Abstract
Given a planar convex body $B$ centered at the origin, we denote by ${\cal M}^2(B)$ the Minkowski plane (i.e., two-dimensional linear normed space) with the unit ball $B.$ For a triangle $T$ in ${\cal M}^2(B)$ we denote by $R_B(T)$ the least radius of a ball containing $T$. This paper establishes a sharp upper bound for $R_B(T)$ based on the side lengths of $T$ and identifies the extremal unit ball as a specific hexagon.
Introduction
In Euclidean geometry, the circumradius of a triangle is uniquely determined by its side lengths. However, in a Minkowski plane ${\cal M}^2(B)$, where distance is defined by a centrally symmetric convex body $B$ acting as the unit ball, the geometry of triangles is more complex. A triangle $T$ might have multiple circumcenters or none at all, and the minimal enclosing ball might not touch all vertices.
This research addresses a fundamental minimax location problem: determining the maximum possible value for the minimal enclosing radius $R_B(T)$ given only the side lengths $(a_1, a_2, a_3)$ of the triangle, without prior knowledge of the underlying norm.
Methodology
The study utilizes a combination of linear algebra and convex geometry to derive geometric inequalities. Key technical steps include:
- Barycentric Coordinates: Used to locate potential circumcenters relative to the triangle's interior.
- Determinant Analysis: Expressing areas of polygons formed by normalized vector differences as determinants to set up geometric inequalities.
- Linear Programming: The core of the proof involves transforming geometric constraints into a linear program. The author then uses the Farkas Lemma to prove that the proposed upper bound is the optimal solution.
- Hexagonal Representation: A lemma characterizing convex hexagons via alternating vertices allows for the identification of the specific unit ball shape that yields the maximum radius.
Key Insights: The Hexagonal Extremum
A central finding is that the "worst-case" geometry for a triangle's circumradius occurs when the unit ball $B$ is a specific type of hexagon whose vertices are determined by the triangle's side lengths. While the lower bound for the radius is simply half the longest side ($R_B(T) \geq \frac{1}{2} \max\{a_i\}$), achieved when the unit ball is a parallelogram, the upper bound is a complex rational function of the side lengths.
The paper also establishes that if the minimal enclosing radius is sufficiently large (specifically, larger than half the triangle's diameter), the circumcenter is unique and must lie strictly inside the triangle.
Key Results
The primary result is a sharp inequality for the minimal enclosing radius $R_B(T)$ in terms of side lengths $a_1, a_2, a_3$:
$$R_B(T) \leq \frac{2a_1a_2a_3}{a_1(a_2 + a_3 - a_1) + a_2(a_1 + a_3 - a_2) + a_3(a_1 + a_2 - a_3)}$$
Additionally, the author provides a corollary that bounds the radius using the triangle's diameter ($\text{diam}_B$) and perimeter ($\text{perim}_B$):
$$R_B(T) \leq \frac{2 \text{diam}_B(T)^2}{6 \text{diam}_B(T) - \text{perim}_B(T)}$$
This result links three fundamental geometric measurements of a triangle in a way that holds true across any possible 2D norm.
Limitations and Future Work
- Dimensionality: The paper is strictly limited to two-dimensional Minkowski planes. Sharp bounds for simplices in $n$-dimensional normed spaces remain an open challenge.
- Smoothness Assumptions: Some existence results rely on the smoothness or strict convexity of the unit ball, which are not present in common norms like $L_1$ (Manhattan) or $L_\infty$ (Chebyshev) distances.
- Future Directions: The techniques could be extended to study other geometric inequalities in Minkowski spaces, potentially involving inradii or higher-order simplices.
Summary updated