[Paper Review] Lower Bounds for Pinning Lines by Balls

Otfried Cheong, Xavier Goaoc+1 more|ArXiv.org|Jun 16, 2009

Computational Geometry and Mesh Generation22 references3 citations

TL;DR

This paper establishes a tight lower bound for the Helly number of line transversals to disjoint congruent balls in ℝᵈ, proving that 2d−1 is the minimal number of unit balls required to form a family with no common transversal, despite every subfamily of size 2d−2 having one. The construction uses stable pinning configurations of balls to show that transversals cannot be guaranteed below this threshold, resolving a longstanding question from Danzer (1957).

ABSTRACT

A line L is a transversal to a family F of convex objects in R^d if it intersects every member of F. In this paper we show that for every integer d>2 there exists a family of 2d-1 pairwise disjoint unit balls in R^d with the property that every subfamily of size 2d-2 admits a transversal, yet any line misses at least one member of the family. This answers a question of Danzer from 1957.

Motivation & Objective

• To resolve a longstanding open problem posed by Danzer in 1957 regarding the minimal number of disjoint unit balls in ℝᵈ with no common transversal.
• To establish a matching lower bound for the Helly number of line transversals to disjoint congruent balls in higher dimensions.
• To demonstrate the existence of minimal pinning configurations of size 2d−1 for a line transversal in ℝᵈ, showing that such configurations are necessary and sufficient in extremal cases.
• To show that no stable pinning of a line by fewer than 2d−1 disjoint unit balls can exist, implying that 2d−1 is the minimal size for such configurations.
• To clarify the tightness of the upper bound 4d−1 from prior work by constructing extremal examples that achieve the lower bound.

Proposed method

• Constructs a stable pinning configuration of 2d−1 disjoint congruent balls in ℝᵈ such that a unique line ℓ is pinned and cannot be perturbed to remain a transversal.
• Uses the concept of screens (projected shadows of balls onto a hyperplane orthogonal to the transversal line) to analyze transversal stability and detect pinning.
• Applies topological arguments involving the space of transversals and the Grassmannian to show that configurations with fewer than 2d−1 balls cannot yield stable pinning.
• Employs a compactness argument on the set of 3-dimensional affine subspaces through a fixed line to ensure the existence of such extremal configurations.
• Perturbs the configuration slightly to ensure disjointness and closedness of balls while preserving the pinning property.
• Uses the fact that a family of screens has no strict transversal if and only if the corresponding ball configuration is a pinning, linking geometric and topological properties.

Experimental results

Research questions

• RQ1 What is the minimal number of disjoint congruent balls in ℝᵈ such that no line transversal exists, yet every subfamily of size 2d−2 does admit one?
• RQ2 Can a stable pinning of a line by disjoint balls in ℝᵈ exist with fewer than 2d−1 balls?
• RQ3 Is the upper bound of 4d−1 for the Helly number of line transversals to disjoint congruent balls tight, and what is the corresponding lower bound?
• RQ4 Does the existence of minimal pinning configurations of size 2d−1 imply that 2d−1 is the exact Helly number for transversals to disjoint unit balls in ℝᵈ?
• RQ5 Can the pinning mechanism for lines by balls be generalized to more general convex sets in higher dimensions?

Key findings

• For every d ≥ 3, there exists a family of 2d−1 pairwise disjoint unit balls in ℝᵈ with no common line transversal.
• Every subfamily of size 2d−2 from this family admits a line transversal, proving that 2d−1 is a lower bound for the Helly number of line transversals to disjoint unit balls.
• There exists a minimal pinning configuration of size 2d−1 consisting of disjoint congruent balls in ℝᵈ, where a unique line is pinned and no proper subfamily pins it.
• Any pinning of a line by fewer than 2d−1 disjoint balls in ℝᵈ is unstable, meaning the transversal can be perturbed to remain a transversal.
• The construction confirms that the upper bound of 4d−1 from prior work is tight up to a factor of 2, with the true Helly number lying between 2d−1 and 4d−1.
• The result resolves Danzer’s 1957 question on the existence of such extremal configurations in dimensions d ≥ 3.