Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Claim. Theorem 3 of P. Erdős and A. Gyárfás, A variant of the classical Ramsey problem, Combinatorica 17 (1997), no. 4, 459--467, DOI 10.1007/BF01195000, received 15 September 1996; Crossref dates the issue to December 1997 without a day, so this page is named by the first day of that month. The paper is cited as [EG97] on the problem page. In its notation f(n,4,5)f(n,4,5) is the least number of colors in an edge-coloring of KnK_n in which every 44-clique spans at least five colors, the f(n)f(n) of Problem 136. The theorem states three bounds:

  • f(n)≥56(n−1)f(n)\ge\frac56(n-1) for all n≥4n\ge4;
  • f(n)≤nf(n)\le n for odd nn, by the rotational one-factorization of KnK_n;
  • f(n)≤n−1f(n)\le n-1 for infinitely many even nn, those with n−1n-1 prime.

The paper's introduction, and Bennett, Cushman, Dudek and Prałat (p. 2 of arXiv:2207.02920), state the upper bound as f(n)≤nf(n)\le n without the parity condition. The lower bound had been claimed without proof in Erdős's survey in Congr. Numer. 32 (1981); the paper marks such earlier claims with an asterisk, and Bennett, Cushman, Dudek and Prałat attribute the earlier statement to Erdős, Elekes and Füredi and restate the lower-bound proof in their Section 2. The paper also states f(9)=8f(9)=8 and exhibits an eight-coloring of K9K_9, but prints no proof that seven colors fail.

Covers. The lower half of f(n)∼56nf(n)\sim\frac56n, that is lim inf⁡n→∞f(n)/n≥56\liminf_{n\to\infty}f(n)/n\ge\frac56. The upper bounds settle nothing by themselves; the matching upper half f(n)≤56n+o(n)f(n)\le\frac56n+o(n) is Bennett, Cushman, Dudek and Prałat's ([[problems/extremal_graph_theory/E0136/claims/2022_07_06_bennett_cushman_dudek_pralat|claim page]]) and, by a second proof, Joos and Mubayi's ([[problems/extremal_graph_theory/E0136/claims/2022_08_26_joos_mubayi|claim page]]).

Depends on. Nothing in this wiki.

Acceptance. Refereed publication in Combinatorica, cited with its venue above. Reviewed: the site's curator, Thomas Bloom, credits the bounds 56(n−1)<f(n)<n\frac56(n-1)<f(n)<n and f(9)=8f(9)=8 to Erdős and Gyárfás in the commentary of a problem he labels SOLVED, and took no part in the paper. Bennett, Cushman, Dudek and Prałat and Joos and Mubayi each take the lower bound from this paper. No proof step was checked by this corpus.