Wiki
Wiki

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

Updated


Statement

Setting (pp. 1--3, 5). er(S)e_r(S), fd(n)f_d(n) and ud(n)u_d(n) are as in Theorem A. For x∈Sx\in S, D(x)D(x) is the largest distance from xx to a point of SS; the furthest neighbour digraph is the one determined by r=Dr=D, and gd(n)g_d(n) is the maximum of eD(S)e_D(S) over nn-point S⊂RdS\subset\mathbb{R}^d. Md(n)M_d(n) is the maximum number of unordered pairs at distance diam⁡(S)\operatorname{diam}(S) in an nn-point S⊂RdS\subset\mathbb{R}^d.

Lenz configurations (p. 5). For even d≥4d\ge4 let p=d/2p=d/2, split Rd=V1⊕⋯⊕Vp\mathbb{R}^d=V_1\oplus\cdots\oplus V_p orthogonally into planes, and let Ci⊂ViC_i\subset V_i be the circle about the origin of radius rir_i, with ri2+rj2=λ2r_i^2+r_j^2=\lambda^2 for all i≠ji\ne j (so every ri=λ/2r_i=\lambda/\sqrt2 when d≥6d\ge6). An even-dimensional Lenz configuration for the distance λ>0\lambda>0 is a finite subset of a translate v+⋃iCiv+\bigcup_iC_i. For odd d≥5d\ge5 let p=⌊d/2⌋p=\lfloor d/2\rfloor, take V1V_1 of dimension 33 and the other ViV_i of dimension 22, replace C1C_1 by the 22-sphere Σ1\Sigma_1 in V1V_1 of radius r1r_1, with the same condition on the radii (all equal to λ/2\lambda/\sqrt2 when d≥7d\ge7); an odd-dimensional Lenz configuration is a finite subset of a translate of Σ1∪⋃i≥2Ci\Sigma_1\cup\bigcup_{i\ge2}C_i. Its associated partition S1,…,SpS_1,\ldots,S_p is the trace of SS on the pp pieces.

Theorem B (p. 5). For every d≥4d\ge4 there is n0∈Nn_0\in\mathbb{N} such that:

  1. If S⊂RdS\subset\mathbb{R}^d and r ⁣:S→(0,∞)r\colon S\to(0,\infty) satisfy ∣S∣=n≥n0|S|=n\ge n_0 and er(S)=fd(n)e_r(S)=f_d(n), then rr is identically some c>0c>0 and SS is a Lenz configuration for the distance cc. The one exception is d=4d=4 with 8∣n−18\mid n-1, where also possible is: for some a∈Sa\in S and c>0c>0, S∖{a}S\setminus\{a\} is a Lenz configuration for the distance cc on two circles C1,C2C_1,C_2 both of radius c/2c/\sqrt2, aa is their common centre, each Ci∩SC_i\cap S is the vertex set of (n−1)/8(n-1)/8 squares inscribed in CiC_i, r≡cr\equiv c on S∖{a}S\setminus\{a\} and r(a)=c/2r(a)=c/\sqrt2.
  2. If S⊆RdS\subseteq\mathbb{R}^d satisfies ∣S∣=n≥n0|S|=n\ge n_0 and eD(S)=gd(n)e_D(S)=g_d(n), then r≡diam⁡(S)r\equiv\operatorname{diam}(S) and SS is a Lenz configuration for the distance diam⁡(S)\operatorname{diam}(S).

In particular (p. 5), fd(n)=2ud(n)f_d(n)=2u_d(n) and gd(n)=2Md(n)g_d(n)=2M_d(n) for all d≥4d\ge4 and n≥n0(d)n\ge n_0(d).

The paper presents Theorem B as a corollary of Theorem C and says (p. 5) that the extremal digraphs are exactly the sets maximizing ud(n)u_d(n) (respectively Md(n)M_d(n)) for nn large in terms of dd, with an exceptional construction when d=4d=4 for all sufficiently large n≡1(mod8)n\equiv1\pmod 8.

Proof pointer

Section 6, pp. 12--16. Theorem C leaves an extremal pair a scaled Lenz configuration with r≡1r\equiv1 off a set S0S_0 of o(n)o(n) points; let TT be the points of S0S_0 with r≠1r\ne1, ∣T∣=k|T|=k. Lower bounds for ud(n)−ud(n−k)u_d(n)-u_d(n-k) and Md(n)−Md(n−k)M_d(n)-M_d(n-k) (Lemma 7, p. 13, from the exact values in Lemmas 5 and 6 and from the Lenz structure of extremal unit distance sets, Theorem 3) are compared with the at most k(n−k)k(n-k) edges between TT and the rest. This rules out k>0k>0 at once for d≥6d\ge6; for d=4,5d=4,5 the points of TT are pinned down geometrically, leaving only the centre in dimension 4 (and only for favourite distances, with 8∣n−18\mid n-1 from the extremal unit distance configurations of Brass and van Wamelen) and nothing in dimension 5. With TT empty, Theorem 3 (cited, p. 5) makes SS a Lenz configuration.

Read depth

Claims checked: the definitions and Theorem B were read clause by clause on the page image of p. 5 of the arXiv preprint, and the proof in Section 6 was read for structure. Theorem 3 and Lemmas 5 and 6 are cited, not proved, in the paper and were not read at their sources. Nothing here is independently reviewed.

Dependencies

Theorem C. External inputs named by the paper: Theorem 3 (Brass 1997; Swanepoel, Unit distances and diameters in Euclidean spaces, 2009), Lemma 5 (Brass, van Wamelen) and Lemma 6 (Swanepoel 2009).

Source. K. J. Swanepoel, Favorite distances in high dimensions, in Thirty Essays on Geometric Graph Theory (J. Pach, ed.), Algorithms and Combinatorics 29, Springer, New York, 2013, 499--519; read in the arXiv preprint arXiv:1108.4817 (24 August 2011), whose labels and pages are used here; see the source card.

Bears on

None directly. The paper's bound behind Problem 754 is Theorem A; Theorem B describes the extremal sets for f4(n)f_4(n) only for n≥n0(4)n\ge n_0(4) and gives no explicit n0n_0.