Riemann Hypothesis for Non-Abelian Zeta Functions of Curves over Finite Fields

Abstract

In this paper, we develop some basic techniques towards the Riemann hypothesis for higher rank non-abelian zeta functions of an integral regular projective curve of genus $g$ over a finite field $\mathbb F_q$. As an application of the Riemann hypothesis for these genuine zeta functions, we obtain some explicit bounds on the fundamental non-abelian $\alpha$- and $\beta$-invariants of $X/\mathbb F_q$ in terms of $X$ and $n,\, q$ and $g$: $$\alpha_{X,\mathbb F_q;n}(mn) = \sum_{V}\frac{q^{h^0(X,V)}-1}{\#\mathrm{Aut}(V)} \qquad{\rm and}\qquad \beta_{X,\mathbb F_q;n}(mn ):= \sum_{V}\frac{1}{\#{\mathrm Aut}(V)}\qquad(m\in \mathbb Z)$$ where $V$ runs through all rank $n$ semi-stable $\mathbb F_q$-rational vector bundles on $X$ of degree $mn$. In particular, $$ \prod_{k=1}^{n}\frac{\ \big( \sqrt q^k-1\big)^{2g-1}\ }{(\sqrt q^k+1)}\leq q^{-\binom{n}{2}(g-1)} \beta_{X,\mathbb F_q;n}(0) \leq \prod_{k=1}^{n}\frac{\ \big( 1+\sqrt q^k\big)^{2g-1}\ }{(\sqrt q^k-1)}, $$ Finally, we demonstrate that the related bounds in lower ranks in turn play a central role in establishing the Riemann hypothesis for higher rank zetas.