[Paper Review] A Proof of Fejes Toth's Conjecture on Sphere Packings with Kissing Number Twelve

Thomas Hales|arXiv (Cornell University)|Sep 26, 2012

Mathematics and Applications3 references3 citations

TL;DR

This paper proves Fejes Tóth's conjecture that any three-dimensional sphere packing in which each sphere touches exactly twelve others must be composed of hexagonal close-packed (HCP) or face-centered cubic (FCC) layers. Using a combination of spherical geometry, hypermap classification, and computer-assisted linear programming, the authors show that the only possible kissing configurations with tame contact parameters are the FCC and HCP patterns, thereby confirming the conjecture.

ABSTRACT

In 1969, Fejes Toth conjectured that in Euclidean 3-space any packing of equal balls such that each ball is touched by twelve others consists of hexagonal layers. This article verifies this conjecture.

Motivation & Objective

• To resolve Fejes Tóth's long-standing conjecture on the structure of 3D sphere packings with kissing number twelve.
• To establish that such packings must consist of hexagonal layers, as seen in FCC and HCP arrangements.
• To prove that the only possible local configurations of twelve mutually touching unit spheres around a central sphere are the FCC and HCP patterns.
• To classify all possible contact configurations using tame contact parameters and eliminate non-realizable cases via geometric and linear programming constraints.

Proposed method

• Define a packing as a set of points in R³ with mutual distances ≥2, and a kissing configuration as a set of 12 points on a sphere of radius 2 around a central point.
• Introduce a truncation parameter h₀ = 1.26 and use a weight function L(h) = (h₀ − h)/(h₀ − 1) to bound the total weight of configurations.
• Model the contact graph as a hypermap H = hyp(V, E₂(V)) where edges represent unit-distance contacts between spheres on the sphere S²(2).
• Apply a computer-assisted classification algorithm to enumerate all hypermaps with 'tame contact' parameters, defined by separation constraints (distances 2 or ≥2h₀).
• Use geometric arguments to rule out one hypermap with a hexagonal face (perimeter would exceed 2π, violating geodesic convexity).
• Apply linear programming to show infeasibility of the remaining five hypermaps under angle sum and edge-length constraints.

Experimental results

Research questions

• RQ1 Is every 3D sphere packing with kissing number twelve necessarily composed of hexagonal layers?
• RQ2 Are the FCC and HCP configurations the only possible local arrangements of twelve unit spheres touching a central sphere?
• RQ3 Can non-hexagonal arrangements of 12 kissing spheres exist under the constraints of unit distance and minimal separation?
• RQ4 What are the complete set of combinatorial types of contact hypermaps that can arise from such packings?
• RQ5 Can geometric and linear programming constraints be used to eliminate non-realizable configurations?

Key findings

• The only valid contact hypermaps with tame contact are the FCC and HCP configurations, as proven by elimination of all other candidates.
• A hypermap with a hexagonal face of side length π/3 cannot be realized geometrically because its perimeter would be 2π, exceeding the strict upper bound for geodesically convex polygons on a sphere.
• Linear programming constraints on angles and edge lengths rule out five of the eight candidate hypermaps, confirming only FCC and HCP are realizable.
• The total weight of the weight assignment over all faces is bounded by (4π − 20sol₀) < tgt, supporting the main inequality used in the proof.
• Every node in the contact graph has degree at most four, and only types (0,3), (1,3), and (2,2) are possible when the node type is (p,q,0).
• The FCC and HCP configurations are rigid: their eight equilateral triangles of side length 2 uniquely determine the entire configuration up to congruence.