A Trace-Path Integral Formula over Function Fields

Imagine you are trying to solve a puzzle where two very different worlds—physics and number theory—suddenly start speaking the same language. This paper, written by Yan Yau Cheng, is about finding a specific "translation key" that connects a formula used by physicists to calculate the behavior of particles with a formula used by mathematicians to count points on geometric shapes over finite fields.

Here is the story of the paper, broken down into simple concepts.

1. The Two Worlds: Physics vs. Math

The Physics Side (The "Path Integral"):
In quantum physics, imagine a particle moving from point A to point B. It doesn't just take one straight line; in a way, it takes every possible path at the same time. Physicists calculate the total "probability" of the particle's behavior by adding up a contribution from every single one of these infinite paths. This is called a Path Integral.

If you wrap this path around a circle (like a loop), there is a famous rule in physics: The sum of all these paths (the Path Integral) is exactly equal to the Trace of a specific action.

The "Trace" is like a summary score. If you have a machine that transforms a system, the "Trace" is a single number that tells you how much the machine "stretches" or "rotates" the whole system.
The Analogy: Imagine a spinning top. The Path Integral is like watching the top spin through every possible wobble. The Trace is just the final number you get when you ask, "How much did the top spin in total?" The physics rule says: Sum of all wobbles = Final Spin Number.

The Math Side (The "Arithmetic World"):
Now, switch to number theory. Instead of a spinning top, imagine a geometric shape (a curve) sitting over a "finite field." A finite field is like a clock with only a few numbers (e.g., 0 to 6). On this shape, there are special points called Jacobian points.

Think of these points as tiny dots scattered on a grid.
The mathematician wants to count these dots, but not just by counting them one by one. They want to do it using a "Path Integral" style sum.
The "Action" here isn't energy; it's a pairing of numbers derived from deep number theory rules (Class Field Theory).

2. The Big Discovery

The author asks: Does the physics rule hold in this math world?

Physics Rule: Sum of Paths = Trace of the Action.
Math Question: If we sum up the "arithmetic paths" (which are just the rational points on our shape), does it equal the "Trace" of the Frobenius action (a special math operation that shuffles these points)?

The Answer: Yes! The paper proves that for a specific type of curve, the sum of these arithmetic paths is exactly equal to the Trace of the Frobenius action, with one tiny catch: there might be a plus or minus sign difference.

3. The "Secret Sauce": Determining the Sign

In physics, getting the sign right is often easy or handled by convention. In this math world, getting the sign right is incredibly hard and delicate. It's like trying to guess whether a coin flip will land heads or tails, but the coin is made of pure logic.

Previous mathematicians (Minhyong Kim and Akshay Venkatesh) had found this formula but didn't know the sign. They were stuck with "It equals the Trace, maybe positive, maybe negative."

Yan Yau Cheng's Contribution:
The paper provides the exact formula for the sign. It's not a guess; it's a precise calculation involving:

The shape of the curve (its genus, $g$ ).
A special number called a "regularized determinant" (a fancy way of measuring how much the Frobenius shuffles the points, ignoring the ones that don't move).
A "Legendre symbol" (a mathematical switch that flips between +1 and -1 based on whether a number is a perfect square in the finite field).

The paper says: "Here is the exact sign. It is $(-1)^g$ times this determinant."

4. How They Proved It

The author didn't just guess the sign; they calculated both sides of the equation separately and showed they matched perfectly.

Step 1: The Trace Side. They treated the points on the curve like a quantum system. They built a "Hilbert Space" (a mathematical container for all possible states) using something called a "Theta Line Bundle" (a fancy geometric structure). They then calculated exactly how the Frobenius shuffles the contents of this container.
Step 2: The Path Integral Side. They treated the points as "paths." They summed up the "action" (the pairing of points) for every single point on the curve. This turned out to be a giant sum of complex numbers (like adding up waves).
Step 3: The Match. When they compared the result of Step 1 and Step 2, they found they were identical, provided they used the specific sign formula they derived.

5. Why This Matters (In Simple Terms)

This paper is a bridge. It shows that the deep, mysterious formulas used to describe the quantum universe have a direct, rigid counterpart in the world of numbers and finite fields.

The Analogy: Imagine you have a recipe for a cake in a foreign language (Physics). You find a translation (Math) that says, "If you mix these ingredients, you get this result." But the translation was missing a crucial word: "Add a pinch of salt OR don't." This paper finds that missing word. It tells us exactly when to add the "salt" (the sign) and when not to.

Summary of the Claim

The paper claims that for a curve over a finite field, the sum of arithmetic paths (a discrete sum over points) is equal to the trace of the Frobenius action (a measure of how points are shuffled), up to a specifically calculated sign. This sign depends on the geometry of the curve and the specific way the points are shuffled.

The paper does not claim this has immediate uses in engineering, medicine, or predicting the stock market. It is a pure mathematical discovery that strengthens the analogy between the topology of 3D shapes and the arithmetic of numbers.

Technical Summary: A Trace–Path Integral Formula over Function Fields

Problem Statement and Motivation
This paper establishes an arithmetic analogue of the trace–path integral formulae found in topological quantum field theory (TQFT). In the physical setting, the trace of a monodromy action $F$ on a Hilbert space $\mathcal{H}$ (arising from the geometric quantization of a phase space $M$ ) is equal to a path integral of an action functional $A$ over the space of sections of a fibration $M \times [0,1]/F \to S^1$ .

The author seeks to translate this correspondence to the arithmetic setting of a smooth projective curve $X$ over a finite field $\mathbb{F}_q$ . The paper constructs a dictionary where:

The phase space $M$ corresponds to the $\ell$ -torsion subgroup $J[\ell]$ of the Jacobian $J$ of $X$ .
The space of sections corresponds to the set of rational points $J[\ell](\mathbb{F}_q)$ .
The monodromy action $F$ corresponds to the Frobenius automorphism $\text{Fr}_q$ .
The Hilbert space $\mathcal{H}$ corresponds to the space of global sections of a theta line bundle over the Jacobian.
The action functional $A$ corresponds to a pairing derived from geometric class field theory.

The central problem is to prove that the trace of the Frobenius action on this arithmetic Hilbert space equals a discrete sum (the "arithmetic path integral") of an exponential of the pairing $A$ , up to an explicitly determined sign involving Legendre symbols and regularized determinants.

Methodology
The proof proceeds by evaluating the two sides of the proposed equality independently and then comparing the results.

Trace Side (Section 3):
- Construction of $\mathcal{H}$ : The Hilbert space is defined as $\mathcal{H} = \Gamma(J_Y, \Theta^{\otimes \ell})$ , where $J_Y$ is the lift of the Jacobian to the Witt vectors $W(\mathbb{F}_q)$ and $\Theta$ is the theta line bundle. This space is identified as the unique irreducible representation of the Heisenberg group $H(J[\ell])$ with a specific central character, analogous to the Stone-von Neumann theorem.
- Trace Computation: The action of the Frobenius $\text{Fr}_q$ on $\mathcal{H}$ is analyzed using the machinery of symplectic representations of finite Heisenberg groups (citing [GH09]). The author decomposes the symplectic vector space $J[\ell]$ into $\text{Fr}_q$ -invariant symplectic subspaces.
- Explicit Formula: By computing the trace on these subspaces (handling cases where $\text{Fr}_q$ acts as identity, $-I$ , or has no eigenvalues $\pm 1$ ), the author derives an explicit formula for $\text{tr}(\text{Fr}_q | \mathcal{H})$ . This formula involves the Legendre symbol, the dimension of the fixed-point subspace, and the regularized determinant of $1 - \text{Fr}_q$ .
Path Integral Side (Section 4):
- Defining the Action $A$ : The arithmetic action $A: J[\ell](\mathbb{F}_q) \times J[\ell](\mathbb{F}_q) \to \frac{1}{\ell}\mathbb{Z}/\mathbb{Z}$ is defined via the reciprocity map of geometric class field theory. Specifically, $A(\gamma, \beta)$ is the evaluation of a torsor associated with $\gamma$ on the Frobenius element associated with $\beta$ .
- Relation to Chern-Simons: The paper demonstrates that this pairing $A$ coincides with a function field analogue of the Abelian Arithmetic Chern-Simons pairing defined in [Chu+19]. This involves showing the symmetry of the pairing and its compatibility with Artin-Verdier duality.
- Evaluation: The path integral is computed as a discrete sum $\sum_{\gamma \in J[\ell](\mathbb{F}_q)} e^{2\pi i A(\gamma)}$ . Since $A$ is a non-degenerate symmetric bilinear form, this sum is evaluated as a Gaussian integral over a finite field, yielding a result involving the square root of the number of points and the Legendre symbol of the determinant of the pairing matrix.
Synthesis (Section 5):
- The two independently derived expressions are combined. The author shows that they match exactly, provided the sign is corrected by a specific Legendre symbol term involving the genus $g$ , the regularized determinant of $1-\text{Fr}_q$ , and the determinant of the pairing $A$ .

Key Contributions and Results

Theorem A (Theorem 5.1): The main result states that for a curve $X$ over $\mathbb{F}_q$ and an odd prime $\ell$ with $q \equiv 1 \pmod \ell$ , if $\text{Fr}_q$ acts semisimply on $J[\ell]$ , then:
$\text{tr}(\text{Fr}_q | \mathcal{H}) = \left( \frac{(-1)^g \det'(1 - \text{Fr}_q) \det(A)}{\ell} \right) \sum_{\gamma \in J[\ell](\mathbb{F}_q)} e^{2\pi i A(\gamma)}$
where $\det'$ denotes the regularized determinant (determinant on the quotient by the kernel).
Explicit Sign Determination: A primary contribution is the explicit determination of the sign in the formula. Previous unpublished work by Minhyong Kim and Akshay Venkatesh established the formula up to an undetermined sign under the restrictive assumption that an invariant Lagrangian subspace exists. This paper removes the Lagrangian assumption and provides the precise sign for the general semisimple case.
Regularized Determinants: The paper highlights the appearance of regularized determinants in the arithmetic formula, strengthening the analogy with physical path integrals where such determinants arise in infinite-dimensional settings.
Computational Examples: Section 6 provides explicit computations for elliptic curves over $\mathbb{F}_7$ with $\ell=3$ . These examples illustrate cases where the trace is non-trivial but the path integral is trivial (due to a lack of rational points), and vice versa, verifying the theorem in concrete settings.

Significance and Scope
The paper claims to add to the series of analogies between topology and arithmetic (referencing Mazur and Morishita), specifically by providing a rigorous arithmetic counterpart to the trace–path integral formula in TQFT. The work is framed as a function field analogue of arithmetic path integrals previously studied over number fields.

The author notes that the assumption $\mu_\ell \subset \mathbb{F}_q$ (i.e., $q \equiv 1 \pmod \ell$ ) is necessary for the current construction, as it ensures the Weil pairing takes values in the base field. The paper suggests that generalizing the results to remove this condition is a direction for future research. Additionally, the author outlines potential extensions to number fields (using Iwasawa theory actions as analogues to Frobenius) and non-abelian arithmetic actions, but these are presented as open questions rather than results of this work.