In standard Euclidean geometry, the relationships between a triangle's side lengths, area, and enclosing radius are rigid and familiar. But what happens when we step into a world where distance is measured differently? In a general normed space—such as a Minkowski plane or a high-dimensional Banach space—the "unit ball" (the set of all points at distance one from the origin) does not have to be a perfect sphere. It can be a square, a hexagon, or any symmetric convex shape.
When you warp the definition of distance, you warp the geometry of triangles. A triangle that looks equilateral in one norm might have wildly different side lengths in another. Despite this variability, a powerful line of research demonstrates that we can establish universal, norm-independent bounds and geometric constants by studying these simple three-sided figures. Triangles and simplices act as fundamental probes, capturing the intrinsic structural properties of non-Euclidean spaces regardless of the specific norm applied.
The Geometry of Warped Distances
To understand why this works, imagine drawing a triangle on a sheet of paper, but measuring its perimeters and area using a square unit ball instead of a round one. As you rotate or stretch the triangle, its properties fluctuate based on how its edges align with the corners of the unit ball.
This variability makes proving universal geometric inequalities challenging. However, Averkov et al. tackled this problem by looking for the absolute minimum and maximum area a triangle can occupy given specific edge lengths across all possible symmetric convex unit balls. The authors discovered a complete system of sharp inequalities bounding the relationship between a triangle's perimeter, diameter, and area. They proved that some rules are completely unbreakable: for instance, a triangle's perimeter is always strictly between two and three times its diameter, no matter what norm you choose to measure it with.
Euclidean Norm (Round Ball) Alternative Norm (Square Ball)
/\ /\
/ \ / \
/____\ /____\
[Rigid Relations] [Warped Relations]
But both obey: 2 × Diameter ≤ Perimeter ≤ 3 × Diameter
Enclosing the Triangle: Norm-Independent Radii
A classic geometric problem is finding the circumradius—the radius of the smallest ball that can completely enclose a triangle. In Euclidean space, this is a straightforward function of the side lengths. In an arbitrary normed space, the enclosing ball must adapt to the shape of the norm's unit ball, making the circumradius seem unpredictable.
Averkov solved this by demonstrating that the minimal enclosing radius in any Minkowski plane can be bounded by a rational function of its side lengths alone. This means that if you know the relative distances between three vertices, you can calculate a universal limit on the distance to a central facility, completely independent of the underlying norm. The spatial extent (diameter) and the sum of the edge lengths (perimeter) tightly constrain the maximum possible distance from a central point to the furthest vertex.
Fingerprinting Spaces with Geometric Constants
If triangles are constrained by the spaces they inhabit, we can invert this relationship: we can use triangles to diagnose and classify the geometric properties of abstract, infinite-dimensional Banach spaces.
Two critical properties of a Banach space are rotundity (how curved the unit ball is) and uniform non-squareness (a measure preventing the unit ball from behaving like a flat-sided square). To quantify these traits, researchers design "geometric constants"—numerical fingerprints of a space.
Chen et al. introduced novel geometric constants ($M_1(X), M_2(X), M_3(X)$) derived specifically from the midlines of equilateral triangles inscribed in the unit sphere of a Banach space. By analyzing how these midlines distort, they established new criteria for space classification. For example, the authors proved that if the constant $M_1(X) < 2$, the underlying space is guaranteed to be uniformly non-square.
In a parallel approach, Chen et al. extended the classic James constant—which measures the orthogonality of vector pairs—by expanding its domain to vectors bound to specific geometric configurations. Their new constants, $J_1(X)$ and $J_4(X)$, incorporate metric terms that evaluate how "square" or "smooth" the space is, providing sharp thresholds ($J_1(X) < 3/2$ or $J_4(X) < 2$) that guarantee uniform non-squareness.
The Universality of Triangular Embeddings
The strategy of reducing complex, abstract structures to the geometry of triangles and simplices extends far beyond functional analysis. Because triangles are the simplest non-trivial polygons, they serve as an ideal bridge between abstract metrics and concrete geometric intuition:
- Markov Chains: Devriendt revealed that Kemeny's constant ($K$), a fundamental metric dictating how fast a Markov chain mixes, can be calculated entirely through the geometry of a simplex constructed from the chain's commute times. The author proved that $K = R^2 - ||\gamma - l||^2$, where $R$ is the circumradius of the simplex and $||\gamma - l||$ is the distance between its circumcenter and its Lemoine point.
- Spherical Partitioning: In data clustering and computational complexity, Heilman et al. resolved the 3-dimensional Propeller Conjecture by analyzing the moments and vertex angles of spherical triangles, proving that a specific three-cluster partition maximizes geometric stability.
- Algorithmic Analysis: Even in data structures, Gupta showed that the runtime of greedy binary search trees can be bounded by reformulating the access sequences geometrically, translating algorithmic execution into paths in a geometric plane.
Ultimately, whether analyzing the mixing time of a network, the complexity of a clustering algorithm, or the abstract structure of a Banach space, the underlying mechanism remains the same. By bounding the behavior of triangles, we uncover universal geometric truths that survive even when our standard notions of distance are stripped away.
Key Foundational Work
On area and side lengths of triangles in normed planes (2009) — contributed 2 concept elements to this direction.
Circumradius versus side lengths of triangles in linear normed spaces (2007) — contributed 2 concept elements to this direction.