Wiki
Wiki

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

Updated


Statement

Setting (p. 1). For an integer k≥3k\ge3, a regular (k−1)(k-1)-simplex is a set of kk points that are pairwise equidistant. Sdk(n)S_d^k(n) is the largest number of regular (k−1)(k-1)-simplices spanned by nn points of Rd\mathbb R^d, and Td(n)T_d(n) is the largest number of equilateral triangles determined by nn points of Rd\mathbb R^d, so that Td(n)=Sd3(n)T_d(n)=S_d^3(n) (p. 2). Triangles and simplices of every side length are counted together.

Theorem 2 (p. 2). Let dd and kk be fixed integers with d≥2k≥6d\ge 2k\ge 6, and put r=⌊d/2⌋r=\lfloor d/2\rfloor. Then

Sdk(n)=(rk)(nr)k+o(nk).S_d^k(n)=\binom rk\Bigl(\frac nr\Bigr)^k+o(n^k).

Conjecture 1 (p. 1), which the paper attributes to Erdős's 1994 paper in Math. Pannon. (its reference [12]), quoted: "Any nn points in R6\mathbb{R}^6 can span at most n3/27+o(n3)n^3/27+o(n^3) equilateral triangles, i.e., T6(n)≤n3/27+o(n3)T_6(n)\leq n^3/27+o(n^3)." The paper notes (p. 2) that Theorem 2 with d=6d=6 and k=3k=3 gives it, since r=3r=3 and (33)(n/3)3=n3/27\binom33(n/3)^3=n^3/27. The paper also recalls (p. 1) the Erdős--Purdy lower bound T6(n)≥n3/27−O(n2)T_6(n)\ge n^3/27-O(n^2) from nn points spread evenly over three pairwise orthogonal circles, so in fact T6(n)=n3/27+o(n3)T_6(n)=n^3/27+o(n^3).

For even dd, Corollary 6 sharpens the error term to Θ(nk−1)\Theta(n^{k-1}).

Proof pointer

§ 4, p. 9. The key step is Lemma 17 (p. 9): for integers d≥2k≥3d\ge2k\ge3 and r=⌊d/2⌋r=\lfloor d/2\rfloor, the kk-uniform hypergraph whose vertices are points of Rd\mathbb R^d and whose edges are the regular (k−1)(k-1)-simplices among them contains no copy of Hr+1(k)(3)H_{r+1}^{(k)}(3), the 33-blowup of the hypergraph Hr+1(k)H_{r+1}^{(k)} obtained from Kr+1K_{r+1} by adding k−2k-2 new vertices to each edge (Definition 8, p. 6). A copy would give, by Lemma 15 (p. 8), r+1r+1 pairwise orthogonal affine spaces of dimension at least 22 inside Rd\mathbb R^d, which is impossible since d<2(r+1)d<2(r+1). Mubayi's theorem ex(n,Hr+1(k))=(rk)(n/r)k+o(nk)\mathrm{ex}(n,H_{r+1}^{(k)})=\binom rk(n/r)^k+o(n^k) (Lemma 9, p. 6) and the standard fact that blowing up changes the Turán number by o(nk)o(n^k) give the upper bound (display (3), p. 6). The lower bound is the construction of § 2: rr pairwise orthogonal unit circles with a common center and nn points spread over them as evenly as possible, every two points on different circles being at distance 2\sqrt2. The proof on p. 9 opens "Let r=⌊d/2⌋≥k≥6r=\lfloor d/2\rfloor\geq k\geq 6 be fixed integers" [sic]; the theorem's hypothesis is d≥2k≥6d\ge2k\ge6.

Read depth

Claims checked: the definitions, Conjecture 1, Theorem 2, Lemma 17 and its proof (p. 9) were read clause by clause on the page images of the print (arXiv version 4). Lemmas 9 and 15 were read as statements only. Nothing here is independently reviewed.

Dependencies

Lemma 9 is Mubayi's theorem (the paper's [18]); the blowup estimate is cited as well known (the paper's [17]); Lemma 15 is proved in the paper (p. 8) from Lemma 14 (p. 7).

Source. F. C. Clemen, A. Dumitrescu and D. Liu, The number of regular simplices in higher dimensions, arXiv:2507.19841 (2025), read in version 4 (28 July 2026); see the source card.

Bears on

  • Problem 755: the case d=6d=6, k=3k=3 bounds the number of equilateral triangles of all side lengths together spanned by nn points of R6\mathbb R^6 by n3/27+o(n3)n^3/27+o(n^3); the problem asks for this bound for triangles of side 11 only, which are among those counted, so the theorem gives the problem's bound.