Arboreal Galois Representation

In arithmetic dynamics, an arboreal Galois representation is a continuous group homomorphism between the absolute Galois group of a field and the automorphism group of an infinite, regular, rooted tree.

The study of arboreal Galois representations of goes back to the works of Odoni in 1980s.

Definition

Let $K$ be a field and $K^$ be its separable closure. The Galois group $G_K$ of the extension $K^/K$ is called the absolute Galois group of $K$. This is a profinite group and it is therefore endowed with its natural Krull topology.

For a positive integer $d$, let $T^d$ be the infinite regular rooted tree of degree $d$. This is an infinite tree where one node is labeled as the root of the tree and every node has exactly $d$ descendants. An automorphism of $T^d$ is a bijection of the set of nodes that preserves vertex-edge connectivity. The group $Aut(T^d)$ of all automorphisms of $T^d$ is a profinite group as well, as it can be seen as the inverse limit of the automorphism groups of the finite sub-trees $T^d_n$ formed by all nodes at distance at most $n$ from the root. The group of automorphisms of $T^d_n$ is isomorphic to $S_d\wr S_d\wr \ldots \wr S_d$, the iterated wreath product of $n$ copies of the symmetric group of degree $d$.

An arboreal Galois representation is a continuous group homomorphism $G_K \to Aut(T^d)$.

Arboreal Galois representations attached to rational functions

The most natural source of arboreal Galois representations is the theory of iterations of self-rational functions on the projective line. Let $K$ be a field and $f \colon \mathbb P^1_K\to \mathbb P^1_K$ a rational function of degree $d$. For every $n\geq 1$ let $f^n=f\circ f\circ \ldots \circ f$ be the $n$-fold composition of the map $f$ with itself. Let $\alpha\in K$ and suppose that for every $n\geq 1$ the set $(f^n)^(\alpha)$ contains $d^n$ elements of the algebraic closure $\overline$. Then one can construct an infinite, regular, rooted $d$-ary tree $T(f)$ in the following way: the root of the tree is $\alpha$, and the nodes at distance $n$ from $\alpha$ are the elements of $(f^n)^(\alpha)$. A node $\beta$ at distance $n$ from $\alpha$ is connected with an edge to a node $\gamma$ at distance $n+1$ from $\alpha$ if and only if $f(\beta)=\gamma$.

The absolute Galois group $G_K$ acts on $T(f)$ via automorphisms, and the induced homomorphism $\rho_\colon G_K\to Aut(T(f))$ is continuous, and therefore is called the arboreal Galois representation attached to $f$ with basepoint $\alpha$.

Arboreal representations attached to rational functions can be seen as a wide generalization of Galois representations on Tate modules of abelian varieties.

Arboreal Galois representations attached to quadratic polynomials

The simplest non-trivial case is that of monic quadratic polynomials. Let $K$ be a field of characteristic not 2, let f=(x-a)^2+b\in K /math> and set the basepoint $\alpha=0$. The adjusted post-critical orbit of $f$ is the sequence defined by $c_1=-f(a)$ and $c_n= f^n(a)$ for every $n\geq 2$. A resultant argument shows that $(f^n)^(0)$ has $d^n$ elements for ever $n$ if and only if $c_n\neq 0$ for every $n$. In 1992, Stoll proved the following theorem:

:Theorem: the arboreal representation $\rho_$ is surjective if and only if the span of $\$ in the $\mathbb F_2$-vector space $K^*/(K^*)^2$ is $n$-dimensional for every $n\geq 1$.

The following are examples of polynomials that satisfy the conditions of Stoll's Theorem, and that therefore have surjective arboreal representations.

* For $K=\mathbb Q$, $f=x^2+a$, where $a\in \mathbb Z$ is such that either $a>0$ and $a\equiv 1,2\bmod 4$ or $a<0$, $a\equiv 0\bmod 4$ and $-a$ is not a square.

* Let $k$ be a field of characteristic not $2$ and $K=k(t)$ be the rational function field over $k$. Then f=x^2+t\in K /math> has surjective arboreal representation.

Higher degrees and Odoni's conjecture

In 1985 Odoni formulated the following conjecture.

:Conjecture: Let $K$ be a Hilbertian field of characteristic $0$, and let $n$ be a positive integer. Then there exists a polynomial f\in K /math> of degree $n$ such that $\rho_$ is surjective.

Although in this very general form the conjecture has been shown to be false by Dittmann and Kadets, there are several results when $K$ is a number field. Benedetto and Juul proved Odoni's conjecture for $K$ a number field and $n$ even, and also when both :\mathbb Q/math> and $n$ are odd, Looper independently proved Odoni's conjecture for $n$ prime and $K=\mathbb Q$.

Finite index conjecture

When $K$ is a global field and $f\in K(x)$ is a rational function of degree 2, the image of $\rho_$ is expected to be "large" in most cases. The following conjecture quantifies the previous statement, and it was formulated by Jones in 2013.

:Conjecture Let $K$ be a global field and $f\in K(x)$ a rational function of degree 2. Let $\gamma_1,\gamma_2\in \mathbb P^1_K$ be the critical points of $f$. Then ut(T(f)):Im(\rho_)\infty if and only if at least one of the following conditions hold:
# The map $f$ is post-critically finite, namely the orbits of $\gamma_1,\gamma_2$ are both finite.
# There exists $n\geq 1$ such that $f^n(\gamma_1)=f^n(\gamma_2)$.
# $0$ is a periodic point for $f$.
# There exist a Möbius transformation $m=\frac\in PGL_2(K)$ that fixes $0$ and is such that $m\circ f \circ m^=f$.

Jones' conjecture is considered to be a dynamical analogue of Serre's open image theorem.

One direction of Jones' conjecture is known to be true: if $f$ satisfies one of the above conditions, then ut(T(f)):Im(\rho_)\infty. In particular, when $f$ is post-critically finite then $Im(\rho_)$ is a topologically finitely generated closed subgroup of $Aut(T(f))$ for every $\alpha\in K$.

In the other direction, Juul et al. proved that if the abc conjecture holds for number fields, $K$ is a number field and f\in K /math> is a quadratic polynomial, then ut(T(f)):Im(\rho_)\infty if and only if $f$ is post-critically finite or not eventually stable. When f\in K /math> is a quadratic polynomial, conditions (2) and (4) in Jones' conjecture are never satisfied. Moreover, Jones and Levy conjectured that $f$ is eventually stable if and only if $0$ is not periodic for $f$.

Abelian arboreal representations

In 2020, Andrews and Petsche formulated the following conjecture.

:Conjecture Let $K$ be a number field, let f \in K /math> be a polynomial of degree $d\ge 2$ and let $\alpha\in K$. Then $Im(\rho_)$ is abelian if and only if there exists a root of unity $\zeta$ such that the pair $(f,\alpha)$ is conjugate over the maximal abelian extension $K^$ to $(x^d,\zeta)$ or to $(\pm T_d,\zeta+\zeta^)$, where $T_d$ is the Chebyshev polynomial of the first kind of degree $d$.

Two pairs $(f,\alpha),(g,\beta)$, where $f,g\in K(x)$ and $\alpha,\beta\in K$ are conjugate over a field extension $L/K$ if there exists a Möbius transformation $m=\frac\in PGL_2(L)$ such that $m\circ f \circ m^=g$ and $m(\alpha)=\beta$. Conjugacy is an equivalence relation. The Chebyshev polynomials the conjecture refers to are a normalized version, conjugate by the Möbius transformation $2x$ to make them monic.

It has been proven that Andrews and Petsche's conjecture holds true when $K=\mathbb Q$.

References

Further reading

*
Arboreal Galois Representations Over Finite Fields
*{{Cite journal , last=Li , first=Hua-Chieh , date=4 October 2019 , title=Arboreal Galois representation for a certain type of quadratic polynomials , url=https://link.springer.com/article/10.1007/s00013-019-01390-x , journal=Archiv der Mathematik , volume=114 , issue=3 , pages=265–269 , doi=10.1007/s00013-019-01390-x, url-access=subscription

Arithmetic dynamics
Galois theory