---
title: Periodic Points of Polynomials in Finite Fields
url: https://www.emergentmind.com/topics/periodic-points-of-polynomials-over-finite-fields
type: topic
---

# Periodic Points of Polynomials in Finite Fields

A periodic point of a polynomial $\phi$ over a finite field is an element $x$ such that $\phi^k(x) = x$ for some $k \geq 1$, where composition is iterated $k$ times. These points constitute foundational objects in arithmetic dynamics, with their density, distribution, and structural properties conveying rich arithmetic and group-theoretic information. Recent advances provide explicit computations for select polynomial families, effective bounds and average behaviors for generic classes, and group-theoretic interpretations based on monodromy and random mappings.

## 1. Fundamental Notions and Historical Context

Given a finite field $\mathbb{F}_q$ of $q = p^r$ elements, and a polynomial $\phi \in \mathbb{F}_q[x]$ of degree at least $2$, the study of $\phi$'s dynamics on $\mathbb{P}^1(\mathbb{F}_{q^n}) = \mathbb{F}_{q^n} \cup \{\infty\}$ involves classifying points as periodic or strictly preperiodic. The set of periodic points under $\phi$ in $\mathbb{P}^1(\mathbb{F}_{q^n})$ is
$$
\operatorname{Per}(\phi, \mathbb{P}^1(\mathbb{F}_{q^n})) = \{x : \phi^k(x) = x \text{ for some } k \geq 1\}.
$$
The relative density of periodic points is defined as
$$
a_n = \frac{|\operatorname{Per}(\phi, \mathbb{P}^1(\mathbb{F}_{q^n}))|}{q^n+1}.
$$
Basic structure theory shows that each point in a finite field’s orbit under $\phi$ is eventually periodic (i.e., eventually falls into a cycle). Power maps $\phi(x) = x^d$ and Chebyshev polynomials have long been known to admit explicit periodic-point counts due to their algebraic group origins and cyclotomic structure, serving as canonical examples for both random-like and highly structured dynamical behavior [1301.6158][1710.07821].

## 2. Counting Periodic Points: Explicit Formulas and Families

### 2.1. Power Maps and Chebyshev Polynomials

The periodic points of $\phi(x) = x^d$ can be classified via group-theoretic and valuation techniques. If $p^n - 1 = q^{e} m$ with $\gcd(q, m) = 1$, the number of periodic points is $m+1$, resulting in limiting proportions in carefully chosen towers:
$$
\lim_{n\to\infty,~\Delta|n,~v_q(n)=v}  \frac{|\operatorname{Per}(\phi, \mathbb{F}_{p^n})|}{p^n} = \frac{1}{q^{\lambda + v}}
$$
where $\Delta = \operatorname{ord}_q(p)$, $\lambda = v_q(p^\Delta - 1)$ [1301.6158].

Chebyshev polynomials $T_d$ satisfy $T_d(z + z^{-1}) = z^d + z^{-d}$ and decompose further through the properties of cyclotomic fields and their relation to lifts through quadratic extensions. For odd $d = q$, their limiting density is
$$
\lim_{n\to\infty}  \frac{|\operatorname{Per}(T_q,\mathbb{F}_{p^n})|}{p^n} = \frac{q^{\lambda+v} + 1}{2 q^{\lambda+v}}
$$
with analogous expressions in the $d=2$ case and for composite degrees.

### 2.2. Additive Polynomials

For additive (i.e., $\mathbb{F}_p$-linear) polynomials $A(X) = \sum_{i=0}^d a_i X^{p^i}$, the count of periodic points in $\mathbb{F}_{q^n}$ is explicitly given by
$$
T_A(n) = q^n - q^{s_A(t(n))}
$$
where $s_A(n)$ is the degree of the splitting field of $A^{(n)}(X)$, and $t(n)$ is maximal such that $s_A(t(n))$ divides $n$. The function $s_A(n)$ exhibits a $p$-adic step-function form up to finitely many initial irregularities and determines both oscillating and limiting behaviors for periodic-point proportions [2502.19141]. In “regular” cases,
$$
s_A(n) = \alpha p^{\lceil \log_p (\beta n) \rceil}
$$
with rational constants $\alpha, \beta$. The proportion $T_A(n)/q^n$ may not converge, typically oscillating or tending to $1$ only for special exceptional cases.

### 2.3. Split Polynomial Maps and Higher Dimensions

In the split map setting $f = f_1 \times \cdots \times f_N$, the total number of periodic points in affine or projective space is multiplicative:
$$
|\operatorname{Per}(f, \mathbb{A}^N(\mathbb{F}_q))| = \prod_{i=1}^N |\operatorname{Per}(f_i, \mathbb{A}^1(\mathbb{F}_q))|
$$
This reduction applies especially to powering and Chebyshev coordinates, allowing for explicit periodic-point enumeration in high-dimensional settings [1710.07821].

## 3. Asymptotic and Average Behavior

Explicit calculations for general polynomials are rare. However, probabilistic and average results have been established:

- For random degree-$d$ polynomials over $\mathbb{F}_q$, the expected number of points of exact period $k \leq d$ is:
$$
\mathbb{E}_{f \in \mathcal{P}_d(\mathbb{F}_q)} [\#\{x \in \mathbb{F}_q : f^k(x) = x,\; x \text{ period } k\}] = \frac{q(q-1)\cdots(q-k+1)}{q^k}
$$
The total expected number of periodic points is therefore a truncated sum over $k$ [1108.4132].

- For fixed $d$ and $q \to \infty$, it is conjectured but not yet proved that the expected number behaves asymptotically as $C q^{1/4}$ with $C = \sqrt{\pi/2}$, which matches the behavior for random mappings.

- For quadratic polynomials $f \in \mathbb{F}_{p^r}[x]$, the average proportion of periodic points in $\mathbb{P}^1(\mathbb{F}_{p^r})$ is less than $22/\log\log(p^r)$ for large $r$,
$$
\mathbb{E}_{\deg f=2} \left[P_{p, r}(f)\right] = O(1/\log\log(p^r))
$$
A uniformity theorem on Galois specializations ensures this decay can be obtained for broad families via Chebotarev-theoretic arguments [2103.16533].

## 4. Non-Exceptional Quadratic Polynomials: Monodromy and Martingales

The periodic-point density for quadratic polynomials with strictly preperiodic critical points and not conjugate to Chebyshev or Lattès maps displays a qualitatively different vanishing phenomenon.

- For $q$ an odd square and such a quadratic $\phi$, $\lim_{n \to \infty} a_n = 0$.
- More generally, $\liminf_{n \to \infty} a_n = 0$ as long as $\phi$ is non-exceptional.

The proof employs a lift of $\phi$ to a post-critically finite polynomial $f$ over $\mathbb{C}$ and constructs the geometric iterated monodromy group (IMG). At finite levels, the proportion of periodic points is controlled by the proportion of IMG elements fixing some vertex in the preimage tree:
$$
a_n \leq \operatorname{FPP}(G_n) + o(1),\quad \operatorname{FPP}(G_n) = \frac{|\{g \in G_n: g \text{ fixes some vertex at level $n$}\}|}{|G_n|}
$$
A martingale argument, the “fixed-point process,” shows that the limiting fixed-point proportion of the IMG is zero, forcing $a_n$ to $0$ [2107.10310].

## 5. Algorithmic and Structural Applications

Explicit periodic-point formulas for powering, Chebyshev, and additive polynomials underpin classification algorithms for polynomial dynamics over finite fields. For a split map $F: \mathbb{P}^N \to \mathbb{P}^N$, the structure of its cycle statistics under varying primes or field extensions enables identification of its decomposition type (random, power, Chebyshev, or mixed). Such algorithms use growth rates, explicit periodic-point counts, and tail-length statistics, as described in [1710.07821].

Beyond model families, the interaction of permutation structure, functional graph component counts, and random-mapping analogies raises further long-cycle and "generic" behavior questions [1108.4132]. These are linked to unresolved heuristics in number theory and algorithms, such as Pollard rho collision frequencies.

## 6. Open Problems and Research Directions

Major unresolved directions include:

- A precise understanding of exceptional cases (notably Lattès maps) over finite fields and explicit limiting densities therein [2107.10310].
- Extension of martingale and monodromy techniques to higher-degree polynomials and general rational maps, especially for towers of function field extensions.
- Improved asymptotic bounds and explicit constants for expected periodic-point densities for general polynomial families.
- Analysis of constant-field extensions in the case of polynomial reductions from number fields.

A plausible implication is that further advances in Galois-theoretic uniformity, monodromy group classification, and random mapping models will deepen the classification of periodic-point behavior for more general dynamical systems over finite fields.

---

**Key References (by arXiv id):**
- [2107.10310]: Iterated monodromy groups and periodic points for rational functions over finite fields (Bridý–Jones–Kelsey–Lodge)
- [2502.19141]: Periodic points of additive polynomials and step-function law (Reis)
- [1301.6158]: Periodic points in towers for power maps and Chebyshev polynomials (Manes–Thompson)
- [2103.16533]: Average density bounds for quadratic polynomials (Logan)
- [1710.07821]: Split polynomial maps, cycle statistics, and classification (Hutz–Patel)
- [1108.4132]: Functional-graph statistics and average numbers of periodic points (Boston–Jones–Konyagin–Sutherland)

Source: https://www.emergentmind.com/topics/periodic-points-of-polynomials-over-finite-fields