Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Complete tripartite subgraphs in the coprime graph of integers
theorem_1: Sárközy's theorem that for n >= n_0 every A in {1,...,n} with more than f(n,2) elements, the number of integers up to n divisible by 2 or 3, has a coprime graph containing a complete tripartite graph K(1, l, l) with l = floor(c log n / log log log n).
theorem_2: Sárközy's case of Theorem 1 in which A has between 1 and c_1 n elements congruent to 1 or 5 modulo 6: then the coprime graph of A contains K(1, l, l) with l = floor(c_2 log n / log log log n).
theorem_3: Sárközy's case of Theorem 1 in which A, of more than f(n,2) elements, has at least eps n elements congruent to 1 or 5 modulo 6: then the coprime graph of A contains K(1, l, l) with l = floor(c_3(eps) log n).
Sárközy, Gábor N., Complete tripartite subgraphs in the coprime graph of integers. Discrete Math. 202 (1999), 227--238. The copy read for this card is the publisher's typeset version obtained from the author's web page; it prints "© 1999 Elsevier Science B.V. All rights reserved" on its first page, after the abstract and again in its front-matter line.
In the coprime graph G(A) on A subset of {1,...,n}, with edges between coprime pairs, let f(n,k) count the integers m <= n having a prime factor among the first k primes (p. 227). Theorem 1 (p. 229) answers a question from Erdős's last problem collection, quoted on p. 228: there are constants c, n_0 such that if n >= n_0 and |A| > f(n,2) then G(A) contains K(1,l,l) with l = floor(c log n / log log log n), so one class is a single vertex and the two others have that many vertices each. Theorem 1 is deduced from two cases split by the size of |A_{(6,1)}| + |A_{(6,5)}| = s_1 + s_2 (residues 1 and 5 mod 6): Theorem 2 (p. 229) handles 1 <= s_1 + s_2 <= c_1 n and gives l = floor(c_2 log n / log log log n), while Theorem 3 (p. 229) handles s_1 + s_2 >= eps n and gives the much larger l = floor(c_3 log n). The proofs adapt the earlier Erdős–Sárközy method for odd cycles; in the proof of Theorem 2 (Section 2.1, pp. 229-233, with s_1 >= s_2) a vertex a in A_{(6,1)} with large phi(a)/a is the singleton class and the two l-element classes are chosen from A_{(6,2)} and A_{(6,3)}, while the proof of Theorem 3 (Section 2.2, pp. 233-238) works with residue classes modulo the product P_r of the primes not exceeding r. Lemmas 1 to 3 (pp. 230 and 234) are quoted from other papers and get no pages of their own. The author notes (p. 229) the singleton class cannot be improved, since A = {m <= n : 2|m or 3|m} union {5} forces every complete tripartite subgraph to have a class of one vertex, and asks for the best possible l. This is the source for the second question of problem 883.
Source: https://web.cs.wpi.edu/~gsarkozy/papers/paper.html.
Bears on. #883: Theorem 1 gives, for all large and every with more than elements, a in with , which tends to infinity; this answers the problem's second question yes. The paper does not address the first question, on odd cycles.
Results.
- Theorem 1 (p. 229): for and , with ; the page also records the remark (p. 229) that the singleton class cannot be enlarged.
- Theorem 2 (p. 229): the case , with .
- Theorem 3 (p. 229): the case , with .
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.