Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Ruzsa 1985 note additive bases integers
conjecture_1: Ruzsa and Turjányi's modified form of the Erdős-Graham conjecture, in which the twofold sumset is counted up to 2x against the basis counted up to x; the paper proves the threefold analogue and leaves this open.
theorem_1: Ruzsa and Turjányi's construction, for every order h at least 3, of a basis of density zero whose (h-1)-fold sumset has counting function within a constant factor of the basis's own along a sequence tending to infinity; the case h = 3 answers Problem 337 in the negative.
theorem_2: Ruzsa and Turjányi's theorem that for every basis of density zero the number of threefold sums below 3x is eventually larger than any constant multiple of the number of elements below x; it is deduced from Theorem 3.
theorem_3: Ruzsa and Turjányi's bound on the iterated sumsets of a finite set of integers in terms of the doubling-type constant of its threefold sumset, proved from Ruzsa's 1976 difference-set inequality.
I. Z. Ruzsa and S. Turjányi, A note on additive bases of integers, Publ. Math. Debrecen 32 (1985), 101--104. Received November 28, 1983.
The copy read for this card is an image-only scan of the four printed pages
(physical PDF p. is printed p. ; the first page carries no page
number, the others carry the journal's running heads). Its metadata title is
Pub_Mat_1985_32__1_2_13. It has no text layer; the statements below were
read on the page images of pp. 101--102, with an OCR pass used only to locate
them. Provenance: downloaded in September 2026 from a URL that was not
recorded; 861,413 bytes. No notice is printed on the scanned pages; the
journal's site (https://publi.math.unideb.hu/, read 2026-10-02) states on its
page for authors that "the authors agree to transfer the copyright to the
publisher" and that "the version published in PMD cannot be uploaded to any
repository", every other right reserved.
Read status: claims checked. Theorems 1, 2 and 3 and Conjectures 1 and 2 were read clause by clause on the page images; none of the proofs (pp. 101--103) was checked.
Contents
Notation (p. 101): , is the -fold sumset, and count the elements of and of below . A basis of order is a set of natural numbers whose sums of at most elements include every sufficiently large integer.
- Introduction (p. 101): records the conjecture of Erdős and Graham (1980) that for every basis with , and Turjányi's (1981) counterexamples: for each , a basis of order with .
- Theorem 1 (p. 101): for each some basis of order has density zero, , and for a constant and arbitrarily large , that is, . The construction (pp. 101--102) adds to a basis of order with (printed , a misprint: a basis of order has , and the proof needs only the bound) the integer intervals for a rapidly increasing sequence and an exponent .
- Conjecture 1 (p. 102), as posed: "If is a basis and , then ." The authors motivate it by the example: jumps in a short interval, but sums of two numbers near lie near .
- Theorem 2 (p. 102): every basis with has . It is deduced from Theorem 3 on p. 103.
- Theorem 3 (p. 102): a finite set of integers with satisfies for every . The proof (p. 103) uses the inequality of Ruzsa (1976).
- Conjecture 2 (p. 102; recorded on the page of Conjecture 1): and imply for a function of and alone; the authors expect it to follow from Freiman's theorem with and guess the true order . It would imply Conjecture 1 in the same way.
Compiled scope
The five statements above were checked on the page images; the proofs of Theorems 1, 2 and 3 were not read beyond the pointers given. Nothing here is independently reviewed.
Bears on. #337: the problem asks whether every basis with has ; Theorem 1 with gives a basis of order with and , so the ratio does not tend to infinity for it, and the introduction records Turjányi's earlier counterexamples of every order . Conjecture 1, , is the paper's "modified form" (p. 101) of the conjecture, which the paper leaves open; Theorem 2, , is the threefold variant it proves, deduced from Theorem 3. Neither decides the problem as stated.
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.