Wiki
Wiki

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

Updated


Paul Erdős, András Gyárfás and Miklós Ruszinkó, How to Decrease the Diameter of Triangle-Free Graphs, Combinatorica 18 (1998), no. 4, 493--501, DOI 10.1007/s004930050035 (card). The page is dated by the year alone: the paper prints only 1998, with the line "Received October 12, 1997", and Crossref dates every 1998 issue of Combinatorica by its number rather than by a month of publication.

The result. Write h(G)h(G) for the least number of edges whose addition to a triangle-free graph GG gives a triangle-free graph of diameter two. In the discussion before their Problem 4.1 (pp. 498--499), the authors note that the proof of their Theorem 2.3 gives h(G)≤n(2+d+d2) cc(G‾)h(G)\le n(2+d+d^2)\,cc(\overline G) for a triangle-free graph GG on nn vertices of maximum degree dd, where cc(G‾)cc(\overline G) is the least number of cliques covering the edges of the complement. Alon's bound, their Theorem 2.1, gives cc(G‾)=O(d2log⁡n)cc(\overline G)=O(d^2\log n), so h(G)=O(nd4log⁡n)h(G)=O(nd^4\log n); the print drops the factor nn that its own preceding display forces, and the Problem 4.1 page restores it. The paper concludes that maximum degree o(n1/4/log⁡n)o(n^{1/4}/\log n) gives h(G)=o(n2)h(G)=o(n^2). For the problem's fixed ϵ>1/4\epsilon>1/4, the degree bound n1/2−ϵn^{1/2-\epsilon} is o(n1/4/log⁡n)o(n^{1/4}/\log n), so h(G)=O(n3−4ϵlog⁡n)h(G)=O(n^{3-4\epsilon}\log n), which is below δn2\delta n^2 for every fixed δ>0\delta>0 once nn is large. So the answer is yes for every ϵ>1/4\epsilon>1/4.

Covers. Problem 134 for every ϵ>1/4\epsilon>1/4 and every δ>0\delta>0. It settles nothing for ϵ≤1/4\epsilon\le1/4, which Alon's theorem settles.

Depends on. Nothing in this wiki.

Acceptance. Refereed: the paper is published in Combinatorica. The paper says that its topic grew from problems Erdős and Gyárfás studied in 1995, special cases of which Erdős's 1997 problem paper [Er97b] mentions on p. 229 with misprints. The site credits Erdős and Gyárfás with the narrower case of maximum degree ≪log⁡n/log⁡log⁡n\ll\log n/\log\log n, which [Er97b] item 7 reports without proof; the site's label credits Alon, so its commentary is not listed as review of this result.