Wiki
Wiki

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

Updated


Source. Lemma 1, p. 69, of Peter Fishburn, "Convex polygons with few intervertex distances," Computational Geometry 5 (1995), no. 2, 65--93, doi:10.1016/0925-7721(94)00020-v, the edition named on the source card. The paper introduces Lemmas 1--3 as lemmas from Altman, its reference [1] (Amer. Math. Monthly 70 (1963), 148--157), and does not prove them.

Read depth. Claims checked: the definition of a max side and the statement were read clause by clause on the page image. Nothing here is independently reviewed.

Statement

Definition (p. 69). In a convex polygon whose distinct intervertex distances are d1>d2>⋯d_1>d_2>\cdots, a side xyxy is "max" if xy=d1xy=d_1, and "uniquely max" if xy=d1xy=d_1 and no other side or diagonal has length d1d_1. m(C)m(C) is the number of distinct intervertex distances of CC.

Lemma 1 (p. 69, quoted). "If a side of convex nn-gon CC is max, then m(C)⩾n−2m(C)\geqslant n-2. If a side of CC is uniquely max, then m(C)⩾n−1m(C)\geqslant n-1."

Lemmas 2 and 3 (p. 69, also from Altman) describe, in the equality cases m(C)=n−1m(C)=n-1 with a uniquely max side (1,n)(1,n) and m(C)=n−2m(C)=n-2 with a max side (1,n)(1,n), the vertices labelled 1,…,n1,\ldots,n around the perimeter, which of d1,d2,…d_1,d_2,\ldots each chord (k,n−k+1)(k,n-k+1), (k,n−k+2)(k,n-k+2) and (k−1,n−k+1)(k-1,n-k+1) carries (Fig. 3, p. 70).

Proof pointer

Not proved in this paper; cited from Altman [1]. The paper applies the lemma to subpolygons of consecutive vertices that have a longest segment as a side, in Sections 2--5 (for example Lemma 4, p. 72).

Bears on

  • Problem 132: a tool of the convex-position analysis behind Theorem 2; the lemma itself counts distinct distances and says nothing about their multiplicities.