Variations on five-dimensional sphere packings

This paper presents new geometrically distinct five-dimensional sphere packings and a nine-dimensional kissing configuration that match existing optimal records without surpassing them, thereby expanding the known constructions of conjecturally optimal arrangements.

The Gist

Imagine you are trying to pack as many identical oranges as possible into a giant, invisible box. This is the Sphere Packing Problem. But there's a related, slightly different puzzle called the Kissing Problem: If you place one orange in the very center, how many other oranges can you arrange around it so that they all touch the center one without overlapping each other?

In low dimensions (like our 3D world), we know the answers. But as you add more "directions" or dimensions (4D, 5D, 9D, etc.), the problem becomes incredibly complex, like trying to solve a Rubik's Cube that has infinite layers.

This paper by Henry Cohn and Isaac Rajagopal is about finding new ways to arrange these oranges in 5-dimensional and 9-dimensional space. They didn't find a way to pack more oranges than we already knew was possible, but they found new, geometrically distinct patterns that achieve the same record-breaking density.

Here is a breakdown of their discoveries using simple analogies:

1. The "Layer Cake" Strategy

To understand their method, imagine building a tower out of layers of oranges.

• The Old Way: For a long time, mathematicians thought the best way to stack these high-dimensional oranges was to use very specific, rigid layers (like a perfect crystal).
• The New Twist: The authors realized you can take a standard layer, cut it in half, flip one side over, or swap some pieces around, and still get a stable, dense stack. It's like taking a perfect sandwich, swapping the top slice of bread with a slightly different shape, and realizing the sandwich still holds together perfectly.

2. The Five-Dimensional Breakthrough (The "Four Cousins")

In 5D, the record for the "Kissing Number" (how many oranges can touch the center one) is 40.

• The Knowns: For decades, we only knew of two main ways to arrange these 40 oranges: the D5 pattern (very symmetrical, like a perfect snowflake) and the L5 pattern (a bit more twisted).
• The Discovery: Recently, a researcher named Szöllősi found a third way (Q5). In this paper, the authors found a fourth way (R5).
• The Analogy: Think of these four patterns as four different families of cousins. They all have the same number of people (40), and they all fit in the same-sized room, but they stand in different poses.
  • D5 is the "Classic" cousin: Symmetrical and orderly.
  • L5 is the "Reflected" cousin: Like D5 but with a mirror image twist.
  • Q5 and R5 are the "New" cousins: They look different enough that you can tell them apart just by counting how many people are standing opposite each other.

The authors proved that these four are the only "Uniform" arrangements (where every orange sees the exact same neighborhood) that use these specific local patterns. This expands the "menu" of optimal packing options.

3. The "Coloring" Puzzle (How they built the 5D stacks)

How did they build the full 5D packing from these kissing patterns? They used a clever coloring trick.

• Imagine a flat 2D floor covered in a pattern of triangles.
• You have four colors (Red, Blue, Green, Yellow).
• The rule is: If two triangles touch, they must have specific color relationships (e.g., Red can only touch Blue or Green, but never Yellow).
• By following these color rules, they could stack 3D "slices" of oranges on top of each other to build a 5D structure.
• The Result: They found two new "Uniform" 5D structures (where every orange is in an identical spot relative to the whole) that hadn't been seen before. One repeats every 2 layers, the other every 4.

4. Why 6 and 7 Dimensions Failed

The authors tried to use this same "layer swapping" trick in 6 and 7 dimensions, but it didn't work.

• The Analogy: It's like trying to fold a piece of paper into a specific origami shape. In 5D, the paper is flexible enough to fold the way you want. In 6D, the paper is too stiff; if you try to make the same fold, the paper rips or the shape collapses. They suspect new, completely different ideas are needed to solve the packing problem in those dimensions.

5. The Nine-Dimensional Surprise

Finally, they looked at 9 dimensions.

• For 50 years, there was only one known way to arrange 306 oranges around a central one in 9D. It was built using a specific mathematical code (like a secret language of 0s and 1s).
• The Discovery: The authors took that code, swapped a few numbers around (like swapping the 2nd and 3rd letters in a word), and created a brand new arrangement.
• The Difference: The new arrangement is less symmetrical. It has fewer "perfectly opposite" pairs of oranges. Interestingly, this new arrangement is actually "smoother" in a mathematical sense (it has lower energy), even though it doesn't pack any more oranges than the old one.

The Big Picture

Why does this matter if they didn't break the record for "most oranges"?

1. Completing the Map: It shows us that the "optimal" packing isn't just one single rigid structure. There are multiple, distinct ways to achieve the same density.
2. New Tools: Their method of "modifying layers" gives mathematicians a new tool to hunt for solutions in other dimensions.
3. Surprise Factor: The fact that they found a new 5D and 9D configuration suggests that even in dimensions we thought we understood, there are still hidden surprises waiting to be found.

In short, they didn't build a bigger tower, but they discovered new blueprints for building towers that are just as tall and stable as the old ones, proving that the world of high-dimensional geometry is still full of undiscovered patterns.

Technical Summary

Here is a detailed technical summary of the paper "Variations on Five-Dimensional Sphere Packings" by Henry Cohn and Isaac Rajagopal.

1. Problem Statement

The paper addresses two fundamental problems in discrete geometry:

• The Sphere Packing Problem: Determining the maximum density at which congruent spheres can be arranged in $\mathbb{R}^n$ without overlapping interiors.
• The Kissing Problem: Determining the maximum number of non-overlapping unit spheres that can touch a central unit sphere (the kissing number).

While these problems are solved in dimensions 1–4, 8, and 24, they remain open in other dimensions. Specifically, the authors focus on five dimensions, where the optimal kissing number is conjectured to be 40, and nine dimensions, where the record kissing number is 306. The paper investigates whether the known optimal configurations are unique or if other geometrically distinct constructions exist that achieve the same density or kissing numbers.

2. Methodology

The authors employ a "layer modification" strategy, which involves analyzing sphere packings and kissing configurations as stacks of lower-dimensional layers. Their methodology includes:

• Layer Analysis and Reflection: They examine existing configurations (such as the $D_5$ root lattice) by splitting them into layers based on a specific coordinate. They then replace a layer with a modified version (often a reflection of another layer or a different geometric arrangement) to create a new configuration.
• Deep Hole Partitioning: In the context of extending 5D configurations to 6D, they analyze "deep holes" (points furthest from the lattice points) and attempt to partition them into subsets to form new layers.
• Four-Coloring Construction (Fiber Bundles): To construct 5D sphere packings from kissing configurations, they generalize Conway and Sloane's approach. Instead of stacking 4D layers, they stack 3D layers (translates of the $D_3$ root lattice). This requires a four-colored point configuration in $\mathbb{R}^2$ (the base space).
  • Points in $\mathbb{R}^2$ are assigned colors $\{0, 1, 2, 3\}$.
  • Spheres are centered at $\{v\} \times (D_3 + t_i)$, where $v$ is a point in the 2D configuration and $t_i$ is a translation vector corresponding to the color.
  • Constraints on distances between points of the same, adjacent, and opposite colors ensure the spheres do not overlap.
• Computer-Assisted Search: The authors utilize depth-first search algorithms to verify the uniqueness of extensions for kissing configurations and to enumerate symmetry groups. Code for these verifications is provided via DSpace@MIT.

3. Key Contributions and Results

A. Five-Dimensional Kissing Configurations

The authors identify and construct a fourth non-isometric kissing configuration of 40 points in 5D, bringing the total known examples to four:

1. $D_5$: The root lattice configuration (reflection symmetric).
2. $L_5$: Leech's configuration (non-symmetric, constructed by modifying $D_5$ layers).
3. $Q_5$: Szöllősi's configuration (constructed via an $A_4$ cross-section).
4. $R_5$ (New): A new configuration derived by modifying $L_5$. It is constructed by replacing one layer of $L_5$ with the reflection of another, resulting in a configuration with only 12 antipodal pairs.

Theorem 2.1: There are at least four non-isometric kissing configurations of 40 points in five dimensions ($D_5, L_5, Q_5, R_5$).

B. Five-Dimensional Sphere Packings

The authors extend these kissing configurations into families of 5D sphere packings:

• Uniform Packings: They identify two new uniform packings (where the symmetry group acts transitively on the spheres):
  • One based on $Q_5$ (2-periodic).
  • One based on $R_5$ (4-periodic).
• Classification: They prove that there are exactly six uniform sphere packings in $\mathbb{R}^5$ containing these known kissing configurations (1 with $D_5$, 3 with $L_5$, 1 with $Q_5$, and 1 with $R_5$).
• Conjecture 4.2: They propose that every weakly recurrent, optimally dense sphere packing in $\mathbb{R}^5$ can be obtained from a valid four-coloring of an edge-to-edge tiling of $\mathbb{R}^2$ using specific triangles ($\Delta_1$ and $\Delta_2$).

C. Six-Dimensional Limitations

The authors attempted to extend the $Q_5$ and $R_5$ configurations to 6D (aiming for 72 points).

• Result: They found that the necessary "deep hole" subsets required to build the 6D layers do not exist for $Q_5$ and $R_5$.
• Conjecture 3.1: No 6D kissing configuration with $Q_5$ or $R_5$ as a cross-section can contain 72 or more points. This suggests the 5D flexibility does not generalize easily to 6D.

D. Nine-Dimensional Kissing Configuration

The authors successfully constructed a new kissing configuration in 9D with 306 points (matching the known record).

• Method: They started with the unique known configuration (derived from a binary error-correcting code of block length 9) and modified the layer corresponding to the first coordinate equal to 1.
• Modification: They swapped specific coordinates (2 with 3, and 4 with 7) within the layer, breaking the antipodal symmetry.
• Theorem 5.1: There are at least two non-isometric kissing configurations of 306 points in 9D.
• Energy Analysis: The new configuration has fewer pairs of points at the minimal distance compared to the original, implying it has lower Riesz energy for large $s$.

4. Significance

• Completeness of Optimal Lists: The paper demonstrates that even in low dimensions (5 and 9), the classification of optimal sphere packings and kissing configurations is incomplete. The existence of multiple non-isometric configurations achieving the same record density/kissing number challenges the assumption of uniqueness.
• Refutation of Previous Conjectures: The discovery of the $Q_5$ and $R_5$ configurations (and their associated packings) disproves specific formulations of "tightness" proposed by Conway and Sloane and Kuperberg, necessitating new definitions (e.g., weak recurrence).
• New Construction Techniques: The paper introduces a robust framework for generating new packings by modifying layers and utilizing four-coloring of 2D tilings to fiber 3D lattices. This provides a systematic way to explore the space of optimal packings.
• Symmetry and Periodicity: The work highlights the relationship between the periodicity of a packing (1-periodic, 2-periodic, 4-periodic) and its underlying kissing configuration, expanding the known catalog of uniform packings in 5D.

In summary, Cohn and Rajagopal expand the known landscape of optimal sphere packings in dimensions 5 and 9, proving that geometrically distinct solutions exist for the same optimality metrics, and providing a new theoretical framework for constructing them.