🔢 mathematics

On minima of theta and Epstein zeta functions in dimension four

This paper proves that the rescaled root lattice $\mathcal{D}_4$ uniquely minimizes both the theta and Epstein zeta functions among all unit-covolume lattices in dimension four, thereby confirming a 2006 conjecture by Sarnak and Strömbergsson and resolving the Rankin-Sobolev problem in this dimension.

Original authors: Senping Luo, Juncheng Wei

Published 2026-09-30

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Senping Luo, Juncheng Wei

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quiet geometry of the universe, there is a fundamental question about how things pack together. Imagine a vast, empty space where you must place an infinite number of points, arranged in a perfectly repeating pattern, like a grid that stretches forever in every direction. Mathematicians call these patterns lattices. The challenge is to find the specific arrangement that is the most efficient or "optimal" for a given purpose. In the physical world, this question explains why atoms in a crystal settle into specific shapes to minimize their energy, or how electrons arrange themselves to create stable structures. For decades, scientists have known the answer for two-dimensional space (a flat sheet) and for very high dimensions like eight and twenty-four, where the solutions are famous and elegant. However, for the four-dimensional space that sits between our familiar three dimensions and those higher realms, the answer remained a stubborn mystery. The question was not just about which shape looks best, but which mathematical function, representing the total energy of the system, reaches its lowest possible value.

Two researchers, Senping Luo and Juncheng Wei, have finally solved this four-dimensional puzzle. They proved that among all possible repeating patterns in four-dimensional space, one specific arrangement, known as the root lattice D4, is the undisputed champion. This structure is the unique shape that minimizes a specific measure of energy, called the theta function, which sums up the influence of every point in the lattice on every other point. Their work also confirms that this same structure minimizes a related measure called the Epstein zeta function, which describes how the points interact over long distances. Before this proof, mathematicians knew that the D4 lattice was a local winner—it was better than its immediate neighbors—but they could not be certain it was the global winner, the absolute best of all possible shapes. Luo and Wei closed that gap, demonstrating with rigorous certainty that no other arrangement can beat it.

To understand what they did, one must first grasp the nature of the problem. The researchers were looking at a four-dimensional grid where the density of points is fixed; you cannot simply add more points to make the system "better." Instead, you must change the shape of the grid itself. The goal was to find the shape that makes the sum of all the interactions between points as small as possible. In the language of physics, this is finding the "ground state," the most stable configuration a system can achieve. The D4 lattice is a specific, highly symmetric arrangement of points. It is constructed from a set of integer coordinates with a special rule: the sum of the coordinates in any point must be an even number. When scaled to have a standard density, this shape has a unique property where its points are packed as tightly as possible without overlapping, creating a structure that is remarkably balanced.

The path to the solution was not a simple calculation but a complex journey through different regions of possibility. The researchers had to compare the D4 lattice against every other conceivable shape. They knew that for some extreme conditions, the D4 lattice was clearly superior. For instance, when the points are very far apart, the D4 arrangement wins easily. Similarly, when the points are very close together, it also wins. The difficulty lay in the middle ground, the vast region of shapes that were neither extremely stretched nor extremely compressed. In this middle zone, the competition was fierce, and the differences between the D4 lattice and its rivals were incredibly subtle.

Luo and Wei developed a strategy that combined local analysis with global constraints. They first identified two special types of shapes, known as perfect forms, which act as the corners of the mathematical space where all possible lattices live. One of these corners is the D4 lattice itself, and the other is a different shape called A4. They proved that if a lattice is close to either of these corners, the D4 lattice is strictly better, except when the lattice is actually the D4 lattice itself. This eliminated the "neighborhoods" around these corners as places where a rival could hide.

Next, they had to deal with the vast space in between these corners. Here, they used a powerful tool involving the shortest distance between points in a lattice. They showed that if a lattice has a very short distance between its nearest neighbors, it cannot beat the D4 lattice. This ruled out a large class of shapes that were too "tight." Conversely, they proved that if a lattice is too "loose" or stretched out, it also fails to compete. This left them with a specific, manageable region of shapes to examine: those that are neither too tight nor too loose, and not too close to the known corners.

In this remaining region, the researchers employed a sophisticated technique involving projections. They imagined taking a four-dimensional lattice and slicing it along its shortest line, effectively flattening it into a three-dimensional structure. By studying the properties of this three-dimensional slice and the layers of points stacked above and below it, they could derive strict inequalities. They constructed a series of mathematical "guards"—functions that act as lower bounds—to ensure that no matter how the points were arranged in this difficult middle zone, the energy of the D4 lattice would always be lower. They verified these bounds with extreme precision, checking thousands of specific cases to ensure no loophole remained.

The result is a complete and absolute proof. The D4 lattice is not just a local favorite; it is the global minimum. This means that for any four-dimensional system governed by these specific rules of interaction, nature will always choose the D4 arrangement to achieve the lowest energy state. This finding resolves a problem that has been open since the late 1970s, a question that had resisted the efforts of many mathematicians over several decades. It also confirms a conjecture made in 2006 by other leading researchers, who had established the local stability of the D4 lattice but could not prove its global dominance.

The implications of this work extend beyond pure mathematics. In physics, the arrangement of atoms in crystals and the behavior of electrons in certain materials are often modeled using these lattice structures. While we live in a three-dimensional world, the mathematics of four dimensions is crucial for understanding higher-dimensional theories in physics and for solving complex optimization problems in coding and communications. By proving that the D4 lattice is the optimal shape in four dimensions, Luo and Wei have provided a solid foundation for these applications. They have shown that in the hidden geometry of four dimensions, there is a single, perfect order that nature prefers, a structure that is more efficient than any other possible arrangement. The mystery of the four-dimensional minimum is finally solved, revealing a beautiful and precise truth about the geometry of space.

Technical Summary: Minima of Theta and Epstein Zeta Functions in Dimension Four

Problem Statement
The paper addresses the global minimization of the theta function $\Theta(\alpha, L)$ and the Epstein zeta function $E(L, s)$ over the space of full-rank lattices $L \subset \mathbb{R}^4$ with unit covolume. Specifically, it seeks to identify the lattice that minimizes these functions for all $\alpha > 0$ and $s > 2$ , respectively. The functions are defined as:
$\Theta(\alpha, L) = \sum_{v \in L} e^{-\pi \alpha |v|^2}, \quad E(L, s) = \sum_{v \in L \setminus \{0\}} |v|^{-2s}.$
The authors aim to prove that the rescaled root lattice $D_4$ (normalized to unit covolume) is the unique global minimizer up to orthogonal transformations. This resolves the Rankin–Sobolev problem in dimension four, a question with historical roots in the work of Endibaev (1978) and Shushbaev (1978). While Sarnak and Strömbergsson (2006) had previously established the local minimality of $D_4$ , the global minimality remained an open conjecture.

Methodology
The proof employs a combination of analytic number theory, convex geometry, and computer-assisted verification. The strategy involves decomposing the space of lattice shapes into regions where different analytical tools apply, ultimately reducing the problem to a finite set of verifiable inequalities.

Reduction and Compactness: The authors utilize Minkowski reduction to represent lattice shapes by compact sets of Gram matrices. They define the Ryshkov domain $\mathcal{R}$ , the set of positive definite forms with minimum squared length at least 1. The problem is analyzed by normalizing lattices to have minimum length $a(L)$ and studying the behavior of the theta function relative to the perfect forms $D_4$ and $A_4$ .
Determinant Separation: A key geometric step involves proving that outside specific "trace neighborhoods" of the perfect forms $D_4$ and $A_4$ , the determinant of the normalized lattice form is bounded away from the determinant of $D_4$ . This implies that for lattices far from $D_4$ and $A_4$ , the shortest vector length $a(L)$ is bounded above, which in turn provides a lower bound for the theta function via the contribution of the shortest vectors.
Local Comparison via Shell Moments: In the neighborhoods of the perfect forms, the authors use shell moment identities. For $D_4$ , they analyze the distribution of vectors in shells (sets of vectors with equal squared length). By expanding the exponential term in the theta function using Taylor series and applying moment bounds (specifically the first and second moments of the shell vectors), they derive quantitative lower bounds for the difference $\Theta(\alpha, L) - \Theta(\alpha, D_4)$ .
Fourier Minorants and Projection: For the remaining parameter ranges where standard bounds are insufficient, the authors employ a Fourier minorant technique. They project the 4-dimensional lattice along a shortest vector onto a 3-dimensional subspace. Using Laguerre polynomials as a basis, they construct auxiliary polynomials that serve as minorants for the theta difference. The positivity of the Fourier transform of these minorants, combined with Poisson summation, allows for strict comparison.
Computer-Assisted Verification: The proof relies on verifying a finite number of inequalities involving rational coefficients and specific intervals. The authors specify exact rational coefficients for the auxiliary polynomials and parameters (detailed in Appendix B) and use finite interval bounds to ensure the inequalities hold strictly. This constitutes a rigorous computer-assisted proof.
Stationarity and Uniqueness: The proof establishes that any global minimizer must be a stationary point for the function $\Phi(\alpha, L) = \alpha(\Theta(\alpha, L) - \Theta(\alpha, D_4))$ . By analyzing the first variations with respect to scale and shape, they show that the only stationary point satisfying the derived inequalities is the lattice $D_4$ itself.

Key Contributions and Results

Theorem 1.1 (Theta Minimization): For every $\alpha > 0$ , $\Theta(\alpha, L) \ge \Theta(\alpha, D_4)$ for any unit covolume lattice $L \subset \mathbb{R}^4$ . Equality holds if and only if $L$ is orthogonally equivalent to $D_4$ .
Theorem 1.2 (Epstein Zeta Minimization): For every $s > 2$ , $E(L, s) \ge E(D_4, s)$ . Equality holds if and only if $L$ is orthogonally equivalent to $D_4$ . This result follows from Theorem 1.1 via the Mellin transform.
Resolution of the Rankin–Sobolev Problem: The paper confirms that $D_4$ is the global minimizer, settling the conjecture regarding the Rankin–Sobolev problem in dimension four.
Extension of Local to Global: The work extends the known local minimality of $D_4$ (established by Sarnak–Strömbergsson) to a global result, overcoming the difficulty of comparing lattices across the entire parameter space of scales and shapes.

Significance
The paper claims to resolve a long-standing problem in the theory of lattices and quadratic forms. By proving the global minimality of $D_4$ , it completes the picture for dimension four, analogous to the celebrated results for the hexagonal lattice in dimension two, the $E_8$ lattice in dimension eight, and the Leech lattice in dimension twenty-four. The significance lies in:

Mathematical Rigor: Providing a complete proof that combines geometric reduction, analytic estimates, and computer-assisted verification to handle the complexity of the 4-dimensional space.
Physical Relevance: The minimization of theta and Epstein zeta functions corresponds to finding ground states for lattice systems with Gaussian and Riesz potentials, respectively. These models are relevant in statistical physics and chemistry for understanding crystal structures and phase transitions.
Methodological Advancement: The paper demonstrates the efficacy of combining "shell moment" analysis with Fourier minorants and projection techniques to solve high-dimensional minimization problems, offering a template for similar problems in other dimensions or with different potentials.

The authors note that their interest stems from the study of $f$ -potential lattice energies, where Gaussian and Riesz potentials are specific cases. The proof confirms that $D_4$ is the optimal configuration for these specific interaction energies in four dimensions.

More like this