Riemann Hypothesis for Drinfeld Modules

• The paper establishes a Hasse–Weil bound on the zero distribution of L-functions linked to Drinfeld modules over global function fields.
• It employs module-theoretic and valuation-theoretic methods to constrain Satake parameters and verify the predicted functional equations.
• The findings offer new insights into prime equidistribution and trace formulas, bridging analogues of classical RH with t-motive theory.

The Riemann Hypothesis for Drinfeld modules establishes a Hasse–Weil type bound on the zero distribution of L-functions and zeta functions associated to Drinfeld modules over global function fields. This analogue of the classical Riemann Hypothesis arises within the arithmetic of function fields and is substantiated by module-theoretic and valuation-theoretic methods. Notably, the proof furnished by Micheli demonstrates that, for any Drinfeld module of arbitrary rank, the local factors of the associated L-series possess roots constrained precisely as predicted by the function field analogy of the classical hypothesis (Micheli, 13 Dec 2025).

1. Global Function Fields and Valuations

Let $\mathbb{F}_q$ denote a finite field of $q$ elements. A global function field $K$ is a finite extension of the rational function field $\mathbb{F}_q(T)$. The ring of integers of $K$ is $A = \mathbb{F}_q[T]$, and the distinguished place at infinity, $\infty$, corresponds to the zero of $1/T$ on $\mathbb{F}_q(T)$. For nonarchimedean places $v$ of $K$, the normalized valuation extends the degree-valuation at infinity: $$v_\infty(f/g) = \deg(g) - \deg(f), \quad (f, g \in A).$$ Two valuations $v, w$ on $K$ are equivalent if $v = c w$ for $c > 0$, and likewise for their induced absolute values. Every nonarchimedean absolute value derives from a unique place of $K$, per standard function field theory (cf. Stichtenoth 1.3.1).

2. Drinfeld Modules: Construction and Arithmetic

Given a finite field extension $k = \mathbb{F}_q^n$, a Drinfeld module of rank $r$ over $k$ is defined by an $\mathbb{F}_q$-algebra homomorphism

$\phi: A = \mathbb{F}_q[T] \rightarrow k\{\tau\},$

where $k\{\tau\}$ is the twisted polynomial ring acted on by the Frobenius automorphism, $\tau a = a^q \tau$. For Drinfeld modules, the prototype polynomial is

$$\phi_T = \tau^r + g_{r-1} \tau^{r-1} + \cdots + g_1 \tau + g_0,$$

with $g_{r-1}, \ldots, g_0 \in k$ and nonzero leading coefficient. As $q$-polynomials, these enact a generalized “exponential map” when $T$ is inverted (Goss, Thm. 4.2.8). The module’s characteristic is the minimal monic $p \in A$ annihilated by $g_0$.

For a prime $\ell \ne p$ of $A$, the corresponding $\ell$-adic Tate module is

$$T_\ell(\phi) = \varprojlim \phi[\ell^n] \cong (A_\ell)^r,$$

which is subject to a Frobenius endomorphism $\pi = \tau^n \in \operatorname{End}(\phi)$. The associated characteristic polynomial

$$P_{\phi, \ell}(T, x) = x^r + a_{r-1}(T)x^{r-1} + \cdots + a_0(T),$$

is in $A[x]$ and invariant under the choice of $\ell$ (Papikian Thm. 3.6.6).

3. Zeta and L-Series: Definitions and Local Factors

For every prime $\mathfrak{p} \subset A$ of good reduction, the local factor of the L-series is

$$P_{\phi, \mathfrak{p}}(x) = \det(1 - \pi_{\mathfrak{p}} x \mid V_\ell(\phi)) = \prod_{i=1}^r (1 - \alpha_{i, \mathfrak{p}} x),$$

where the $\alpha_{i,\mathfrak{p}}$ are the local eigenvalues (“Satake parameters”). Two equivalent forms are used:

• The Weil zeta function:

$$Z(\phi, u) = \prod_{\mathfrak{p} \subset A} P_{\phi, \mathfrak{p}}(u^{\deg \mathfrak{p}})^{-1}$$

• The L-series:

$$L(\phi, s) = Z(\phi, q^{-s}) = \prod_{\mathfrak{p}} \left(1 - a_\mathfrak{p} q^{-s \deg \mathfrak{p}}\right)^{-1},$$

where $$a_\mathfrak{p} = \sum_{i=1}^r \alpha_{i, \mathfrak{p}}$$.

4. Functional Equation and Completed L-Function

The completed L-function is defined as

$$\Lambda(\phi, s) = \Gamma_\infty(\phi; s) L(\phi, s),$$

with $\Gamma_\infty(\phi; s)$ the “infinite-place” factor constructed via Goss’s gamma-function in positive characteristic. This function satisfies the functional equation

$\Lambda(\phi, s) = W(\phi) \Lambda(\phi, 1-s),$

where $W(\phi) \in \{\pm 1\}$ is computable from local data at $\infty$ and the conductor. The Hasse–Weil zeta function, expressible as

$Z(\phi, u) = \frac{P(\phi, u)}{(1-u)(1-qu)},$

with $P(\phi, u)$ a polynomial of degree $r-1$, satisfies

$$Z(\phi, q^{-1} u^{-1}) = \epsilon q^{-(r-1)} u^{-2} Z(\phi, u),$$

leading to $\Lambda(s) = \epsilon \Lambda(1-s)$ under the change of variables $s = -\log_q u$.

5. Riemann Hypothesis for Drinfeld Modules

The Riemann Hypothesis in this context asserts that all zeros of $\Lambda(\phi, s)$ lie on the line $\mathrm{Re}\, s = \frac{1}{2}$, equivalently that

$$|\alpha_{i, \mathfrak{p}}| = (q^{\deg \mathfrak{p}})^{1/2}$$

for every Satake parameter $\alpha_{i, \mathfrak{p}}$. The zeros of the numerator polynomial $P(\phi, u)$ for $Z(\phi, u)$ reside on the circle $|u| = q^{-1/2}$. Micheli’s Theorem 1.1 formalizes these assertions for rank $r$ Drinfeld modules over $k = \mathbb{F}_{q^n}$: all roots $\alpha$ of the Frobenius characteristic polynomial satisfy

$|\alpha|_* = q^{n/r}.$

Moreover, for the characteristic polynomial

$$P_{\phi, \ell}(T, x) = x^r + \sum_{i=0}^{r-1} a_i(T)x^i \in A[x],$$

the coefficients obey $\deg_T a_i \leq (r-i)n/r$ and $(a_0) = (p^{n/d})$ for $d = \deg p$.

6. Outline and Methodology of the Proof

The proof proceeds in three principal components: A) Determinant vs. $\tau$-degree on Tate Modules: The reduction of $P_{u, \phi}(T, x)$ modulo $\ell$ matches the characteristic polynomial of $u$ on $\phi[\ell]$ (Prop. 2.3). Separable endomorphisms $u$ satisfy $\deg_T \det u = \deg_\tau u$, extended to all $u$ via integrality (Thm. 2.6). B) Uniqueness of the Infinite Place: Lemma 2.1 ensures prescribed valuations at finitely many places, while Lemma 3.2 and Prop. 3.3 establish that the sole place above $\infty$ in $L = \mathbb{F}_q(T, u)$ aligns with the $\tau$-degree. C) Symmetric Polynomial Bounds: A pseudo-absolute value $|x|_* = q^{-v_\infty(N_{F(x):F}(x))/[F(x):F]}$ is multiplicative and, when applied to Frobenius eigenvalues, yields the desired modulus bound via the determinant’s $\tau$-degree (deg$\det \pi = n$). The symmetric coefficients $a_i$ inherit corresponding degree bounds by classical estimates on symmetric polynomials in $r$ variables.

7. Corollaries and Mathematical Consequences

The validation of the Riemann Hypothesis for Drinfeld modules imparts several direct consequences:

• Explicit Formulae: Local Frobenius factors $P_{\phi, \mathfrak{p}}(x)$ with Satake parameters $|\alpha_{i, \mathfrak{p}}| = (N\mathfrak{p})^{1/2}$ yield trace formulas relating sums over test functions to zeros of $\Lambda(\phi, s)$.
• Prime Equidistribution: Equidistribution of Frobenius conjugacy classes in the motivic Galois group is deduced, producing prime-counting error terms of size $O(q^{n/2})$.
• Contextual Integration: These arguments relate to the t-motive theory over function fields and echo the cohomological proofs of the Riemann Hypothesis for varieties over finite fields (Deligne), executed here by means of elementary module and valuation theory.

The full exposition and proof, along with technical refinement and explicit structure, is found in Micheli’s work and is contextualized within the framework laid by Drinfeld, Goss, Laumon, Papikian, and Stichtenoth (Micheli, 13 Dec 2025).