Abstract
Motivated by the problem of counting finite BPS webs, we count certain immersed metric graphs, tripods, on the flat torus. Classical Euclidean geometry turns this into a lattice point counting problem in $\mathbb{C}^2$, and we give an asymptotic counting result using lattice point counting techniques.
Introduction
In the study of flat geometry, a fundamental problem is counting "simple" objects such as closed geodesics. On a torus, this task is equivalent to counting primitive lattice points within a circle, a problem with roots in Gaussian mathematics. However, modern theoretical physics—particularly the study of BPS webs in supersymmetric field theories—demands the analysis of more complex structures.
This paper focuses on "tripods": metric graphs consisting of a central vertex (the Fermat point) and three legs meeting at $120^\circ$ angles. When embedded in a flat torus, these legs can wrap around the surface in intricate ways. The authors provide the first precise asymptotic formula for the number of such primitive tripods as a function of their total length $R$.
Geometric Framework and Methodology
The transition from geometry to number theory involves several sophisticated steps. First, the authors leverage classical Euclidean geometry, specifically the work of Toricelli and Viviani on Fermat points. A critical insight is that the total length of a tripod is equivalent to the side length of an equilateral triangle constructed on its endpoints.
To count these objects, the tripod is "lifted" from the torus (a quotient space) to the complex plane. Since a tripod is defined by the relative positions of its three endpoints, the problem is naturally parameterized in four real dimensions, or $\mathbb{C}^2$. This transformation allows the authors to treat the problem as one of counting pairs of lattice points $(z, w)$ subject to specific angular and length constraints. The lead constant of the growth rate is then derived through a volume calculation in $\mathbb{R}^4$ using a volume-preserving coordinate transformation.
Key Results
The central contribution of this work is Theorem 1.1, which establishes the asymptotic growth of primitive tripods. For any unit covolume lattice $\Lambda$, the number of primitive tripods $N_\gamma(R, \Lambda)$ with length at most $R$ satisfies:
$$\lim_{R \to \infty} \frac{N_\gamma(R, \Lambda)}{R^4} = \frac{15\sqrt{3}}{4\pi^3} \approx 0.2095$$
This $R^4$ growth rate is significantly faster than the $R^2$ growth observed for simple closed geodesics, reflecting the increased degrees of freedom in the tripod's configuration.
Insights and Discussion
Several key observations emerge from this analysis:
- The $\zeta(4)$ Connection: Just as 2D lattice counting involves $1/\zeta(2)$, the parameterization of tripods in $\mathbb{C}^2$ introduces a $1/\zeta(4)$ factor due to the primitivity condition.
- Lattice Independence: Surprisingly, the asymptotic growth rate is independent of the specific shape of the unit-area lattice $\Lambda$. Whether the torus is square or highly skewed, the long-term growth remains constant.
- Primitive vs. Reduced Tripods: The authors distinguish between "primitive" tripods (not scaled versions of others) and "reduced" tripods (legs containing no lattice points). While these are asymptotically equivalent for generic lattices, special symmetric lattices (like the triangular lattice) show a higher frequency of non-reduced tripods.
Conclusion and Future Work
While this paper provides a definitive answer for the torus (genus 1), extending these results to higher-genus surfaces presents significant challenges due to the presence of "cone points" or singularities. Furthermore, tripods represent the simplest class of BPS webs; future research may explore counting more complex networks, such as "H-shaped" graphs, which would necessitate lattice point counting in even higher-dimensional spaces.
Note: The authors verified their theoretical results using SageMath simulations. For a length $R=35$, numerical experiments on Gaussian integers yielded 312,488 tripods, matching the prediction within 0.1%.
Summary updated