Wiki
Wiki

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

Updated


Claim. Let d=2p≥4d=2p\ge4 be even and let tp(n)t_p(n) be the number of edges of the complete pp-partite Turán graph on nn vertices, so that tp(n)=p−12pn2+O(n)t_p(n)=\frac{p-1}{2p}n^2+O(n). Theorem 1 of the paper states that for all n>n0(d)n>n_0(d)

tp(n)+n−p≤fd(n)≤tp(n)+n,t_p(n)+n-p\le f_d(n)\le t_p(n)+n,

with equality fd(n)=tp(n)+nf_d(n)=t_p(n)+n whenever 2d2d divides nn, in the notation of Problem 1085. This determines fd(n)f_d(n) for every even d≥4d\ge4 up to an additive constant, and exactly along the multiples of 2d2d.

Covers. The estimate of fd(n)f_d(n) for every even d≥4d\ge4 and all large nn, to within the additive constant pp. The exact value for every nn is not claimed; Brass's result for d=4d=4 and Swanepoel's for even d≥6d\ge6, on their own claim pages in this folder, supply it. Nothing is claimed for odd dd or for d≤3d\le3.

Depends on. No page of this wiki.

The argument. The lower bound places the points on pp mutually orthogonal circles of radius 1/21/\sqrt2, Lenz's construction, with the n/pn/p points on each circle taken as the vertices of n/(4p)n/(4p) squares, which adds the nn unit distances inside the circles to the tp(n)t_p(n) between them. The upper bound uses a lemma of Erdős and Simonovits, that a graph on nn vertices with tp(n)+n+1t_p(n)+n+1 edges contains the complete (p+1)(p+1)-partite graph Kp+1(1,3,…,3)K_{p+1}(1,3,\dots,3), together with the geometric fact that pp mutually orthogonal planes cannot all be at unit distance from a further point in R2p\mathbb R^{2p}. The library's [[../library/distance_problems/erdos_1967_applications_graph_theory_geometry/_index|card for the paper]] records the theorem and the lemma.

Acceptance. Refereed: P. Erdős, On some applications of graph theory to geometry, Canadian Journal of Mathematics 19 (1967), 968–971. Not reviewed: the site's remarks say that this paper determined fd(n)f_d(n) up to O(1)O(1) for all even d≥4d\ge4, but the site labels the problem OPEN, so the remark is not an acceptance of the problem or of a part.