) and the construction (6) were read clause by clause on the page images of pp. 420--421. The proof was read for its structure (below) and not checked step by step.
Proof pointer
Pages 421--423. Write , . A prime is associated with if at least members of are multiples of , and with if at least members of are. Deleting the multiples of every prime not associated with (and repeating), and likewise for , leaves with members and with members, all of whose prime factors are associated with , respectively , since . It suffices to prove (10) . Let () be the largest integer for which more than primes in are associated with both and , and these primes (11). The pairs (12) with , number more than (13), and they are distinct: two equal pairs taken at primes , with common quotients and , would give the cross products with different indices, against the distinct-products hypothesis. From above: by the maximality of at most primes of each higher dyadic block are associated with both, so (14) the sum of their reciprocals is less than ; the quotients are free of the primes not associated with , and of the primes not associated with (15); Brun's method bounds the two counts by (16) and (17), Mertens's theorem with (14) gives (19) , so the number of pairs is less than (20) . Comparing with (13) gives , hence (10).
Dependencies
Brun's sieve in the form used in (16)--(17) and Mertens's theorem (19), quoted without proof; the bound . External premises are taken at statement level; none was checked here.
Bears on
- Problem 490: the statement is the problem's inequality for with all products distinct; the site attributes the theorem to Szemerédi's 1976 paper, whose main theorem this library files, and this paper gives a second, simpler proof. The conjecture (5) and the construction (6), filed as Conjecture (5), bear on Erdős's question about the limit of .