Mordell--Lang and disparate Selmer ranks of odd twists of some superelliptic curves over global function fields
Extracted source text · arXiv 2504.20594
\documentclass[12pt, reqno]{amsart}
\input{preamble}
\begin{document}
\title[Mordell--Lang. Selmer groups. superelliptic curves. global function field]{Mordell--Lang and disparate Selmer ranks of odd twists of some superelliptic curves over global function fields}
\author{Sun Woo Park}
\address{Max Planck Institute for Mathematics, Vivatsgasse 7, 53115 Bonn, Germany}
\email{} %Please enter email here.
\email{\href{mailto:s.park@mpim-bonn.mpg.de}{s.park@mpim-bonn.mpg.de}}
\begin{abstract}
Fix a prime number $\ell \geq 5$. Let $K = \mathbb{F}_q(t)$ be a global function field of characteristic $p$ coprime to $2,3$, and $q \equiv 1 \text{ mod } \ell$. Let $C:y^\ell = F(x)$ be a non-isotrivial superelliptic curve over $K$ such that $F$ is a degree $3$ polynomial over $\mathbb{F}_q(t)$. Denote by $C_f: fy^\ell = F(x)$ the twist of $C$ by a polynomial $f$ over $\mathbb{F}_q$. Assuming some conditions on $C$, we show that the expected number of $K$-rational points of $C_f$ is bounded, and at least $99\%$ of such curves $C_f$ have at most $(3p)^{5\ell} \cdot \ell!$ many $K$-rational points, as $f$ ranges over the set of polynomials of sufficiently large degree over $\mathbb{F}_q$. To achieve this, we compute the distribution of dimensions of $1-\zeta_\ell$ Selmer groups of Jacobians of such superelliptic curves. This is done by generalizing the technique of constructing a governing Markov operator, as developed from previous studies by Klagsbrun--Mazur--Rubin, Yu, and the author. As a byproduct, we prove that the density of odd twist families of such superelliptic curves with even Selmer ranks cannot be equal to $50\%$, a disparity phenomenon observed in previous works by Klagsbrun--Mazur--Rubin, Yu, and Morgan for quadratic twist families of principally polarized abelian varieties.
\end{abstract}
\maketitle
\input{Section1-Introduction}
\input{Section2-Rank2Galois}
\input{Section3-Application}
\bibliographystyle{alpha}
\bibliography{main}
\end{document}
\usepackage[margin=0.75in]{geometry}
\usepackage[%pdftex,
bookmarks=true]{hyperref}
\usepackage[OT2, T1]{fontenc}
\usepackage[english]{babel}
\usepackage[utf8]{inputenc}
\usepackage{csquotes}
\usepackage[final]{microtype}
\usepackage{lmodern}
\usepackage{amsthm}
\usepackage{amssymb}
\usepackage{mathrsfs}
\usepackage{enumerate}
\usepackage{tikz-cd} % commutative diagrams
\usepackage{tikz}
\usepackage{multicol}
\usepackage{multirow}
\usepackage{tikz-qtree}
\usepackage{rotating}
\usepackage{graphicx}
\usepackage{comment}
\usepackage{enumitem}
\usepackage{manfnt}
\usetikzlibrary{arrows,calc,matrix,trees,arrows.meta,positioning,decorations.pathreplacing,bending}
% Theorem-style environments
\newtheorem{theorem}{Theorem}[section]
\newtheorem{conjecture}[theorem]{Conjecture}
\newtheorem{problem}[theorem]{Problem}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{claim}[theorem]{Claim}
\theoremstyle{definition}
\newtheorem{fact}[theorem]{Fact}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{example}[theorem]{Example}
\newtheorem{note}[theorem]{Note}
\newtheorem{exercise}[theorem]{Exercise}
\newtheorem{question}[theorem]{Question}
\newtheorem{condition}[theorem]{Condition}
%Short-Hand Commands
\newcommand{\Z}{\mathbb{Z}}
\newcommand{\N}{\mathbb{N}} %Integers or Norm Map
\newcommand{\F}{\mathbb{F}}
\newcommand{\A}{\mathbb{A}} %A^1
\newcommand{\G}{\mathbb{G}}
\newcommand{\SL}{\mathrm{SL}}
\newcommand{\GL}{\mathrm{GL}}
\newcommand{\Prob}{\mathbb{P}} %P^1
\newcommand{\Q}{\mathbb{Q}}
\newcommand{\Oh}{\mathcal{O}} %Notation for ring of integers
\newcommand{\Tr}{\mathrm{Tr}} %Trace Map
\newcommand{\Sel}{\mathrm{Sel}} %Selmer Groups
\newcommand{\Gal}{\mathrm{Gal}} %Galois Groups
\newcommand{\Aut}{\mathrm{Aut}} %Automorphism Groups
\newcommand{\Conf}{\mathrm{Conf}}
\newcommand{\Frob}{\mathrm{Frob}}
\newcommand{\Hom}{\mathrm{Hom}}
\newcommand{\et}{\text{\'et}}
\newcommand{\Res}{\mathrm{Res}}
\newcommand{\Rank}{\mathrm{Rank}}
\newcommand{\genlegendre}[4]{%
\genfrac{(}{)}{}{#1}{#3}{#4}%
\if\relax\detokenize{#2}\relax\else_{\!#2}\fi
}
\newcommand{\legendre}[3][]{\genlegendre{}{#1}{#2}{#3}}
\newcommand{\dlegendre}[3][]{\genlegendre{0}{#1}{#2}{#3}}
\newcommand{\tlegendre}[3][]{\genlegendre{1}{#1}{#2}{#3}}
\newcommand*\cube{\mbox{\mancube}}
\DeclareSymbolFont{cyrletters}{OT2}{wncyr}{m}{n}
\DeclareMathSymbol{\Sha}{\mathalpha}{cyrletters}{"58}
\setcounter{tocdepth}{1}
\section{Introduction}
\subsection{Main result}
This paper obtains upper bounds on the moments of the number of rational solutions to twist families of some superelliptic curves over global function fields.
Given a global field $K$ and a smooth projective curve $C$ of genus at least two, the Mordell--Lang conjecture states that the number of $K$-rational solutions of $C$ is finite. The proof of the conjecture with non-effective bounds was obtained in groundbreaking studies by Faltings \cite{Faltings83} and Vojta \cite{Vojta91} when $K$ is a number field, and by Manin \cite{Manin63}, Grauert \cite{Grauert65}, and Vojta \cite{Vojta89} when $K$ is a global function field with finite characteristic. The uniform effective upper bounds for the number of $K$-rational solutions of $C$ can be found in a collection of seminal works, such as those of Buium--Voloch \cite{BV96}, Katz--Rabinoff--Zureick-Brown \cite{KRZB16}, and Dimitrov--Gao--Habegger \cite{DGH21}. We state below one of the effective versions of these results.
\begin{theorem}[Uniform Mordell--Lang for curves]
Let $g \geq 2$ be an integer, and let $K$ be a global field. Given any smooth projective curve $C$ of genus $g$ over $K$ which is not defined over $K^p$, there exists an absolute constant $c := c(g,K) > 0$ such that
\begin{equation*}
\# C(K) \leq c^{1 + \mathrm{Rank}_\mathbb{Z} \mathrm{Jac}(C)(K)},
\end{equation*}
where $\mathrm{Jac}(C)$ is the Jacobian variety of the curve $C$ over $K$.
\end{theorem}
A key parameter which appears in the statement of the uniform Mordell--Lang conjecture is the Mordell-Weil rank of $\mathrm{Jac}(C)(K)$. It is a natural question to ask to which extent the uniform Mordell--Lang conjecture can be utilized to analyze the distribution of number of $K$-rational points of families of smooth projective curves $\mathcal{C}$ of fixed genus at least two.
Before we give a partial answer to this question, we fix a few notations. Fix a prime number $\ell$. Let $K = \mathbb{F}_q(t)$ be the global function field whose characteristic $p$ is coprime to $2, 3,$ and $\ell$. We denote by $\mu_\ell$ the set of primitive $\ell$-th roots of unity. Throughout this manuscript we assume that $\mu_\ell \subset K$. Given a non-negative integer $n$, we denote by $F_n(\mathbb{F}_q)$ the set of polynomials $f$ over $\mathbb{F}_q$ of degree $n$. We denote by $C$ a non-isotrivial superelliptic curve over $K$ whose Weierstrass equation is
\begin{equation}
C: y^\ell = F(x),
\end{equation}
where $F$ is a degree $3$ polynomial with coefficients in $K$ but not in $K^p$. The family of superelliptic curves we consider is given by a family of twists of $C$:
\begin{equation}
\mathcal{C}_n := \left\{ C_f : fy^\ell = F(x) \; | \; f \in F_n(\mathbb{F}_q) \right\}.
\end{equation}
The main result of this paper computes effective probabilistic stratification on the number of $K$-rational points of superelliptic curves $C_f \in \mathcal{C}_n$ as $n$ grows arbitrarily large. Provided below is a simplification of the main result of the manuscript.
\begin{theorem}[A simplification of Theorem \ref{thm:main}] \label{thm:main_simple}
Suppose the following three conditions hold for the superelliptic curve $C: y^\ell = F(x)$ over $K$.
\begin{itemize}
\item $C$ is not defined over $K^p$.
\item The splitting field of $F(x)$ is an $S_3$-Galois extension over $K$, whose constant field is $\mathbb{F}_q$.
\item $\mathrm{Jac}(C)$ has a place $v$ of totally split multiplicative reduction.
\end{itemize}
Then for sufficiently large $n$, there exists a fixed constant $B(\ell,p)$ depending only on $\ell$ and $p$ such that
\begin{equation}
\limsup_{n \to \infty} \frac{\sum_{f \in F_n(\mathbb{F}_q)} \# C_f(K)}{\# F_n(\mathbb{F}_q)} \leq B(\ell,p).
\end{equation}
Furthermore, at least $99\%$ of polynomials $f \in F_n(\mathbb{F}_q)$ satisfy $\# C_f(K) \leq (3p)^{5\ell} \cdot \ell!$, and for every positive integer $k \geq 1$, the following inequality holds:
\begin{equation}
\limsup_{n \to \infty} \frac{\#\{f \in F_n(\mathbb{F}_q) \; : \; \# C_f(K) > (3p)^{(3+k) \cdot \ell} \cdot \ell!\}}{\# F_n(\mathbb{F}_q)} < 2 \cdot {\ell^{-\frac{k(k+1)}{2}}}.
\end{equation}
\end{theorem}
\subsection{Selmer groups of Jacobians}
The main result follows from computing the dimensions of Selmer groups of Jacobians of superelliptic curves in $\mathcal{C}_n$. The computation of distribution of Selmer groups of families of abelian varieties is one of the widely used techniques to understand the distribution of Mordell-Weil ranks of their $K$-rational points. Given an abelian variety $A$ over a global field $K$ and a $K$-rational isogeny $\phi \in \mathrm{End}(A/K)$, the $\phi$-Selmer group of $A$, denoted as $\Sel_\phi(A/K)$, is a finite subset of the first cohomology group $H^1_\et(K, A[\phi])$. In case $\phi$ is a multiplication by $\ell$ map for some prime $\ell$, the group $\Sel_\ell(A/K)$ is a finite dimensional $\mathbb{F}_\ell$ vector space of $H^1_\et(K, A[\ell])$ whose dimension gives an upper bound on the Mordell-Weil rank of $A(K)$. Most notably, the Poonen--Rains heuristics \cite{PR12} predict the following distribution of $\Sel_\ell(A/K)$ for a ``nice'' family of abelian varieties $\mathcal{A}$ endowed with a height function $h: \mathcal{A} \to \mathbb{R}_{\geq 0}$ satisfying the Northcott property (i.e. there are only finitely many isomorphism classes of abelian varieties of bounded height $X$):
\begin{equation}
\lim_{X \to \infty} \frac{\# \{A \in \mathcal{A} \; | \; h(A) \leq X, \dim_{\mathbb{F}_\ell} \Sel_\ell(A/K) = r\}}{\# \{A \in \mathcal{A} \; | \; h(A) \leq X\}} = \prod_{k=0}^\infty \frac{1}{1+\ell^{-k}} \cdot \prod_{k=1}^r \frac{\ell}{\ell^k-1}.
\end{equation}
As the upper bound on the height $X$ grows arbitrarily large, the average size of $\Sel_\ell(A/K)$ as $A$ ranges over elements of $\mathcal{A}$ converges to $\ell + 1$, and the probability that the dimension of $\Sel_\ell(A/K)$ is even (or odd, respectively) as $A$ ranges over elements of $\mathcal{A}$ converges to $\frac{1}{2}$. The formulation of this heuristic rests upon identifying Selmer groups as intersections of two random maximal isotropic subspaces of an infinite dimensional vector space over finite fields \cite[Theorem 4.14]{PR12}.
There is a wealth of previous studies which compute the moments and distributions of Selmer groups of families of Jacobians of higher genus curves over global fields $K$. When $K$ is a number field, the works by Bhargava--Gross\cite{BG13}, Shankar--Wang \cite{SW18}, and Shankar \cite{Sh19} compute moments of 2-Selmer groups of universal families of hyperelliptic curves. The two works by Yu computes that there are infinitely many twist families of hyperelliptic and superelliptic curves over number fields whose Selmer groups have a given dimension \cite{Yu16}, and that the distribution of 2-Selmer groups of quadratic twist families of hyperelliptic curves conforms to the Poonen-Rains heuristics assuming certain conditions on the local conditions defining the desired Selmer groups \cite{Yu19}. These results, combined with Chabauty's method, may be leveraged to prove strong results on the number of $\mathbb{Q}$-rational points of such hyperelliptic curves, as shown in the groundbreaking work by Poonen--Stoll \cite{PS14}. For cases where $K$ is a global function field, the work by Thinh computes the first moments of 2-Selmer groups of universal families of even hyperelliptic curves \cite{Thinh23}. The recent groundbreaking work by Ellenberg--Landesman computes the distribution of $n$-Selmer groups of quadratic twist families of principally polarized abelian varieties \cite{EL23}. Both results compute the upper and lower bounds of the distribution, the two values of which converge to the predicted moments conforming to the Poonen-Rains heuristics assuming that the size of the constant field of $K$ grows arbitrarily large.
Among a variety of techniques available to analyze Selmer groups of families of curves in $\mathcal{C}_n$, we focus on the approach of constructing a Markov operator over the state space of non-negative integers, as previously studied in the works of Klagsbrun--Mazur--Rubin \cite{KMR14}, Yu \cite{Yu16, Yu19}, and the author \cite{park2022prime}. In particular, we focus on computing the Selmer groups of $\mathrm{Jac}(C_f)$ with respect to the isogeny $1-\zeta_\ell: \mathrm{Jac}(C_f) \to \mathrm{Jac}(C_f)$ induced from the map $\zeta_\ell: C_f \to C_f$ that sends $(x,y)$ to $(x, \zeta_\ell y)$.
The two key merits for utilizing this approach are the ability to compute distribution of desired Selmer groups for any size of the constant field of $K$ and the capability to explicitly determine the rate of convergence in terms of $n$.
By generalizing the previous work of the author \cite{park2022prime}, we are able to compute the distribution of $1-\zeta_\ell$ Selmer groups of superelliptic curves $C_f \in \mathcal{C}$ with both merits fulfilled. Interestingly, a non-trivial byproduct we obtain is that the distribution of $1-\zeta_\ell$ Selmer groups of Jacobians of curves $C_f \in \mathcal{C}_n$ deviates from the prediction of Poonen-Rains heuristics. Most notably, the density of curves $C_f \in \mathcal{C}$ whose Jacobian has even dimensional $1-\zeta_\ell$ Selmer group does not converge to $\frac{1}{2}$.
\begin{theorem}[A simplification of Theorem \ref{thm:superelliptic}]
Assume all the conditions and notations from Theorem \ref{thm:main_simple}. Then for sufficiently large $n$, there exist fixed constants $B(C,q) > 0$ and $0 < \alpha(\ell) < 1$ independent of $n$ such that
\begin{equation}
\left| \frac{\#\{f \in F_n(\mathbb{F}_q) \; | \; \dim_{\mathbb{F}_\ell} \Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K) = r\}}{\# F_n(\mathbb{F}_q)} - \rho_C(r) \cdot \prod_{k=0}^\infty \frac{1}{1+\ell^{-k}} \cdot \prod_{k=1}^r \frac{\ell}{\ell^k-1} \right| < \frac{B(C,q)}{n^{\alpha(\ell)}},
\end{equation}
where $\rho_C(r) \in [0,1] \setminus \left\{ \frac{1}{2} \right\}$ is a rational number whose denominator is a power of $\ell$.
\end{theorem}
We refer to Theorem \ref{thm:superelliptic} and Proposition \ref{prop:major_contribution} for an explicit description of the parameter $\rho_C$. More generally, we identify Chebotarev conditions which give rise to such disparity in dimensions of Selmer groups of $2$-dimensional $\Gal(\overline{K}/K)$ modules $T$ satisfying a number of axioms, see in particular Condition \ref{condition:admissible} and Theorem \ref{thm:Galois_module}. These results are generalizations of remarkable observations made by Klagsbrun--Mazur--Rubin \cite[Corollary 7.10, Theorem 8.2]{KMR13}, Yu \cite[Theorem 1]{Yu16}, and Morgan \cite[Theorem 7.4]{Mor19} where they demonstrate that there are disparities in the dimensions of 2-Selmer groups of quadratic twist families of principally polarized abelian varieties. Likewise, the function $\rho_C(r)$ can be considered as a parameter that signifies slight deviations from one of the assumptions inherent in the Poonen-Rains heuristics, see Remark \ref{remark:PR} for more details.
\subsection{Organization}
Section \ref{sec:Selmer} generalizes the main theorem of \cite{park2022prime} to admissible $\Gal(\overline{K}/K)$ modules $T$, a $2$-dimensional vector space over the finite field $\mathbb{F}_\ell$ satisfying a number of mild technical conditions outlined in Condition \ref{condition:admissible}. The highlight of the section is Theorem \ref{thm:Galois_module}, where we extend the main result of \cite{KMR14} to any such admissible $\Gal(\overline{K}/K)$ modules $T$ over global function fields $K$. Section \ref{sec:superelliptic} focuses on applying Theorem \ref{thm:Galois_module} to obtain Theorem \ref{thm:main} and Theorem \ref{thm:superelliptic}.
\subsection*{Acknowledgements}
I would like to thank Valentin Blomer, Jordan Ellenberg, Daniel Keliher, Chris Keyes, Zev Klagsbrun, and David Zureick-Brown for helpful discussions and comments. I would like to also thank the Max Planck Institute for Mathematics for its generous support.
\section{Distribution of twisted Selmer structures of rank 2 Galois modules}
\label{sec:Selmer}
Throughout this manuscript we fix a set $T$ equipped with a continuous action of $\Gal(\overline{K}/K)$.
\begin{condition} \label{condition:admissible}
We say that a $\Gal(\overline{K}/K)$-module $T$ is \textbf{admissible} if $T$ satisfies all the seven conditions provided below.
\begin{enumerate}
\item $T$ is a 2-dimensional $\mathbb{F}_\ell$ vector space.
\item There exists a $\Gal(\overline{K}/K)$-equivariant non-degenerate pairing $e_T: T \times T \to \mu_\ell$ such that the following composition is a symmetric pairing:
\begin{equation*}
H^1_{\et}(K_v, T) \times H^1_{\et}(K_v, T) \to H^2_{\et}(K_v, T \otimes T) \to \mathbb{F}_\ell,
\end{equation*}
where the first map is the cup product, and the second map is the map induced from $e_T$.
\item The constant field of $K(T)$ is equal to $\mathbb{F}_q$.
\item $T$ is a simple $\Gal(\overline{K}/K)$-module.
\item $\mathrm{Hom}_{\Gal(\overline{K}/K)}(T,T) = \mathbb{F}_\ell$.
\item $H^1(\Gal(K(T)/K),T) = 0$.
\item There exists a place $v$ of $K$ such that $T^{\Gal(\overline{K}_v/K_v\overline{\mathbb{F}}_q)} \cap \mu_{\ell^\infty} = \mu_\ell$.
\end{enumerate}
\end{condition}
The goal of this section is to generalize the results by \cite{KMR14} and \cite{park2022prime} on the distribution of Selmer groups of cyclic twists of a fixed admissible $\Gal(\overline{K}/K)$-module $T$. To state the proposed generalization, we introduce the following notations and results from \cite[Section 5]{KMR14}.
\begin{definition}[Definition 5.8, Lemma 5.9 of \cite{KMR14}, Definition 4.2 of \cite{park2022prime}]
We define the following subsets of places $v$ of $K$.
\begin{itemize}
\item $\Sigma$: a set of places which includes places at which $T$ is ramified and which are archimedean.
\item $\Sigma_T$: the set of places whose elements are exactly places at which $T$ is ramified and which are archimedean.
\item $\Sigma(\mathfrak{d})$: a set of places which includes places in $\Sigma$ and places dividing a square-free product of places $\mathfrak{d}$
\item $\mathcal{P}_0$: the set of all places $v \not\in \Sigma$ and $\dim_{\mathbb{F}_\ell} T^{\Gal(\overline{K}_v/K_v)} = 0$. It is the set of all places $v$ such that $\Frob_v \in \Gal(K(T)/K)$ has non-trivial order prime to $\ell$.
\item $\mathcal{P}_1$: the set of all places $v \not\in \Sigma$ and $\dim_{\mathbb{F}_\ell} T^{\Gal(\overline{K}_v/K_v)} = 1$. It is the set of all places $v$ such that $\Frob_v \in \Gal(K(T)/K)$ has order equal to $\ell$.
\item $\mathcal{P}_2$: the set of all places $v \not\in \Sigma$ and $\dim_{\mathbb{F}_\ell} T^{\Gal(\overline{K}_v/K_v)} = 2$. It is the set of all places $v$ such that $\Frob_v \in \Gal(K(T)/K)$ has order equal to $1$.
\end{itemize}
For each $i$, we denote by $\delta(\mathcal{P}_i)$ the density of places $v$ of $T$ such that $v \in \mathcal{P}_i$.
\end{definition}
\begin{definition}
Given a place $v$ of $K$, we denote by $\mu_{T,v}$ the symmetric pairing obtained from part (2) of Condition \ref{condition:admissible}:
\begin{align*}
\mu_{T,v} &: H_\et^1(K_v,T) \times H_\et^1(K_v,T) \to \mathbb{F}_\ell.
\end{align*}
\end{definition}
\begin{definition}[Definition 5.4 of \cite{KMR14}]
For each place $v$ of $K$, we denote by $\mathcal{H}(\mu_{T,v})$ the set of maximal isotropic subspaces of $H^1_{\et}(K_v,T)$ with respect to the pairing $\mu_{T,v}$. Note that $H^1_\et(\Oh_{K_v},T) \in \mathcal{H}(\mu_{T,v})$. We denote by $\mathcal{H}_{ram}(\mu_{T,v})$ the set of subspaces in $\mathcal{H} \in \mathcal{H}(\mu_{T,v})$ such that $\mathcal{H} \cap H^1_\et(\Oh_{K_v},T) = 0$.
\end{definition}
\begin{lemma}[Lemma 5.5 of \cite{KMR14}]
The following statements hold for any place $v$ of $K$.
\begin{itemize}
\item For every $\mathcal{H} \in \mathcal{H}(\mu_{T,v})$ we have $\dim_{\mathbb{F}_\ell} \mathcal{H} = \dim_{\mathbb{F}_\ell} T^{\Gal(\overline{K}_v/K_v)}$.
\item If $v \in \mathcal{P}_1 \cup \mathcal{P}_2$, then we have $\# \mathcal{H}_{ram}(\mu_{T,v}) = \ell^{\dim T^{\Gal(\overline{K}_v/K_v)} - 1}$.
\end{itemize}
\end{lemma}
The lemma above gives a leverage to parametrize subspaces in $\mathcal{H}(\mu_{T,v})$ and $\mathcal{H}_{ram}(\mu_{T,v})$ using order $\ell$ Galois extensions over $K_v$.
\begin{definition}[Definition 5.10 of \cite{KMR14}]
A twisting data $\alpha := (\alpha_v)_v$ is a collection of maps $\alpha_v$ for each place $v$ of $K$ defined as follows.
\begin{itemize}
\item If $v \in \Sigma$, then $\alpha_v$ is a set-theoretic map
\begin{equation*}
\alpha_v: \frac{\mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell)}{\mathrm{Aut}(\mu_\ell)} \to \mathcal{H}(\mu_{T,v}).
\end{equation*}
\item If $v \in \mathcal{P}_0$, then $\alpha_v$ is the zero map
\begin{equation*}
\alpha_v: \frac{\mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell)}{\mathrm{Aut}(\mu_\ell)} \to \{0\}.
\end{equation*}
\item If $v \in \mathcal{P}_1$, then $\alpha_v$ is the map
\begin{equation*}
\alpha_v: \frac{\mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell)}{\mathrm{Aut}(\mu_\ell)} \to \{H^1_{\et}(\Oh_{K_v}, T)\} \cup \mathcal{H}_{ram}(\mu_{T,v}),
\end{equation*}
which maps the unramified $\mathbb{Z}/\ell \mathbb{Z}$ extension to $H^1_{\et}(\Oh_{K_v},T)$, and maps all other ramified extensions to the unique element of $\mathcal{H}_{ram}(\mu_{T,v})$.
\item If $v \in \mathcal{P}_2$, then $\alpha_v$ is a bijection
\begin{equation*}
\alpha_v: \frac{\mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell)}{\mathrm{Aut}(\mu_\ell)} \to \{H^1_{\et}(\Oh_{K_v}, T)\} \cup \mathcal{H}_{ram}(\mu_{T,v}).
\end{equation*}
\end{itemize}
\end{definition}
\begin{definition}[c.f. Definition 5.6, Definition 5.12 of \cite{KMR14}]
Given an equivalence class of characters $\overline{\chi} \in \mathrm{Hom}(\Gal(\overline{K}/K), \mu_\ell) / \mathrm{Aut}(\mu_\ell)$ and a choice of a twisting data $\alpha := (\alpha_v)_v$, we denote by $\Sel(T,\alpha : \overline{\chi})$ a finite dimensional $\mathbb{F}_\ell$ vector space defined as
\begin{equation}
\Sel(T,\alpha : \overline{\chi}) := \mathrm{Ker} \left( H_\et^1(K,T) \to \prod_v \frac{H_\et^1(K_v,T)}{\alpha_v(\overline{\chi}_v)} \right),
\end{equation}
where $\overline{\chi}_v$ is the restriction of $\overline{\chi}$ to an element of $\mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell) / \mathrm{Aut}(\mu_\ell)$.
\end{definition}
\begin{definition}
Given a polynomial $f \in \mathbb{F}_q[t]$, we denote by $\overline{\chi}^f \in \mathrm{Hom}(\Gal(\overline{K}/K), \mu_\ell) / \mathrm{Aut}(\mu_\ell)$ an equivalence class of cyclic order $\ell$ characters whose fixed field is $K(\sqrt[\ell]{f})$.
\end{definition}
We state the main result to be proven in this section.
\begin{theorem}
\label{thm:Galois_module}
Fix a prime number $\ell$. Let $K = \mathbb{F}_q(t)$ be a global function field whose characteristic is coprime to $2$, $3$ and $q \equiv 1 \text{ mod } \ell$. Let $T$ be an admissible $\Gal(\overline{K}/K)$-module. Fix a twisting data $\alpha := (\alpha_v)_v$. Then for sufficiently large $n$, there exists a fixed constant $B(T,\alpha) > 0$ depending only on $T$ and $\alpha$ and independent of $n$ such that
\begin{align}
\left| \frac{\# \{f \in F_n(\mathbb{F}_q) \; | \; \dim_{\mathbb{F}_\ell} \Sel(T, \alpha : \overline{\chi}^f) = r \}}{\#F_n(\mathbb{F}_q)} - \rho(r) \cdot \prod_{k=1}^\infty \frac{1}{1+\ell^{-k}} \prod_{k=1}^r \frac{\ell}{\ell^k-1} \right| < \frac{B(T,\alpha)}{n^{\alpha(T)}},
\end{align}
where
\begin{equation}
\alpha(T) := \sup_{0 < \rho < 1} \left( \min \left( \rho \log \rho + 1 - \rho, \; \; \rho \log \gamma_T, -\rho \log \left( 1 - \delta(\mathcal{P}_0) \right) \right) \right)
\end{equation}
with $\gamma_T$ an explicit constant depending only on $T$ (see Theorem \ref{thm:Markov} for its construction), and
\begin{equation}
\rho(r) := \lim_{n \to \infty} \left(\frac{1}{2} + (-1)^r \cdot \left(\frac{1}{2} - \sum_{\substack{k \geq 0 \\ k \equiv r \text{ mod } 2}} \frac{\#\{f \in F_n(\mathbb{F}_q) \cap \mathcal{P}_0 \; | \; \dim_{\mathbb{F}_\ell} \Sel(T, \alpha : \overline{\chi}^f) = r\}}{\#F_n(\mathbb{F}_q) \cap \mathcal{P}_0} \right) \right).
\end{equation}
We note that $\rho(r) = \frac{1}{2}$ for all non-negative integer $r \geq 0$ if and only if $\delta(\mathcal{P}_1) \neq 0$.
\end{theorem}
The proof of the theorem follows from a direct line-by-line adaptation of the proof of the main result of \cite{KMR14} and \cite{park2022prime}. We borrow all the ingredients and relevant results from \cite{KMR14} and \cite{park2022prime}, filling in the gaps of the proof if needed. For the interest of potential readers who are aware of the two references, we will adhere to the notations utilized in both references, and introduce new notations when necessary. For the rest of the section, we assume that $T$ is an admissible $\Gal(\overline{K}/K)$-module.
\subsection{Effective Chebotarev density theorem}
We recall the two types of effective Chebotarev density theorems from Section 3.1 of \cite{park2022prime}.
\begin{theorem}[Proposition 6.4.8 of \cite{FJ08}, Theorem 3.1 of \cite{park2022prime}] \label{theorem:effective_chebotarev}
Let $L/K$ be a Galois extension of global function fields over $\F_q(t)$. Pick a conjugacy class $C \subset G = \Gal(L/K)$. Denote by $g_L$ and $g_K$ genera of $L$ and $K$. If the constant fields of $L$ and $K$ are both equal to $\F_q$, then
\begin{align*}
& \left| \# \{ v \; \text{a place over} \; K \; | \; \Frob_v \in C, \; \dim_{\F_q} (\Oh_K/v) = n \} - \frac{|C|}{|G|} \frac{q^n}{n} \right| \\
& < \frac{2|C|}{n|G|} \left[ (|G| + g_L) q^{\frac{n}{2}} + |G|(2g_K + 1)q^{\frac{n}{4}} + (|G| + g_L) \right].
\end{align*}
\end{theorem}
\begin{corollary}[Corollary 3.2 of \cite{park2022prime}]
\label{corollary:effective_chebotarev}
Assume all conditions and notations from Theorem \ref{theorem:effective_chebotarev}. Pick two non-empty subsets $S, S' \subset G = \Gal(L/K)$ stable under conjugation. Suppose that the size of the constant field $q$ satisfies
\begin{equation*}
q^{\frac{n}{2}} - q^{\frac{n}{4}} > 2(|G| + g_L + 2g_K).
\end{equation*}
Then the following inequality holds.
\begin{align*}
& \left| \frac{\{ v, \; \text{a place over} \; K \; | \; \Frob_v \in S, \; \dim_{\F_q} (\Oh_K/v) = n \}}{\{ v, \; \text{a place over} \; K \; | \; \Frob_v \in S', \; \dim_{\F_q} (\Oh_K/v) = n \}} - \frac{|S|}{|S'|} \right| \\
& < 4\frac{|S|}{|S'|} (|G| + g_L + 2g_K) \left[ \frac{1}{q^\frac{n}{2} - q^\frac{n}{4} - 2(|G| + g_L + 2g_K) } \right].
\end{align*}
In particular, if $n \geq 2 \cdot (\log 8 + \log(|G| + g_L + 2g_K)) \cdot (\log q)^{-1}$, then
\begin{equation*}
\left| \frac{\{ v, \; \text{a place over} \; K \; | \; \Frob_v \in S, \; \dim_{\F_q} (K/v) = n \}}{\{ v, \; \text{a place over} \; K \; | \; \Frob_v \in S', \; \dim_{\F_q} (K/v) = n \}} - \frac{|S|}{|S'|} \right| < 16 \frac{|S|}{|S'|} (|G| + g_L + 2g_K) q^{-\frac{n}{2}}.
\end{equation*}
\end{corollary}
\subsection{Irreducible factors}
Throughout this section, we define two constants $m_{n,q}$ and $\mathfrak{n}$ as
\begin{align}
\begin{split}
m_{n,q} &:= \log n + \log \log q, \\
\mathfrak{n} &:= \frac{4m_{n,q}^2}{\log q}.
\end{split}
\end{align}
We denote by $g_T$ the genus of the geometric Galois extension $K(T)/K$.
\begin{definition}[Definition 4.11 of \cite{park2022prime}]
We denote by $\hat{F}_{(n,N),(w,w')}(\F_q)$ the set of polynomials $f \in \mathbb{F}_q[t]$ of degree $n$ which satisfies the following conditions:
\begin{itemize}
\item The number of distinct irreducible factors of $f$ is equal to $w$.
\item The number of distinct irreducible factors of $f$ of degrees greater than $\mathfrak{n}$ is equal to $w'$.
\item The degree of products of all irreducible factors of $f$ (including multiplicities) is equal to $N$.
\item There exists at least one irreducible factor of $f$ of degree greater than $\mathfrak{n}$ that lies in $\mathcal{P}_0$.
\end{itemize}
\end{definition}
With appropriate choices of $w, w'$, and $N$, we can approximate the set $F_n(\mathbb{F}_q)$ with disjoint unions of $\hat{F}_{(n,N),(w,w')}(\F_q)$, with explicit error term computed as below.
\begin{proposition}[Proposition 4.14 of \cite{park2022prime}] \label{proposition:fan_approximation}
Let $\rho \in (0,1)$ be a positive number. Suppose $n$ is a positive integer such that $m_{n,q} > \max \{e^{e^e}, \log6 + \log(\ell^3 + g_{T})\}$. Let $\epsilon = \frac{1}{\log \log m _{n,q}}$. Then
\begin{align}
\begin{split}
& \; \; \; \; \# F_n(\F_q) - \sum_{w = \rho m_{n,q}}^{2 m_{n,q}} \sum_{w' = (1-\epsilon)w}^w \sum_{N = w'\mathfrak{n}}^n \# \hat{F}_{(n,N),(w,w')}(\F_q) \\
&\leq 4 \cdot q^n \cdot \max \left(n^{-\rho \log \rho - 1 + \rho}, 3 m_{n,q}^2 \cdot \left( 1 - \delta(\mathcal{P}_0) \right)^{(1-\epsilon)\rho m_{n,q}} \right).
\end{split}
\end{align}
\end{proposition}
The first component of the error term originates from large and moderate deviation principle for distinct irreducible factors of polynomials over $\mathbb{F}_q$ \cite{FWY20}, whereas the second component of the error term originates from computing the probability that all $w'$ irreducible factors of $f \in \hat{F}_{(n,N),(w,w')}(\F_q)$ do not lie in $\mathcal{P}_0$.
\subsection{Markov chains}
A key insight from \cite{SD08}, \cite{KMR14}, and \cite{park2022prime} is that there is a Markov operator over the countable state space of non-negative integers whose stationary distribution determines the distribution of Selmer groups of twist families of elliptic curves over global fields. We recall the relevant facts about the Markov operator of our interest.
\begin{definition}[Definition 6.1 of \cite{park2022prime}] \label{defn:Markov}
Let $M_\ell := [\ell_{r,s}]$ be the operator over the state space of non-negative integers $\mathbb{Z}_{\geq 0}$ defined as
\begin{equation*}
\ell_{r,s} := \begin{cases}
1 - \ell^{-r} &\text{ if } s = r-1 \geq 0 \\
\ell^{-r} &\text{ if } s = r+1 \\
0 &\text{ else}.
\end{cases}
\end{equation*}
\end{definition}
\begin{definition}
Given an admissible $\Gal(\overline{K}/K)$-module $T$, we denote by $M_T$ the Markov operator over the state space of non-negative integers $\mathbb{Z}_{\geq 0}$ defined as
\begin{equation}
M_T := \delta(\mathcal{P}_0) \cdot I + \delta(\mathcal{P}_1) \cdot M_\ell + \delta(\mathcal{P}_2) \cdot M_\ell^2,
\end{equation}
where $I$ is the identity operator.
\end{definition}
We note that in the case where $\ell = 2$, a necessary condition for a module $T$ to be admissible is to have $\Gal(K(T)/K) \cong S_3$. In such a case, the Markov operator $M_T$ is given by
\begin{equation}
M_T := \frac{1}{3} \cdot I + \frac{1}{2} \cdot M_2 + \frac{1}{6} \cdot M_2^2.
\end{equation}
We note that $M_T$ is an irreducible, aperiodic, and positive-recurrent Markov operator if and only if $\delta(\mathcal{P}_1) \neq 0$. Note that if $\delta(\mathcal{P}_1) = 0$, then $M_T$ has two non-communicating states: the set of odd integers and the set of even integers. Nevertheless, in each non-communicating states, the operator $M_T$ is irreducible, aperiodic, and positive-recurrent. To determine stationary distributions of $M_T$ based on the value of $\delta(\mathcal{P}_1)$, we recall the notion of the parity of a probability distribution.
\begin{definition}[Definition 6.3 of \cite{park2022prime}]
Let $\mu:\mathbb{Z}_{\geq 0} \to [0,1]$ be a probability distribution. The parity of $\mu$, denoted as $\rho(\mu)$, is the sum of probabilities supported over odd integer states, i.e. $\rho(\mu) := \sum_{n \equiv 1 \text{ mod } 2} \mu(n)$.
\end{definition}
The following theorem elucidates the geometric rate of convergence of the Markov operator $M_T$ to a stationary distribution given the parity of the initial probability distribution $\mu$.
\begin{theorem} \label{thm:Markov}
Suppose $T$ is an admissible $\Gal(\overline{K}/K)$-module. Given a probability distribution $\mu: \mathbb{Z}_{\geq 0} \to [0,1]$, we denote by $\mathbb{E}[\ell^\mu] := \sum_{z \in \mathbb{Z}_{\geq 0}} \ell^z \cdot \mu(z)$.
\begin{enumerate}
\item Suppose $\delta(\mathcal{P}_1) \neq 0$. Then for every $n \in \mathbb{N}$ and any probability distribution $\mu$, there exist fixed constants $c_T > 0$ and $0 < \gamma_T < 1$ depending only on $T$ such that
\begin{equation}
\sup_{r \in \mathbb{Z}_{\geq 0}} \left| M_T^n \mu (r) - \prod_{j=0}^\infty \frac{1}{1+\ell^{-j}} \prod_{j=1}^r \frac{\ell}{\ell^j - 1} \right| < c_T \gamma_T^n (\mathbb{E}[\ell^\mu] + 1).
\end{equation}
\item Suppose $\delta(\mathcal{P}_1) = 0$. Then for every $n \in \mathbb{N}$ and any probability distribution $\mu$, there exist fixed constants $c_T > 0$ and $0 < \gamma_T < 1$ depending only on $T$ such that
\begin{equation}
\sup_{r \in \mathbb{Z}_{\geq 0}} \left| M_T^n \mu (z) - \left( \frac{1}{2} + (-1)^r \cdot \left(\frac{1}{2} - \rho(\mu) \right) \right)\prod_{j=1}^\infty \frac{1}{1+\ell^{-j}} \prod_{j=1}^r \frac{\ell}{\ell^j - 1} \right| < c_T \gamma_T^n (\mathbb{E}[\ell^\mu] + 1).
\end{equation}
\end{enumerate}
In particular, the constant $\gamma_T$ is the geometric rate of convergence of the Markov operator $M_T$ to its stationary distribution.
\end{theorem}
\begin{proof}
The theorem follows from \cite[Proposition 2.4]{KMR14}, \cite[Theorem 15.0.1]{MT93}, and \cite[Corollary 6.7, Lemma 6.8]{park2022prime}. For the case where $\delta(\mathcal{P}_1) = 0$, we take $\gamma_T$ to be the maximum of geometric rates of convergence over each non-communicating state of $M_T$.
\end{proof}
\subsection{Local Selmer groups}
To relate the distribution of Selmer groups $\Sel(T, \alpha : \overline{\chi})$ with stationary behavior of the Markov operator $M_T$, we utilize the notion of local Selmer structures associated to the module $T$.
\begin{definition}[Section 5 of \cite{KMR14}, Definition 5.1 of \cite{park2022prime}]
Given a set $\Sigma$ of places of $K$, we denote by $\mathfrak{d}$ a square-free product of places coprime to elements in $\Sigma$.
\begin{itemize}
\item $\Omega_T$: the set of finite Cartesian products of local characters
\begin{equation*}
(\chi_v)_v \in \Omega_T := \prod_{v \in \Sigma_T} \mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell)
\end{equation*}
\item $\Omega_\mathfrak{d}$: the set of finite Cartesian products of local characters
\begin{equation*}
(\chi_v)_v \in \Omega_\mathfrak{d} := \prod_{v \in \Sigma(\mathfrak{d})} \mathrm{Hom}(\Gal(\overline{K}_v/K_v), \mu_\ell)
\end{equation*}
such that the local character at places $v \mid \mathfrak{d}$ is ramified.
\item $\Omega_{\chi, \mathfrak{v}}$: the set of finite Cartesian products of local characters $\chi' \in \Omega_{\mathfrak{d}\mathfrak{v}}$ given an element $\chi \in \Omega_\mathfrak{d}$ that satisfies the following two conditions:
\begin{itemize}
\item For any $v \in \Sigma(\mathfrak{d})$, we have $\chi'_v = \chi_v$.
\item The component $\chi'_\mathfrak{v}$ is ramified at $\mathfrak{v}$.
\end{itemize}
\end{itemize}
\end{definition}
\begin{definition}[Definition 5.2 of \cite{park2022prime}]
Given an element $(\chi_v)_v \in \Omega_\mathfrak{d}$ and a choice of a twisting data $\alpha$, the local Selmer group of $T$ associated to $\alpha$ and $(\chi_v)_v$ is defined as
\begin{equation}
\Sel(T, \alpha : (\chi_v)_v) := \mathrm{Ker} \left( H_\et^1(K, T) \to \prod_v \frac{H^1_\et(K_v, T)}{\alpha_v(\chi_v)}\right).
\end{equation}
The dimension of $\Sel(T, \alpha : (\chi_v)_v)$, as an $\mathbb{F}_\ell$-vector space, is denoted as $\mathrm{rk}((\chi_v)_v)$. Lastly, given an element $\chi := (\chi_v)_v \in \Omega_\mathfrak{d}$ and a place $\mathfrak{v} \not\in \Sigma(\mathfrak{d})$, we denote by $t_{\chi}(\mathfrak{v})$ the dimension of the following subspace:
\begin{equation}
t_\chi(\mathfrak{v}) := \dim_{\mathbb{F}_\ell} \mathrm{Image} \left( \mathrm{loc}_\mathfrak{v}: \Sel(T, \alpha : (\chi_v)_v) \to H^1_\et(\Oh_{K_v}, T) \right).
\end{equation}
\end{definition}
The variation of the dimensions of local Selmer groups can be detected from the quantity $t_\chi(\mathfrak{v})$, as shown below.
\begin{proposition}[Proposition 7.2 of \cite{KMR14}, Proposition 5.3 of \cite{park2022prime}]
Let $T$ be an admissible $\Gal(\overline{K}/K)$-module. Fix a set of places $\Sigma$, a square-free product of places $\mathfrak{d}$ coprime to any places in $\Sigma$, an element $\chi \in \Omega_\mathfrak{d}$, and a place $\mathfrak{v} \not\in \Sigma(\mathfrak{d})$. Then for any $\chi' \in \Omega_{\chi, \mathfrak{v}}$, we have
\begin{equation*}
\mathrm{rk}(\chi') - \mathrm{rk}(\chi) = \begin{cases}
2 &\text{ if } \mathfrak{v} \in \mathcal{P}_2 \text{ and } t_\chi(\mathfrak{v}) = 0 \text{ for exactly } \ell - 1 \text{ out of } \ell(\ell - 1) \text{ many } \chi' \in \Omega_{\chi, \mathfrak{v}}, \\
1 &\text{ if } \mathfrak{v} \in \mathcal{P}_1 \text{ and } t_\chi(\mathfrak{v}) = 0, \\
-1 &\text{ if } \mathfrak{v} \in \mathcal{P}_1 \text{ and } t_\chi(\mathfrak{v}) = 1, \\
-2 &\text{ if } \mathfrak{v} \in \mathcal{P}_2 \text{ and } t_\chi(\mathfrak{v}) = 2, \\
2 &\text{ otherwise }.
\end{cases}
\end{equation*}
\end{proposition}
Using the proposition above, we obtain the following Chebotarev condition that determines the variation of the dimensions of local Selmer groups.
\begin{proposition}[Proposition 5.4 of \cite{park2022prime}] \label{prop:local-Selmer}
Let $T$ be an admissible $\Gal(\overline{K}/K)$-module. Fix a set of places $\Sigma$, a square-free product of places $\mathfrak{d}$ coprime to any places in $\Sigma$, and an element $\chi \in \Omega_\mathfrak{d}$.
Let $d_{i,j}$ be given by the following table:
\begin{center}
\begin{tabular}{|c||c|c|c|}
\hline
$d_{i,j}$ & $i = 0$ & $i = 1$ & $i = 2$ \\
\hline
\hline
$j = -2$ & $\times$ & $\times$ & $1 - (\ell+1) \ell^{-\text{rk}(\chi)} + \ell^{1-2\text{rk}(\chi)}$ \\
\hline
$j = -1$ & $\times$ & $1-\ell^{-\text{rk}(\chi)}$ & $\times$ \\
\hline
$j = 0$ & $1$ & $\times$ & $(\ell+1)(\ell^{-\text{rk}(\chi)} -\ell^{-2\text{rk}(\chi)})$ \\
\hline
$j = 1$ & $\times$ & $ \ell^{-\text{rk}(\chi)}$ & $\times$ \\
\hline
$j = 2$ & $\times$ & $\times$ & $\ell^{-2\text{rk}(\chi)}$ \\
\hline
\end{tabular}
\end{center}
Here, the term "$\times$" denotes the case where such a difference of ranks cannot occur. Let $D_{T,\ell,q} > 0$ be a constant defined as
\begin{equation}
D_{T,\ell,q} := \ell^{\text{max}_{\chi \in \Omega_T} \left( \text{rk}(\chi) \right)}
\end{equation}
Then there exists a fixed constant $C_{T,\ell,q} > 0$ which depends only on $T$, $\ell$, and $q$ such that for every $$d > \frac{12 \log \ell + 2\log D_{T,\ell,q} + (6 \log \ell) \cdot \# \Sigma_T(\mathfrak{d})}{\log q},$$ we have
\begin{equation}
\left| \frac{\# \{ \mathfrak{v} \in \mathcal{P}_i \; | \; \deg \mathfrak{v} = d, \mathfrak{v} \not\in \Sigma_T(\sigma) \text{ and } t_\chi(\mathfrak{v}) = j \}}{\# \{\mathfrak{v} \in \mathcal{P}_i \; | \; \deg \mathfrak{v} = d, \mathfrak{v} \not\in \Sigma_T(\sigma)\}} - d_{i,j} \right| < C_{T,\ell,q} \cdot \ell^{3 \# \Sigma_T(\sigma)} \cdot q^{-\frac{d}{2}}.
\end{equation}
\end{proposition}
\begin{proof}
The statement follows from adapting the proof presented in \cite[Proposition 5.4]{park2022prime}. We present a short summary of the proof. We denote by $\mathrm{Res}_T$ the following injection obtained from the inflation-restriction sequence:
\begin{equation*}
\mathrm{Res}_T: H^1_\et(K, T) \to \mathrm{Hom}(\Gal(\overline{K}/K(T)), T)^{\Gal(K(T)/K)}.
\end{equation*}
Given an element $\chi := (\chi_v)_v \in \Omega_\mathfrak{d}$, let $F_{\mathfrak{d},\chi}$ be the fixed field of the following subgroup of $\Gal(\overline{K}/K(T))$:
\begin{equation*}
\bigcap_{c \in \Sel(T, \alpha : (\chi_v)_v)} \mathrm{Ker} \left( \mathrm{Res}_T(c): \Gal(\overline{K}/K(T)) \to T \right).
\end{equation*}
As shown in \cite[Proposition 9.3]{KMR14}, the field $F_{\mathfrak{d}, \chi}$ is a Galois extension over $K$ satisfying the following properties.
\begin{itemize}
\item $\Gal(F_{\mathfrak{d},\chi}/K(T)) \cong T^{\mathrm{rk}(\chi)}$,
\item $F_{\mathfrak{d},\chi}/K$ is unramified outside of places of $\Sigma_T(\mathfrak{d})$.
\end{itemize}
The fact that the constant field of $F_{\mathfrak{d}, \chi}$ is identical to that of $K$ follows from statement (7) of Condition \ref{condition:admissible}, from which one can deduce that any basis element $c \in \Sel(T, \alpha : (\chi_v)_v)$ maps the arithmetic Frobenius $\Frob_q \in \Gal(\overline{\mathbb{F}}_q/\mathbb{F}_q)$ to the identity element of $T$. This allows us to apply the effective Chebotarev density theorem over conjugacy-invariant subsets of $\Gal(F_{\mathfrak{d},\chi}/K)$, which relates the densities $d_{i,j}$ with according two conditions on $\mathfrak{v} \in \mathcal{P}_i$ and $t_\chi(\mathfrak{v})=j$, see \cite[Proposition 9.4]{KMR14} for more details.
\end{proof}
\subsection{Proof of Theorem \ref{thm:Galois_module}}
Proposition \ref{prop:local-Selmer} computes the variations of dimensions of local Selmer groups assuming that one can choose finite Cartesian products of local characters uniformly at random from $\Omega_\mathfrak{d}$. To utilize this proposition in computing the variation of dimensions of global Selmer groups, we require that the restriction of global characters $\chi \in \mathrm{Hom}(\Gal(\overline{K}/K), \mu_\ell)$ to finite Cartesian products of local characters $\Omega_\mathfrak{d}$ for some $\mathfrak{d}$ converges in distribution to a uniform distribution. This claim holds true for global characters induced from the set $\hat{F}_{(n,N),(w,w')}(\mathbb{F}_q)$, because every polynomial in such a set has an irreducible factor in $\mathcal{P}_0$ of degree greater than $\mathfrak{n}$. For example, the following effective statement holds true.
\begin{lemma}[Corollary 4.17 of \cite{park2022prime}] \label{lemma:equidistribution}
Let $T$ be an admissible $\Gal(\overline{K}/K)$-module. Let $h_1, h_2, \cdots, h_\omega$ be irreducible polynomials over $\mathbb{F}_q$. Given a place $v$ of $K$, we denote by $\left(\frac{v}{h_j}\right)_\ell \in \mu_\ell$ the $\ell$-th power residue symbol. Let $n$ be an integer such that the sums of degrees of $h_i$'s are at most $n$ and $\omega \leq 2m_{n,q}$.
\begin{itemize}
\item Suppose $\ell \geq 5$ or $K(\sqrt[\ell]{h_1}, \cdots, \sqrt[\ell]{h_\omega}) \cap K(T) = K$. Then for any element $a \in \mu_\ell^{\oplus \omega}$ and $i > \mathfrak{n}$, tehre exists a constant $\hat{C}_{T,\ell,q} > 0$ depending only on $T, \ell$, and $q$ such that
\begin{equation*}
\left| \frac{\#\{ v \in \mathcal{P}_k \; | \; \deg v = i, \oplus_{j=1}^\omega \left( \frac{v}{h_j} \right)_\ell = a \in \mu_\ell^{\oplus \omega}\}}{\#\{v \in \mathcal{P}_k \; | \; \deg v = i\}} - \frac{1}{\ell^\omega}\right| < \hat{C}_{T,\ell,q} \cdot (n \log q)^{-2 m_{n,q} + 2 \log \ell}.
\end{equation*}
\item Suppose $\ell = 2, 3$ and $K(\sqrt[\ell]{h_1}, \cdots, \sqrt[\ell]{h_\omega}) \cap K(T) \neq K$. Then for any $i > \mathfrak{n}$, there are $\ell^\omega - \ell^{\omega-1}$ many elements $a \in \mu_\ell^{\oplus \omega}$ such that $\oplus_{j=1}^\omega \left( \frac{v}{h_j} \right)_\ell \neq a$ for all $v \in \mathcal{P}_k$ with $\deg v = i$. For the other $\ell^{\omega-1}$ many elements $a \in \mu_\ell^{\oplus \omega}$, there exists a constant $\hat{C}_{T, \ell, q} > 0$ depending only on $T, \ell,$ and $q$ such that
\begin{equation*}
\left| \frac{\#\{ v \in \mathcal{P}_k \; | \; \deg v = i, \oplus_{j=1}^\omega \left( \frac{v}{h_j} \right)_\ell = a \in \mu_\ell^{\oplus \omega}\}}{\#\{v \in \mathcal{P}_k \; | \; \deg v = i\}} - \frac{1}{\ell^{\omega - 1}} \right| < \hat{C}_{T,\ell,q} \cdot (n \log q)^{-2 m_{n,q} + 2 \log \ell}.
\end{equation*}
\end{itemize}
\end{lemma}
Using this statement, an adaptation of the analytic argument presented in \cite[Proposition 5.13, Proposition 6.11]{park2022prime} gives the following statement.
\begin{proposition} \label{prop:major_contribution}
Suppose $\omega < 2 m_{n,q}$, and $\omega' = (1-\epsilon) \omega$ for some small enough $0 < \epsilon < 1$. Suppose that $n$ satisfies $m_{n,q} > \max \left(e^{e^e}, \; \sum_{v \in \Sigma_T} \deg v, \; 6 \log \ell + 2 \right).$
Let $T$ be an admissible $\text{Gal}(\overline{K}/K)$-module with a twisting data $\alpha := (\alpha_v)_v$. Then there exist fixed constants $\tilde{B}_{T, \ell, q}$ depending only on $T, \ell, q$ and $\gamma_T$ depending only on $T$ such that
\begin{align}
\begin{split}
& \; \; \; \; \left| \frac{\#\{ f \in \hat{F}_{(n,N),(\omega,\omega')}(\mathbb{F}_q) \; | \; \dim_{\mathbb{F}_\ell} \Sel(T, \alpha : \overline{\chi}^f) = r \}}{\# \hat{F}_{(n,N),(\omega,\omega')}(\mathbb{F}_q)} - \rho(r) \cdot \prod_{k=1}^\infty \frac{1}{1+\ell^{-k}} \prod_{k=1}^r \frac{\ell}{\ell^k-1} \right| \\
&< \tilde{B}_{T, \ell, q} \cdot (n \log q)^{4 \epsilon \log \ell} \cdot \left( (n \log q)^{-m_{n,q}} + \gamma_T^{\omega' - 1} \right),
\end{split}
\end{align}
where
\begin{equation}
\rho(r) := \lim_{n \to \infty} \left(\frac{1}{2} + (-1)^r \cdot \left(\frac{1}{2} - \sum_{\substack{k \geq 0 \\ k \equiv r \text{ mod } 2}} \frac{\#\{f \in F_n(\mathbb{F}_q) \cap \mathcal{P}_0 \; | \; \dim_{\mathbb{F}_\ell} \Sel(T, \alpha : \overline{\chi}^f) = r\}}{\#F_n(\mathbb{F}_q) \cap \mathcal{P}_0} \right) \right).
\end{equation}
We note that $\rho(r) = \frac{1}{2}$ for all non-negative integer $r \geq 0$ if and only if $\delta(\mathcal{P}_1) \neq 0$.
\end{proposition}
\begin{proof}
We summarize the key ingredients needed to prove the desired statement. Further analytic details of the proof can be found in \cite[Proposition 5.13, Proposition 6.11]{park2022prime}.
The global Selmer group $\Sel(T, \alpha: \overline{\chi}^f)$ can be rewritten as a local Selmer group as shown in the following equation:
\begin{equation}
\Sel(T, \alpha : \overline{\chi}^f) := \Sel(T, \alpha : (({\chi}^f)_v)_{v \in \Sigma_T(\mathrm{sqf}(f))}),
\end{equation}
where $\mathrm{sqf}(f)$ is the squarefree product of distinct irreducible factors of $f$.
Let $h$ be a squarefree polynomial of degree at most $n - \mathfrak{n}$ that satisfies the following conditions.
\begin{itemize}
\item $h$ has $\omega-1$ many distinct irreducible factors.
\item $h$ has $\omega'-1$ many distinct irreducible factors of degree greater than $\mathfrak{n}$.
\item All irreducible factors of $h$ are elements in the set $\Sigma_T \cup \mathcal{P}_1 \cup \mathcal{P}_2$.
\end{itemize}
Because every element in $\hat{F}_{(n,N), (\omega,\omega')}(\mathbb{F}_q)$ has an irreducible factor in $\mathcal{P}_0$ of degree greater than $\mathfrak{n}$, Lemma \ref{lemma:equidistribution} implies that the pushforward of the uniform distribution via the canonical projection
\begin{equation}
\begin{split}
\left\{ f \in \hat{F}_{(n,N),(\omega,\omega')}(\mathbb{F}_q) \; : \; h \text{ divides } f, \; v \in \mathcal{P}_0 \text{ for every } v \mid \frac{f}{h} \right\} &\to \Omega_{\Sigma_T(h)}, \\
f &\mapsto (({\chi}^f)_v)_{v \in \Sigma_T(h)},
\end{split}
\end{equation}
converges to a uniform distribution as $n$ grows arbitrarily large with explicit rate of convergence. By Proposition \ref{prop:local-Selmer}, there exists a probability distribution $\mu^*: \mathbb{Z}_{\geq 0} \to [0,1]$ and an explicit constant $B_{T, \ell, q} > 0$ such that
\begin{equation}
\left| \frac{\#\{ f \in \hat{F}_{(n,N),(\omega,\omega')}(\mathbb{F}_q) \; | \; \dim_{\mathbb{F}_\ell} \Sel(T, \alpha : \overline{\chi}^f) = r \}}{\# \hat{F}_{(n,N),(\omega,\omega')}(\mathbb{F}_q)} - \left( M_T^{\omega'-1} \mu^* \right)(r) \right| < B_{T, \ell, q} \cdot (n \log q)^{-m_{n,q}}
\end{equation}
for all $r \geq 0$.
The statement of the proposition then follows from combining the above inequality with the geometric rate of convergence of the Markov operator $M_T$ computed in Theorem \ref{thm:Markov}.
To deal with the case when $\delta(\mathcal{P}_1) = 0$ (which can only happen when $\ell \neq 2$ if $T$ is admissible), we proceed as in the proof of \cite[Theorem 8.2(i)]{KMR13}. The Markov operator $M_T$ satisfies the following relation for any probability distribution $\delta: \mathbb{Z}_{\geq 0} \to [0,1]$:
\begin{equation}
\rho(M_T \delta) = \rho(\delta).
\end{equation}
We denote by $\delta_T: \mathbb{Z}_{\geq 0} \to [0,1]$ the probability distribution
\begin{equation}
\delta_T(r) := \frac{\#\{\chi \in \Omega_T \; | \; \mathrm{rk}(\chi) = r\}}{\# \Omega_T}.
\end{equation}
We note that $\delta_T(r) \neq \frac{1}{2}$ for all non-negative $r \geq 0$, because the cardinality of $\Hom(\Gal(\overline{K}_v/K), \mu_\ell)$ is a power of $\ell$, in particular odd as long as $\ell \neq 2$. By Lemma \ref{lemma:equidistribution}, we obtain
\begin{equation}
\delta_T(r) = \lim_{n \to \infty} \frac{\#\{f \in F_n(\mathbb{F}_q) \cap \mathcal{P}_0 \; | \; \dim_{\mathbb{F}_\ell} \Sel(T, \alpha : \overline{\chi}^f) = r\}}{\#F_n(\mathbb{F}_q) \cap \mathcal{P}_0}.
\end{equation}
By induction on $\omega'$, we obtain
\begin{equation}
\rho(M_T^{\omega'-1} \mu^*) = \rho(\mu^*) = \rho(\delta_T).
\end{equation}
The construction of $\rho(r)$ follows from using the relation that $\rho(r) = \frac{1}{2} + (-1)^r \cdot \left( \frac{1}{2} - \rho(\delta_T) \right)$.
\end{proof}
Combining Proposition \ref{proposition:fan_approximation} and Proposition \ref{prop:major_contribution}, and optimizing the rate of convergence by choosing $\epsilon := (\log \log m_{n,q})^{-1}$ appearing in Proposition \ref{proposition:fan_approximation}, we prove Theorem \ref{thm:Galois_module}. Explicit computations on how such a choice of $\epsilon$ gives rise to the explicit rate of convergence can be found in \cite[Proof of Theorem 1.2, pp.37--38]{park2022prime}.
\section{Applications to twists of some superelliptic curves}
\label{sec:superelliptic}
We recall that $C$ is a non-isotrivial superelliptic curve over $K$ whose Weierstrass equation is
\begin{equation}
C: y^\ell = F(x),
\end{equation}
such that $F$ is a degree $3$ polynomial over $\mathbb{F}_q(t)$. We note that $C$ is a curve of genus $\ell-1$.
Because we assumed that $\mu_\ell \subset K$, the isogeny $1-\zeta_\ell: \mathrm{Jac}(C) \to \mathrm{Jac}(C)$ induced from the map $\zeta_\ell: C \to C$ defined as $(x,y) \mapsto (x,\zeta_\ell y)$ is a well-defined endomorphism of $\mathrm{Jac}(C)$ over $K$. Hence, we let our $\Gal(\overline{K}/K)$-module of interest to be
\begin{equation}
T := \mathrm{Jac}(C)[1-\zeta_\ell].
\end{equation}
Given a polynomial $f \in \mathbb{F}_q[t]$, we recall that $C_f$ is a cyclic order $\ell$ twist of $C$ over $K$ whose Weierstrass equation can be written as
\begin{equation}
C_f: f y^\ell = F(x).
\end{equation}
As an application of Theorem \ref{thm:Galois_module}, we compute the distribution of $1-\zeta_\ell$ Selmer groups of $\mathrm{Jac}(C_f)$ as $f$ ranges over the set of polynomials of degree $n$.
\begin{theorem}
\label{thm:superelliptic}
Fix a prime number $\ell \neq 2,3$. Let $K = \mathbb{F}_q(t)$ be a global function field whose characteristic is coprime to $2,3$, and $q \equiv 1 \text{ mod } \ell$. Let $C: y^\ell = F(x)$ be a non-isotrivial superelliptic curve over $K$. Suppose the following four conditions hold.
\begin{itemize}
\item $C$ is not defined over $K^p$.
\item The constant field of $K(\mathrm{Jac}(C)[1-\zeta_\ell])$ is equal to $\mathbb{F}_q$.
\item $\Gal(K(\mathrm{Jac}(C)[1-\zeta_\ell])/K) \cong S_3$,
\item $\mathrm{Jac}(C)$ has a place $v$ of totally split multiplicative reduction.
\end{itemize}
Denote by $\gamma_\ell$ the geometric rate of convergence of the Markov operator $\frac{5}{6} \cdot I + \frac{1}{6} \cdot M_\ell^2$.
Then for sufficiently large $n$, there exists a fixed constant $B(C,q) > 0$ independent of $n$ such that
\begin{align*}
\left| \frac{\# \{f \in F_n(\mathbb{F}_q) \; | \; \dim_{\mathbb{F}_\ell} \Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K) = r \}}{\#F_n(\mathbb{F}_q)} - \rho_C(r) \cdot \prod_{k=1}^\infty \frac{1}{1+\ell^{-k}} \prod_{k=1}^r \frac{\ell}{\ell^k-1} \right| < \frac{B(C,q)}{n^{\alpha(\ell)}},
\end{align*}
where
\begin{equation*}
\alpha(\ell) := \sup_{0 < \rho < 1} \left( \min \left( \rho \log \rho + 1 - \rho, \; \; \rho \log \gamma_\ell, \; \; \rho \log 6 \right) \right),
\end{equation*}
and
\begin{equation*}
\rho_C(r) := \lim_{n \to \infty} \left(\frac{1}{2} + (-1)^r \cdot \left(\frac{1}{2} - \sum_{\substack{k \geq 0 \\ k \equiv r \text{ mod } 2}} \frac{\#\{f \in F_n(\mathbb{F}_q) \cap \mathcal{P}_0 \; | \; \dim_{\mathbb{F}_\ell} \Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K) = k\}}{\#F_n(\mathbb{F}_q) \cap \mathcal{P}_0} \right) \right).
\end{equation*}
Furthermore, for every integer $r$, $\rho_C(r)$ is a rational number whose denominator is a power of $\ell$. In particular, $\rho_C(r) \neq \frac{1}{2}$ for all non-negative integers $r \geq 0$.
\end{theorem}
\begin{remark}
We recall that the proof of Proposition \ref{prop:major_contribution} states that $\rho_C$ is a rational number whose denominator term is a power of $\ell$. The fact that $\rho_C(r) \neq \frac{1}{2}$ for all non-negative integers $r \geq 0$ implies that there is a disparity in the distribution of dimensions of $1-\zeta_\ell$ Selmer groups of $\mathrm{Jac}(C_f)$ even if the curves $C_f$ are canonically ordered based on the degree of $f$. The disparity in dimensions of Selmer groups was first observed in \cite[Theorem A]{KMR13}, where disparity in distribution of 2-Selmer groups of quadratic twists of elliptic curves may be observed if these curves are ordered in a non-canonical manner (for instance, with respect to a fan structure over number fields). Generalizations of the disparity phenomena for quadratic twist families of principally polarized abelian varieties can be found in the works by Yu \cite{Yu16} and Morgan \cite{Mor19}. Theorem \ref{thm:superelliptic} proves that the disparity of $1-\zeta_\ell$ Selmer groups of $\mathrm{Jac}(C_f)$, unlike the case for $2$-Selmer groups of quadratic twist families of elliptic curves, still remains to be valid even after rearranging the enumeration of superelliptic curves under a canonical ordering. This phenomenon is predicted from \cite[Theorem 8.2(i)]{KMR13}, where such a disparity can occur if $\ell$ does not divide the order of $\Gal(K(T)/K)$. The module $\mathrm{Jac}(C)[1-\zeta_\ell]$ satisfies this condition; the order of the group $S_3$ is not divisible by $\ell \geq 5$.
\end{remark}
Using Theorem \ref{thm:superelliptic}, we prove the following theorem, namely the uniform upper bound on the moments of the number of $K$-rational solutions of superelliptic curves $C_f$ in the twist family $\mathcal{C}_n$. We note that the statement of the main theorem can be refined further if the function $\rho_C(r)$ from Theorem \ref{thm:superelliptic} can be explicitly determined.
\begin{theorem} \label{thm:main}
Suppose the following four conditions hold for the superelliptic curve $C: y^\ell = F(x)$.
\begin{itemize}
\item $C$ is not defined over $K^p$.
\item The constant field of $K(\mathrm{Jac}(C)[1-\zeta_\ell])$ is equal to $\mathbb{F}_q$.
\item $\Gal(K(\mathrm{Jac}(C)[1-\zeta_\ell])/K) \cong S_3$,
\item $\mathrm{Jac}(C)$ has a place $v$ of totally split multiplicative reduction.
\end{itemize}
Given a positive integer $m$, we denote by $S(\ell, p)$, $\mathbb{E}(\ell, p, m)$, and $\mathbb{P}(\ell, m)$ the following constants:
\begin{align*}
S(\ell, p) &:= 3^{\ell - 1} \cdot p^{3\ell - 3} \cdot (8\ell - 10) \cdot (\ell - 1)!, \\
\mathbb{E}(\ell, p, m) &:= \prod_{j=1}^\infty \frac{1}{1+\ell^{-j}} \cdot \max_{\rho \in [0,1]} \left[\sum_{\substack{k \geq 0 \\ k \equiv 0 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell \cdot p^{(\ell - 1)m}}{\ell^j-1}, \sum_{\substack{k \geq 1 \\ k \equiv 1 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell \cdot p^{(\ell-1)m}}{\ell^j-1} \right],\\
\mathbb{P}(\ell, m) &:= \prod_{j=1}^\infty \frac{1}{1+\ell^{-j}} \cdot \min_{\rho \in [0,1]} \left[\sum_{\substack{0 \leq k \leq m \\ k \equiv 0 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell}{\ell^j-1}, \sum_{\substack{1 \leq k \leq m \\ k \equiv 1 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell}{\ell^j-1} \right].
\end{align*}
Then for each positive integer $m$, the following two inequalities hold:
\begin{align}
\limsup_{n \to \infty} \frac{\sum_{\substack{f \in F_n(\mathbb{F}_q)}} (\# C_f(K))^m}{\#F_n(\mathbb{F}_q)} &\leq \mathbb{E}(\ell, p, m) \cdot S(\ell,p)^m \\
\liminf_{n \to \infty} \frac{\#\{f \in F_n(\mathbb{F}_q) \; | \; \#C_f(K) \leq p^{m(\ell - 1)} \cdot S(\ell,p)\}}{\# F_n(\mathbb{F}_q)} &\geq \mathbb{P}(\ell, m)
\end{align}
\end{theorem}
\begin{proof}[Proof of Theorem \ref{thm:main}]
By adapting the proof of Proposition 3.5 of \cite{Sch98} to global function fields, we have
\begin{equation}
\mathrm{rank}_\mathbb{Z}(\mathrm{Jac}(C_f)(K)) \leq (\ell-1) \cdot \dim_{\mathbb{F}_\ell} \Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K).
\end{equation}
Because $C$ is not defined over $K^p$, it follows that for any $f \in F_n(\mathbb{F}_q)$, $C_f$ is also not defined over $K^p$. This is because we have an isomorphism $C \cong C_f$ over $K(\sqrt[\ell]{f})$. If it were the case that $C_f$ is defined over $(K(\sqrt[\ell]{f}))^p = K^p(\sqrt[\ell]{f})$, then $C$ is also defined over $K^p(\sqrt[\ell]{f})$. But because the model of $C$ is already defined over $K$ and $\ell$ is coprime to $p$, we obtain that $C$ is in fact defined over $K_p$, a contradiction.
Hence for any $f \in F_n(\mathbb{F}_q)$, we can use the main theorem of \cite{BV96} and the fact that the genus of $C_f$ is equal to $\ell - 1$ to obtain
\begin{align}
\begin{split}
\# C_f(K) &\leq \# \frac{\mathrm{Jac}(C_f)(K)}{p \cdot \mathrm{Jac}(C_f)(K)} \cdot (3p)^{\ell-1} \cdot (8\ell - 10) \cdot (\ell-1)! \\
&\leq p^{(\ell - 1) \dim_{\mathbb{F}_\ell} \Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K)} \cdot 3^{\ell - 1} \cdot p^{3\ell - 3} \cdot (8 \ell - 10) \cdot (\ell - 1)!.
\end{split}
\end{align}
Applying Theorem \ref{thm:superelliptic} gives the statement of the main theorem. The fact that $\mathbb{E}(\ell, p, m) < \infty$ for each $m \geq 1$ follows from the fact that the quantity $\prod_{j=1}^k \frac{\ell}{\ell^j-1}$ decays quadratic exponentially, i.e. at a rate of $O(\ell^{-k(k-1)/2})$, as $k$ grows arbitrarily large.
\end{proof}
The upper bound computed in Theorem \ref{thm:main} is far from impressive. Nevertheless, we provide in Table \ref{tab:my_label} some approximate values of $\mathbb{E}(\ell,p,1)$, $\mathbb{P}(\ell,1)$, $\mathbb{P}(\ell,2)$ where $p, \ell \in \{5,7,11,13,17\}$, and their asymptotic values when $\ell$ grows arbitrarily large, whose implied constants of the error terms are positive and independent of the choice of $C$. In particular, at least $99\%$ of curves $C_f \in \mathcal{C}_n$ satisfies the inequality $\#C_f(K) \leq p^{2(\ell-1)} \cdot S(\ell,p)$ as $n$ grows arbitrarily large.
\begin{table}[ht]
\centering
\begin{tabular}{|c||c|c|c|c|c|c|}
\hline
& $\ell = 5$ & $\ell = 7$ & $\ell = 11$ & $\ell = 13$ & $\ell = 17$ & $\ell \to \infty$ \\
\hline
\hline
$\mathbb{E}(\ell,5,1)$ & $\times$ & $5.35713$ & $5.09091$ & $5.05494$ & $5.02451$ & $5 + O(1/\ell)$ \\
\hline
$\mathbb{E}(\ell,7,1)$ & $11.07690$ & $\times$ & $7.25454$ & $7.15385$ & $7.06863$ & $7 + O(1/\ell)$\\
\hline
$\mathbb{E}(\ell,11,1)$ & $26.76903$ & $18.57500$ & $\times$ & $11.60440$ & $11.26961$ & $11 + O(1/\ell)$\\
\hline
$\mathbb{E}(\ell,13,1)$ & $37.61501$ & $25.67499$ & $16.40459$ & $\times$ & $13.44608$ & $13 + O(1/\ell)$\\
\hline
$\mathbb{E}(\ell,17,1)$ & $66.07605$ & $43.59997$ & $27.42754$ & $23.29292$ & $\times$ & $17 + O(1/\ell)$\\
\hline
\hline
$\mathbb{P}(\ell,1)$ & $0.79334$ & $0.85459$ & $0.90840$ & $0.92265$ & $0.94098$ & $1 - O(1/\ell)$ \\
\hline
$\mathbb{P}(\ell,2)$ & $0.99167$ & $0.99702$ & $0.99924$ & $0.99954$ & $0.99980$ & $1 - O(1/\ell^{3})$\\
\hline
\end{tabular}
\vspace{0.5pt}
\caption{A table of approximate values of $\mathbb{E}(\ell,p,1)$, $\mathbb{P}(\ell,1))$, and $\mathbb{P}(\ell,2)$ for some first $5$ values of $\ell$ and $p$ up to 5 decimals. We also include the asymptotic values of these constants as $\ell$ grows arbitrarily large with error terms. We omit the cases where $\ell = p$ for which our main theorem is not applicable.}
\label{tab:my_label}
\end{table}
\begin{remark}
In case $\rho_C(r)$ is explicitly computable, we can obtain an improved statement of Theorem \ref{thm:main}. Given a positive integer $m$, we define constants $\mathbb{E}_C(\ell, p, m)$ and $\mathbb{P}_C(\ell, p, m)$ as follows, where $\rho := \rho_C(0)$:
\begin{align*}
\mathbb{E}_C(\ell, p, m) &:= \prod_{j=1}^\infty \frac{1}{1+\ell^{-j}} \cdot \left[\rho \cdot \sum_{\substack{k \geq 0 \\ k \equiv 0 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell \cdot p^{(\ell - 1)m}}{\ell^j-1} + (1-\rho) \cdot \sum_{\substack{k \geq 1 \\ k \equiv 1 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell \cdot p^{(\ell-1)m}}{\ell^j-1} \right],\\
\mathbb{P}_C(\ell, p, m) &:= \rho \cdot \sum_{\substack{0 \leq k \leq m \\ k \equiv 0 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell}{\ell^j-1} + (1-\rho) \cdot \sum_{\substack{1 \leq k \leq m \\ k \equiv 1 \text{ mod } 2}} \prod_{j=1}^k \frac{\ell}{\ell^j-1}.
\end{align*}
Then for each non-negative integer $m$, we have
\begin{align*}
\limsup_{n \to \infty} \frac{\sum_{\substack{f \in F_n(\mathbb{F}_q)}} (\# C_f(K))^m}{\#F_n(\mathbb{F}_q)} &\leq \mathbb{E}_C(\ell, p, m) \cdot S(\ell,p)^m \\
\liminf_{n \to \infty} \frac{\#\{f \in F_n(\mathbb{F}_q) \; | \; \#C_f(K) \leq p^{m(\ell - 1)} \cdot S(\ell,p)\}}{\# F_n(\mathbb{F}_q)} &\geq \mathbb{P}_C(\ell, p, m).
\end{align*}
\end{remark}
Before we prove Theorem \ref{thm:superelliptic}, we state and prove a few lemmas.
\begin{lemma} \label{lemma:T_superelliptic}
The following properties hold for the superelliptic curve $C: y^\ell = F(x)$ over $K$ such that $F$ is a degree $3$ polynomial over $\mathbb{F}_q(t)$.
\begin{itemize}
\item The torsion subgroup $\mathrm{Jac}(C)[1-\zeta_\ell]$ is a 2-dimensional $\mathbb{F}_\ell$ vector space generated by a set of divisors $\{(P_2 - P_1), (P_3 - P_1)\}$, where $P_i$'s are three Weierstrass points of $C$.
\item For any polynomial $f \in \mathbb{F}_q[t]$, there is a $\Gal(\overline{K}/K)$-equivariant isomorphism
\begin{equation}
\mathrm{Jac}(C_f)[1-\zeta_\ell] \cong \mathrm{Jac}(C)[1-\zeta_\ell]
\end{equation}
\end{itemize}
\end{lemma}
\begin{proof}
The proof follows from \cite[Proposition 1.7]{Yu16} and \cite[Proposition 3.2]{Sch98}. The details of the proof for the second statement of the lemma can be obtained as follows. Denote by $\alpha_1, \alpha_2,\alpha_3$ the roots of the cubic polynomial $F$ in $\overline{K}$. Then we can rewrite the Weierstrass equation for $C$ as
\begin{equation*}
y^\ell = (x-\alpha_1)(x-\alpha_2)(x-\alpha_3).
\end{equation*}
As a convention, we denote by $P_1 = (\alpha_1,0)$, $P_2 = (\alpha_2,0)$, and $P_3 = (\alpha_3,0)$.
Let $b$ be an integer such that $b\ell \equiv 1 \text{ mod } 3$. We observe that multiplying both sides of the Weierstrass equation for $C_f$ by $f^{b\ell - 1}$ and applying change of coordinates $(x^{\frac{b\ell-1}{3}},y^b) \mapsto (X,Y)$ gives
\begin{equation*}
Y^\ell = (X - f^{\frac{b\ell - 1}{3}} \alpha_1)(X - f^{\frac{b\ell - 1}{3}} \alpha_2)(X - f^{\frac{b\ell - 1}{3}} \alpha_3)
\end{equation*}
Because $f^{\frac{b\ell-1}{3}} \in \mathbb{F}_q[t]$, it follows that the bijection defined as
\begin{align*}
\mathrm{Jac}(C_f)[1-\zeta_\ell] &\to \mathrm{Jac}(C)[1-\zeta_\ell], \\
((f^{\frac{b\ell - 1}{3}} \alpha_2,0) - (f^{\frac{b\ell - 1}{3}} \alpha_1,0)) &\mapsto ((\alpha_2,0) - (\alpha_1,0)), \\
((f^{\frac{b\ell - 1}{3}} \alpha_2,0) - (f^{\frac{b\ell - 1}{3}} \alpha_1,0)) &\mapsto ((\alpha_3,0) - (\alpha_1,0)),
\end{align*}
is a $\Gal(\overline{K}/K)$-equivariant isomorphism.
\end{proof}
\begin{lemma} \label{lemma:T_conditions}
Suppose the superelliptic curve $C$ satisfies the four conditions from Theorem \ref{thm:superelliptic}. Then $\mathrm{Jac}(C)[1-\zeta_\ell]$ is an admissible $\Gal(\overline{K}/K)$-module.
\end{lemma}
\begin{proof}
Lemma \ref{lemma:T_superelliptic} implies statement (1) from Condition \ref{condition:admissible}. Statement (2) follows from the fact that the map $1-\zeta_\ell: \mathrm{Jac}(C) \to \mathrm{Jac}(C)$ is self-dual. This can be proven by line-by-line adaptation of the argument presented in \cite[Proposition 3.1]{Sch98}. Because $\deg(1-\zeta_\ell) = \ell^2$ is odd, we can use \cite[Section 4, Case (I), page 259]{PR12} to obtain statement (2). The first condition from Theorem \ref{thm:superelliptic} is identical to statement (3) of Condition \ref{condition:admissible}.
We now show that the second condition from Theorem \ref{thm:superelliptic} implies statements (4),(5), and (6) of Condition \ref{condition:admissible}. We represent $(P_2 - P_1)$ and $(P_3 - P_1)$ as column vectors with entries in $\mathbb{F}_\ell$:
\begin{equation*}
(P_2 - P_1) \mapsto \begin{pmatrix} 1 \\ 0 \end{pmatrix}, \hspace{4pt} (P_3 - P_1) \mapsto \begin{pmatrix} 0 \\ 1 \end{pmatrix}.
\end{equation*}
We recall that the absolute Galois group $\Gal(\overline{K}/K)$ acts on $\mathrm{Jac}(C)[1-\zeta_\ell]$ by mapping the divisor $[P_i - P_j]$ to $[\sigma P_i - \sigma P_j]$ for any $\sigma \in \Gal(\overline{K}/K)$. The elements of $\Gal(K(\mathrm{Jac}(C)[1-\zeta_\ell])/K) \cong S_3$ can be represented by elements in $\GL_2(\mathbb{F}_\ell)$ as:
\begin{align*}
() \mapsto \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix}, \hspace{4pt} (1 2) &\mapsto \begin{pmatrix} -1 & -1 \\ 0 & 1 \end{pmatrix}, \hspace{4pt} (1 3) \mapsto \begin{pmatrix} 1 & 0 \\ -1 & -1 \end{pmatrix}, \\
(2 3) \mapsto \begin{pmatrix} 0 & 1 \\ 1 & 0 \end{pmatrix}, \hspace{4pt} (1 2 3) &\mapsto \begin{pmatrix} -1 & -1 \\ 1 & 0 \end{pmatrix}, \hspace{4pt} (1 3 2) \mapsto \begin{pmatrix} 0 & 1 \\ -1 & -1 \end{pmatrix}.
\end{align*}
The group acts on the divisors via matrix multiplication. Statement (4) follows from the fact that $\ell \geq 5$ and there are no proper subspaces of $\mathbb{F}_\ell^{\oplus 2}$ that is invariant with respect to multiplication by $6$ of the matrices above. Statement (5) can be checked from the fact that the only matrix in $\text{M}_{2 \times 2}(\mathbb{F}_\ell)$ such that commutes with all the $6$ matrices above are scalar multiples of the identity matrix. Statement (6) follows from the fact that the order of the group $S_3$ is coprime to $\ell$.
The third condition from Theorem \ref{thm:superelliptic} implies that at the place $v$ the abelian variety $\mathrm{Jac}(C)$ is isomorphic to $\mathbb{G}_m^{\ell-1} / \prod_{i=1}^{\ell-1} q_i^\mathbb{Z}$ for some elements $q_1, q_2, \cdots, q_{\ell - 1}$ of $\mathbb{G}_m$ with non-trivial valuations with respect to $v$. It follows that $\mathrm{Jac}(C)[\ell](\overline{K}_v) \cap \mu_{\ell^\infty} = \mu_\ell$, from which Statement (7) follows.
\end{proof}
We now prove Theorem \ref{thm:superelliptic}. By Lemma \ref{lemma:T_superelliptic}, we have
\begin{equation*}
\Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K) \subset H^1_\et(K, \mathrm{Jac}(C)[1-\zeta_\ell]).
\end{equation*}
Let $\alpha$ be a twisting data defined as
\begin{equation}
\alpha(\overline{\chi}^f) := \begin{cases}
\mathrm{Image} \left( \delta_v: \frac{\mathrm{Jac}(C_f)(K_v)}{(1-\zeta_\ell)(\mathrm{Jac}(C_f)(K_v))} \to H^1_{\et}(K_v, \mathrm{Jac}(C)[1-\zeta_\ell]) \right) &\text{ if } v \in \Sigma \text{ or } v \mid f, \\
H^1_{\et}(\Oh_{K_v}, \mathrm{Jac}(C)[1-\zeta_\ell]) &\text{ otherwise }.
\end{cases}
\end{equation}
We observe that
\begin{equation}
\Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K) = \Sel(\mathrm{Jac}(C)[1-\zeta_\ell], \alpha : \overline{\chi}^f).
\end{equation}
By Lemma \ref{lemma:T_conditions}, the torsion subgroup $\mathrm{Jac}(C)[1-\zeta_\ell]$ is an admissible $\Gal(\overline{K}/K)$-module. Applying Theorem \ref{thm:Galois_module} and Chebotarev density theorem for $S_3$ Galois extensions proves Theorem \ref{thm:superelliptic}. In particular, the governing Markov operator is given by
\begin{equation}
M_{\mathrm{Jac}(C)[1-\zeta_\ell]} := \frac{5}{6} \cdot I + \frac{1}{6} \cdot M_\ell^2.
\end{equation}
We note that the set of places $\mathcal{P}_1$ is empty because no Frobenius elements in $\Gal(K(\mathrm{Jac}(C)[1-\zeta_\ell])/K)$ are of orders equal to $\ell \geq 5$. This is why we have $\delta(\mathcal{P}_0) = \frac{5}{6}$, unlike the case where $\ell = 2$ as explored in \cite{park2022prime}. In the latter case, the governing Markov operator was given by
\begin{equation*}
M_L := \frac{1}{3} \cdot I + \frac{1}{2} \cdot M_2 + \frac{1}{6} \cdot M_2^2.
\end{equation*}
In this case, because $\delta(\mathcal{P}_1) = \delta(\mathcal{P}_0) + \delta(\mathcal{P}_1)$, we obtain that $\mu(M_L \delta) = \frac{1}{2}$ for any probability distribution $\delta: \mathbb{Z}_{\geq 0} \to [0,1]$, which is consistent with the parity of dimensions of 2-Selmer groups of quadratic twist families of elliptic curves over $K$.
\begin{remark} \label{remark:PR}
We compare the statement of Theorem \ref{thm:superelliptic} with the predicted distribution from Poonen-Rains heuristics \cite{PR12}. Because $\mathrm{Jac}(C)[1-\zeta_\ell]$ is a direct summand of the permutation $\mathbb{F}_\ell$ vector space arising from the set of roots of $F$, Proposition 3.3.(b) and Theorem 4.14 of \cite{PR12} implies that $\Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K)$ is an intersection of the images of the two homomorphisms
\begin{equation}
\begin{tikzcd}
& H^1_\et(K,\mathrm{Jac}(C)[1-\zeta_\ell]) \arrow{d}{} \\
\prod_{v \text{ place of } K} \frac{\mathrm{Jac}(C_f)(K_v)}{(1-\zeta_\ell)\mathrm{Jac}(C_f)(K_v)} \arrow{r}{} & \prod_{v \text{ place of } K} H^1_\et(K_v, \mathrm{Jac}(C)[1-\zeta_\ell]),
\end{tikzcd}
\end{equation}
both of which are maximal isotropic subspaces of the codomain of two homomorphisms with respect to quadratic forms induced from the collection of orthogonal pairings $\{\mu_{T,v}\}_v$. The assumption of the Poonen-Rains heuristics can be summarized as follows.
\begin{itemize}
\item One of the images of the two homomorphisms is fixed.
\item The other image of the two homomorphisms varies uniformly at random over the set of maximal isotropic subspaces of $\prod_{v \text{ place of } K} H^1_\et(K_v, \mathrm{Jac}(C)[1-\zeta_\ell])$.
\end{itemize}
The Galois-equivariant isomorphism obtained from Lemma \ref{lemma:T_superelliptic} implies that the image of the restriction map (or the vertical map from the commutative diagram) satisfies the first condition. However, the description of the twisting data $\alpha$ defining the Selmer group $\Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K)$ does not satisfy the second condition. This is because given any place $v \in \mathcal{P}_2$, the Poonen-Rains heuristics assumes that the images of the local Kummer maps (or the horizontal map from the commutative diagram) vary uniformly at random over $2(\ell+1)$ many maximal isotropic subspaces of $H^1_\et(K_v, \mathrm{Jac}(C)[1-\zeta_\ell])$(see Proposition 2.6(b) of \cite{PR12} for the number of maximal isotropic subspaces), whereas the proof of Theorem \ref{thm:superelliptic} shows that the images of the local Kummer maps range over only $\ell + 1$ of the maximal isotropic subspaces of $H^1_\et(K_v, \mathrm{Jac}(C)[1-\zeta_\ell])$. The discrepancy in random matrix theoretic conditions governing Selmer groups of Jacobians of superelliptic curves is represented by the function $\rho_C(r)$ which measures the disparity of dimensions of $\Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_f)/K)$.
As a related note, the discrepancy in random matrix theoretic conditions governing Selmer groups of admissible $\Gal(\overline{K}/K)$-modules $T$ given a twisting data $\alpha$ is also represented by the function $\rho(r)$ constructed in Theorem \ref{thm:Galois_module}. We note that the condition $\delta(\mathcal{P}_1) \neq 0$ is equivalent to the condition that there is a positive density of places $v$ of $K$ such that both assumptions of the Poonen-Rains heuristics hold. Indeed, given any place $v \in \mathcal{P}_1$, the Poonen-Rains heuristics assumes that the images of the local Kummer maps vary uniformly at random over $2$ maximal isotropic subspaces of $H^1_\et(K_v, \mathrm{Jac}(C)[1-\zeta_\ell])$, a condition that is also satisfied by local conditions defining $\Sel(T, \alpha : \overline{\chi})$.
\end{remark}
\begin{remark}
The analogue of the results obtained in this manuscript for twist families of superelliptic curves over number fields will be explored in an upcoming joint work with Daniel Keliher \cite{KP25-2}.
\end{remark}
\begin{comment}
\begin{remark}
We note that the number field analogue of Theorem \ref{thm:superelliptic} can be obtained for twist families of superelliptic curves with respect to a non-canonical ordering known as the ``fan structure'', as introduced in the work by Klagsbrun--Mazur--Rubin \cite{KMR14}.
An exemplary analogue can be formulated as follows. Fix an integer $\ell$. Let $K$ be a number field such that $\mu_\ell \subset K$. Given a positive integer $m$ and a positive number $X$, we denote by $\mathcal{B}_m(X)$ the collection of square-free products of prime ideals in $\Oh_K$ defined as
\begin{equation*}
\mathcal{B}_m(X) := \left\{ \mathfrak{d} := \prod_{i=1}^m \mathfrak{p}_i \; : \; \mathfrak{p}_i \neq \mathfrak{p}_j \text{ for all } 1 \leq i < j \leq m \text{ and } \mathrm{Nm}_\mathbb{Q}^K(\mathfrak{p}_i) \leq X^{i^2} \right\}.
\end{equation*}
We denote by $\mathcal{D}_m(X)$ the collection of square-free elements in $\Oh_K$ defined as
\begin{equation*}
\mathcal{D}_m(X) := \left\{ d \in \Oh_K : d \text{ is } \ell\text{th power free and } \exists! \; \mathfrak{d} \in \mathcal{B}_m(X) \text{ such that } d \in \mathfrak{p} \text{ for every } \mathfrak{p} \mid \mathfrak{d} \right\}.
\end{equation*}
As before, we denote by $C: y^\ell = F(x)$ a superelliptic curve over $K$ that satisfies the condition that $\Gal(K(\mathrm{Jac}(C)[1-\zeta_\ell])/K) \cong S_3$. We denote by $C_d$ the twist of $C$ by $d$, i.e. $C_d: dy^\ell = F(x)$. By applying Theorem 11.6 of \cite{KMR14}, we obtain that there exists a function $\rho_C \in \mathbb{Z}_{\geq 0} \to [0,1]$ such that $\rho_C(r)$ is a rational number with denominator being a power of $\ell$ such that
\begin{equation*}
\lim_{m \to \infty} \lim_{X \to \infty} \frac{\# \left\{ d \in \mathcal{D}_m(X) \; : \; \dim_{\mathbb{F}_\ell} \Sel_{1-\zeta_\ell}(\mathrm{Jac}(C_d)/K) = r \right\}}{\# \mathcal{D}_m(X)} = \rho_C(r) \cdot \prod_{k=1}^\infty \frac{1}{1 + \ell^{-k}} \prod_{k=1}^r \frac{\ell}{\ell^k-1}.
\end{equation*}
In an ongoing joint work with Daniel Keliher, we aim to compute the distribution of $1-\zeta_\ell$ Selmer groups of Jacobians of twisted superelliptic curves $C_d$ where $d$ varies over a canonically ordered set of $\ell$-th power free elements of $\Oh_K$. By doing so, we hope to apply uniform effective Mordell-Lang for curves \cite{KRZB16, DGH21} to canonically ordered twist families of superelliptic curves $C_d$.
\end{remark}
\end{comment}
\begin{thebibliography}{KRZB16}
\bibitem[BG13]{BG13}
Manjul Bhargava and Benedict Gross.
\newblock The average size of the 2-{S}elmer group of {J}acobians of hyperelliptic curves having a rational {W}eierstrass point.
\newblock In {\em Automorphic representations and {$L$}-functions}, volume~22 of {\em Tata Institute of Fundamental Research Studies in Mathematics}, pages 23--91. Tata Institute of Fundamental Research, Mumbai, 2013.
\bibitem[BV96]{BV96}
Alexandru Buium and Jos\'e~Felipe Voloch.
\newblock Lang’s conjecture in characteristic $p$: an explicit bound.
\newblock {\em Compositio Mathematica}, 103(1):1--6, 1996.
\bibitem[DGH21]{DGH21}
Vesselin Dimitrov, Ziyang Gao, and Philipp Habegger.
\newblock Uniformity in {M}ordell–{L}ang for curves.
\newblock {\em Annals of Mathematics}, 194(1):237--298, 2021.
\bibitem[EL23]{EL23}
Jordan Ellenberg and Aaron Landesman.
\newblock Homological stability for generalized {Hurwitz} spaces and {Selmer groups} in quadratic twist families over function fields, 2023.
\newblock arXiv:2310.16286.
\bibitem[Fal83]{Faltings83}
Gerd Faltings.
\newblock Endlichkeitss\"atze f\"ur abelsche variet\"aten \"uber zahlk\"orpern.
\newblock {\em Inventiones Mathematicae (in German)}, 73:349--366, 1983.
\bibitem[FJ08]{FJ08}
Michael~D. Fried and Moshe Jarden.
\newblock {\em {Field Arithmetic}}.
\newblock Springer, Berlin, Germany, 2008.
\bibitem[FWY20]{FWY20}
Tingting Feng, Shaochen Wang, and Guangyu Yang.
\newblock Large and moderate deviation principles for the {Erd\"os-Kac} theorem in function fields.
\newblock {\em Statistics and probability letters}, 163, 2020.
\bibitem[Gra65]{Grauert65}
Hans Grauert.
\newblock Mordells vermutung \"uber rationale punkte auf algebraischen kurven und funktionenk\"orper.
\newblock {\em Publications Math\`ematiques de l'IH\`ES (in German)}, 25(25):131--149, 1965.
\bibitem[KMR13]{KMR13}
Zev Klagsbrun, Barry Mazur, and Karl Rubin.
\newblock Disparity in {Selmer} ranks of quadratic twists of elliptic curves.
\newblock {\em Annals of Mathematics}, 178(1):287--320, 2013.
\bibitem[KMR14]{KMR14}
Zev Klagsbrun, Barry Mazur, and Karl Rubin.
\newblock A {M}arkov model for {S}elmer ranks in families of twists.
\newblock {\em Compositio Mathematica}, 150:1077--1106, 2014.
\bibitem[KP25]{KP25-2}
Daniel Keliher and Sun~Woo Park.
\newblock Distribution of {Selmer} ranks in prime cyclic extensions, 2025.
\newblock Preprint, Draft available upon request.
\bibitem[KRZB16]{KRZB16}
Eric Katz, Joseph Rabinoff, and David Zureick-Brown.
\newblock Uniform bounds for the number of rational points on curves of small {Mordell-Weil} rank.
\newblock {\em Duke Mathematical Journal}, 165(16):3189--3240, 2016.
\bibitem[Man63]{Manin63}
Yuri Manin.
\newblock ``{Rational points on algebraic curves over function fields}''.
\newblock {\em Seriya Matematicheskaya (in Russian)}, 27:1395--1440, 1963.
\bibitem[Mor19]{Mor19}
Adam Morgan.
\newblock Quadratic twists of abelian varieties and disparity in {Selmer} ranks.
\newblock {\em Algebra and Number Theory}, 13(4):839--899, 2019.
\bibitem[MT93]{MT93}
Sean Meyn and Richard Tweedie.
\newblock {\em Markov chains and stochastic stability}.
\newblock Springer-Verlag, Berlin, Germany, 1993.
\bibitem[Par22]{park2022prime}
Sun~Woo Park.
\newblock On the prime {S}elmer ranks of cyclic prime twist families of elliptic curves over global function fields, 2022.
\newblock arXiv:2211.11486. To appear at Compositio Mathematica.
\bibitem[PR12]{PR12}
Bjorn Poonen and Eric Rains.
\newblock Random maximal isotropic subspaces and {S}elmer groups.
\newblock {\em Journal of the AMS}, 25(1):245--269, 2012.
\bibitem[PS14]{PS14}
Bjorn Poonen and Michael Stoll.
\newblock Most odd degree hyperelliptic curves have only one rational point.
\newblock {\em Annals of Mathematics}, 180(3):1137--1166, 2014.
\bibitem[Sch98]{Sch98}
Edward Schaefer.
\newblock Computing a {S}elmer group of a {J}acobian using functions on the curve.
\newblock {\em Mathematische Annalen}, 310:447--471, 1998.
\bibitem[SD08]{SD08}
Peter Swinnerton-Dyer.
\newblock The effect of twisting on the 2-{Selmer} group.
\newblock {\em Mathematical Proceedings of the Cambridge Philosophical Society}, 145(3):513--526, 2008.
\bibitem[Sha19]{Sh19}
Ananth Shankar.
\newblock {2-Selmer groups of hyperelliptic curves with marked points}.
\newblock {\em Transactions of the American Mathematical Society}, 372:267--304, 2019.
\bibitem[SW18]{SW18}
Arul Shankar and Xiaoheng Wang.
\newblock {Rational points on hyperelliptic curves having a marked non-Weierstrass point}.
\newblock {\em Compositio Mathematica}, 154(1):188–222, 2018.
\bibitem[Thi23]{Thinh23}
Dao~Van Thinh.
\newblock {2-Selmer groups of even hyperelliptic curves over function fields}.
\newblock {\em Transactions of the American Mathematical Society}, 376:4679--4712, 2023.
\bibitem[Voj89]{Vojta89}
Paul Vojta.
\newblock Mordell's conjecture over function fields.
\newblock {\em Inventiones Mathematicae}, 98:115--138, 1989.
\bibitem[Voj91]{Vojta91}
Paul Vojta.
\newblock Siegel's theorem in the compact case.
\newblock {\em Annals of Mathematics}, 133(3):509--548, 1991.
\bibitem[Yu16]{Yu16}
Myungjun Yu.
\newblock {Selmer ranks of twists of hyperelliptic curves and superelliptic curves}.
\newblock {\em Journal of Number Theory}, 160:148--185, 2016.
\bibitem[Yu19]{Yu19}
Myungjun Yu.
\newblock {The distribution of Selmer ranks of quadratic twists of Jacobians of hyperelliptic curves}.
\newblock {\em Mathematical Research Letters}, 26(4):1217--1250, 2019.
\end{thebibliography}