---
title: Galois Closure of Rational Maps
url: https://www.emergentmind.com/topics/galois-closure-of-rational-maps
type: topic
---

# Galois Closure of Rational Maps

Searching arXiv for papers on Galois closures of rational maps and closely related frameworks.
arXiv.search query: "all:Galois closure rational maps"
I’m checking arXiv records for directly relevant papers.
The Galois closure of a rational map is the minimal Galois cover that dominates the map, or equivalently the normalization of the target in the Galois closure of the induced function-field extension. For a finite rational map \(\phi:\mathbb{P}^1\to\mathbb{P}^1\) over \(\mathbb{C}\), the finite extension \(\mathbb{C}(x)/\mathbb{C}(T)\) has a Galois closure \(\widetilde F/\mathbb{C}(T)\), and the associated smooth projective curve \(\widetilde X\) with \(\mathbb{C}(\widetilde X)=\widetilde F\) yields a Galois cover \(\widetilde X\to\mathbb{P}^1\) [1807.01937]. In higher dimensions the same construction is expressed by replacing a dominant rational map \(X\dashrightarrow Y\) with the finite cover obtained by normalizing \(Y\) in \(k(X)\), and then normalizing \(Y\) in the classical Galois closure of \(k(X)/k(Y)\) [1507.04433]. The topic therefore links branched coverings, monodromy, Hurwitz spaces, arithmetic dynamics, and birational geometry.

## 1. Function fields, normalization, and the cover-theoretic viewpoint

In the language of covers of \(\mathbb{P}^1\), a \(k\)-regular cover is “a non-constant finite morphism \(f:X\rightarrow \mathbb P^1_k\) with \(X\) a smooth \(k\)-curve,” and if \(k(X)/k(T)\) is Galois then \(f\) is a \(k\)-regular Galois cover. The group and branch point set of a \(k\)-regular cover are defined over \(\overline{k}\): the group is “the Galois group of its Galois closure” and the branch point set is the set of ramified points in \(\mathbb P^1(\overline{k})\) [1807.01937]. This formulation already identifies the Galois closure as the natural invariant attached to a rational map.

For a rational function \(X:\mathbb{P}^1\to\mathbb{P}^1\) of degree \(d\), the normalization viewpoint makes the Galois closure geometric. There exists a compact Riemann surface \(S_X\) and a holomorphic map \(\widetilde X:S_X\to\mathbb{P}^1\) such that \(\widetilde X\) is a Galois covering, \(X=\widetilde X\circ H\) for some holomorphic map \(H:\mathbb{P}^1\to S_X\), and \(\widetilde X\) has the lowest possible degree with these properties. Algebraically, \(\mathbb{C}(S_X)\) is the Galois closure of \(\mathbb{C}(z)/\mathbb{C}(X)\), and the genus \(g(S_X)\) is the genus of the Galois closure [1707.02948].

A concrete realization uses the \(d\)-fold fiber product of \(X\) with itself:
\[
\mathcal{L}=\{(z_1,\dots,z_d)\in(\mathbb{P}^1)^d\mid X(z_i)=X(z_j)\ \forall i,j\}.
\]
Removing components with coincident coordinates and desingularizing an irreducible component \(N\) produces a curve \(N'\) isomorphic to \(S_X\). In this way the Galois closure is encoded by the geometry of the fiber product and by the monodromy action on the sheets of the covering [1707.02948].

In the general setting of finite dominant rational maps between normal projective varieties, the same pattern persists. One replaces \(f:X\dashrightarrow Y\) by the finite morphism \(\tilde f:\tilde X\to Y\) obtained by normalizing \(Y\) in \(k(X)\). The Galois closure is then the normalization \(\hat X\to Y\) of \(Y\) in the Galois closure field of \(k(X)/k(Y)\); this is exactly the categorical Galois closure in the sense of the abstract theory of finite covers [1507.04433].

## 2. Rational pullback, parametricity, and the structure of closures over \(\mathbb{P}^1\)

A central operation on Galois closures of rational maps is rational pullback. Given a \(k\)-regular cover \(f:X\to\mathbb{P}^1_k\) and a non-constant rational map \(T_0\in k(U)\setminus k\), the pullback is defined by taking the fiber product \(X\times_{f,T_0}\mathbb{P}^1_k\), passing to a geometrically irreducible component, and normalizing. If \(f\) is Galois of group \(G\), then the pullback \(f_{T_0}:X_{T_0}\to\mathbb{P}^1_k\) “remains a \(k\)-regular Galois cover of group \(G\)” [1807.01937].

This makes the Galois closure of a composed rational map a pullback phenomenon. If \(f\) has affine equation \(P(T,Y)=0\), then a pullback along \(T_0(U)\) is given by substitution:
\[
P(T_0(U),Y)=0.
\]
Thus composition with rational maps preserves the global Galois group while modifying branch data and genus in a controlled way [1807.01937].

The structural theorem of Debes classifies when all Galois closures with fixed group can be generated from bounded-branch data by pullback. The finite subgroups of \(\mathrm{PGL}_2(\mathbb{C})\), namely the cyclic groups, dihedral groups, \(A_4\), \(S_4\), and \(A_5\), are exactly the finite groups that admit a regularly parametric cover: if \(G\subset\mathrm{PGL}_2(\mathbb{C})\), one well-chosen cover with at most \(3\) branch points suffices to produce every \(G\)-Galois cover by rational pullback; if \(G\not\subset\mathrm{PGL}_2(\mathbb{C})\), no bounded-branch family is regularly parametric [1807.01937]. A recurring simplification is therefore false outside the spherical case: for \(G\not\subset\mathrm{PGL}_2(\mathbb{C})\), letting the branch point number grow produces “truly new” Galois realizations, not mere pullbacks of a bounded family.

Pullback also obeys sharp monotonicity constraints. If \(r\) and \(g\) denote the branch point number and genus of a Galois cover \(f\), and \(r_{T_0}\), \(g_{T_0}\) those of its pullback, then
\[
r\le r_{T_0},\qquad g\le g_{T_0},
\]
and if \(g>1\) and \(T_0\) is not an isomorphism, then \(g<g_{T_0}\) [1807.01937]. A refined inequality expresses \(R_{T_0}\), the number of branch points after pullback, in terms of the degree \(n=\deg T_0\), the original inertia orders \(e_i\), and the ramification profile of \(T_0\):
\[
R_{T_0}\ge (R-4)n+4+\sum_{i=1}^R (e_i-2)a_i+\sum_{q\in U_{T_0,f}}(e(q\mid t_0)-1),
\]
with equality precisely under an explicit unramified-away-from-\(\{t_i\}\) condition [1807.01937].

These results connect directly to the Beckmann–Black-type lifting problem. The “Beckmann–Black regular lifting property” holds exactly for finite subgroups of \(\mathrm{PGL}_2(\mathbb{C})\); outside that class, one cannot dominate all Galois closures with a single universal cover via pullbacks [1807.01937].

## 3. Low-genus Galois closures in semiconjugacy and orbifold geometry

In complex dynamics, the Galois closure of a rational map often appears through semiconjugacy. If \(A,B,X:\mathbb{P}^1\to\mathbb{P}^1\) satisfy
\[
A\circ X=X\circ B
\]
and the solution is primitive, meaning \(\mathbb{C}(B,X)=\mathbb{C}(z)\), then the Galois closure of \(\mathbb{C}(z)/\mathbb{C}(X)\) has genus \(0\) or \(1\); equivalently, \(g(S_X)\le 1\) [1707.02948]. This theorem places strong geometric restrictions on rational maps that occur in primitive semiconjugacies.

One proof characterizes the low-genus condition by meromorphic parametrization: \(g(S_X)\le 1\) if and only if there exist \(d=\deg X\) distinct meromorphic functions \(\psi_1,\dots,\psi_d\) on \(\mathbb{C}\) such that
\[
X(\psi_i(z))=X(\psi_j(z))
\qquad (i\ne j).
\]
Another proof identifies \(S_X\) with an invariant algebraic curve in \((\mathbb{P}^1)^d\) for the product dynamics induced by \(B\); invariant irreducible curves of this type have genus at most \(1\) [1707.02948].

The theorem admits an orbifold reformulation. Under the same semiconjugacy and primitivity hypotheses, the associated orbifolds \(\mathcal{O}_1^X\) and \(\mathcal{O}_2^X\) satisfy
\[
\chi(\mathcal{O}_2^X)\ge 0,\qquad \chi(\mathcal{O}_1^X)\ge 0,
\]
and \(\chi(\mathcal{O}_2^X)\ge 0\) is equivalent to \(g(S_X)\le 1\) [1707.02948]. In this sense, the Galois closure is controlled by spherical and elliptic orbifold geometry.

Typical examples of rational maps with low-genus Galois closure are power maps, Chebyshev maps, and Lattès maps. In genus \(0\), the Galois closure comes from the sphere; in genus \(1\), it comes from an elliptic curve. A plausible implication is that primitive semiconjugacy forces the intermediate map \(X\) into the same restricted “low-genus universe” that also appears in Ritt-type classifications [1707.02948].

## 4. Iterated, arboreal, and dynatomic Galois closures

For iteration, the Galois closure is naturally studied level by level. Given a field \(k\), a rational function \(f\in k(x)\) of degree \(d\ge 2\), and a transcendental \(t\), one defines
\[
K_n=k(f^{-n}(t)),
\]
the splitting field of \(f^n(x)-t\) over \(k(t)\). Thus \(K_n/k(t)\) is the Galois closure of the iterate \(f^n\) in the generic parameter \(t\). For a point \(a\in\mathbb{P}^1(k)\), one likewise defines
\[
K_{a,n}=k(f^{-n}(a)),
\]
the splitting field of \(f^n(x)-a\) over \(k\) [2309.00840].

Passing to the inverse limit produces the generic and specialized iterated Galois groups
\[
G_\infty=\varprojlim_n \mathrm{Gal}(K_n/k(t)),\qquad
G_{a,\infty}=\varprojlim_n \mathrm{Gal}(K_{a,n}/k),
\]
which act on the rooted preimage tree \(T_d\). In the post-critically finite (PCF) case, the ramification set of \(K_\infty/k(t)\) is finite, and this finiteness is a key structural input [2309.00840].

Under a PCF and \(p\)-group hypothesis at level \(1\), finite-level maximality implies global maximality: there exists \(m\ge 1\) such that
\[
G_m=G_{a,m}\quad\Longrightarrow\quad G_\infty=G_{a,\infty}.
\]
For PCF quadratic rational functions over a number field, one has \(G_\infty=G_{a,\infty}\) for all \(a\) outside a thin set [2309.00840]. In the family \(f(x)=x^{p^m}+c\), the paper gives an explicit criterion at the level of the forward orbit size \(N\) of the critical point \(0\):
\[
G_{a,\infty}=G_\infty
\iff
\bigl|\mathrm{Gal}(k'\cdot K_{a,N}/k)\bigr|=|G_N|\,[k':k_1].
\]
This places the Galois closure of iterates inside an explicit finite-level control problem [2309.00840].

A different global constraint concerns abelianity. For a number field \(K\), a PCF rational map \(f(x)\in K(x)\), and a non-preperiodic point \(\alpha\in\mathbb{P}^1(K)\), the arboreal Galois extension
\[
K_\infty(f,\alpha)=\bigcup_{n\ge 0} K(f^{-n}(\alpha))
\]
has non-abelian Galois group:
\[
\mathrm{Gal}(K_\infty(f,\alpha)/K)\ \text{is not abelian}.
\]
Thus the global Galois closure of the backward dynamical system is never abelian in the PCF plus non-preperiodic setting [2407.17415].

Periodic-point fields provide a parallel construction. For the dynatomic polynomial \(\Phi_{n,\phi}(x)\), if the discriminant is nonzero then
\[
G_{n,\phi}=\mathrm{Gal}(\Phi_{n,\phi}),
\]
so the dynatomic Galois group is exactly the Galois group of the splitting field of \(\Phi_{n,\phi}\) [2303.05632]. For quadratic rational maps with a critical point of period \(2\), the generic third dynatomic Galois group in the trivial-automorphism case is
\[
C_3\wr S_2,
\]
and the generic fourth dynatomic Galois group is
\[
C_4\wr S_3;
\]
with nontrivial automorphism group, the generic fourth dynatomic Galois group becomes a subgroup
\[
V\simeq C_4\times S_3
\subset
C_4\wr S_3
\]
[2303.05632]. This suggests that iterated and periodic-point Galois closures are typically “as large as possible,” with exceptional smaller groups occurring along thin parameter sets.

## 5. Explicit monodromy classifications and geometric realizations

In degree \(5\), the Galois closure can be described explicitly from ramification data. For a \(5\)-fold ramified covering \(f:X\to Y\) of compact Riemann surfaces, every possible Galois closure is determined by the ramification divisor of \(f\), and the possible transitive monodromy groups are exactly
\[
C_5,\quad D_5,\quad Aff,\quad A_5,\quad S_5.
\]
The paper classifies which groups occur from which ramification types, both for \(g_Y\ge 1\) and for \(Y=\mathbb{P}^1\), and then describes the group algebra decomposition of the Jacobian of the Galois closure in terms of Jacobians and Prym varieties of intermediate coverings [2211.11840].

The degree-\(5\) case shows how local branching controls the global closure. The geometric signature of the Galois closure is determined by the cycle types in the original covering, the subgroup lattice of the monodromy group organizes the intermediate covers, and the induced action on \(\mathrm{Jac}(\hat X)\) decomposes into explicit abelian subvarieties. The dimensions and induced polarizations of these factors are computed directly from the ramification data [2211.11840].

A different explicit realization appears for degree \(3\) covers of projective space. Let \(V\) be a nonsingular projective variety of dimension \(n\) with a very ample divisor \(D\) such that
\[
D^n=6,\qquad \dim H^0(V,\mathcal O(D))=n+3.
\]
Then \((V,D)\) defines a \(D_6\)-Galois embedding if and only if \(V\) is the Galois closure variety of a smooth cubic in \(\mathbb{P}^{n+1}\) with respect to a suitable projection center, and the pullback of a hyperplane of \(\mathbb{P}^n\) is linearly equivalent to \(D\) [1212.6799]. The paper denotes by \(D_{2m}\) the dihedral group of order \(2m\); in this notation, \(D_6\) is the dihedral group of order \(12\).

For a good projection point \(P\), the Galois closure variety \(A_P\) of the non-Galois triple cover \(A\to\mathbb{P}^n\) is a smooth \((2,3)\)-complete intersection in \(\mathbb{P}^{n+2}\), cut out by
\[
F(\mathbf X)=0,\qquad X_{n+2}^2=\phi_2(X_0,\dots,X_n),
\]
and the Galois group is \(S_3\simeq D_6\) [1212.6799]. In this setting the Galois closure is not merely an abstract field extension: it is a concrete projective variety carrying a distinguished Galois embedding.

## 6. Categorical closure and higher-dimensional birational constraints

A unified abstract treatment replaces special constructions by a categorical notion of finite cover. In this framework, a cover \(f\) is Galois when the automorphism set over the codomain has size equal to the degree,
\[
|(f)|=\deg f,
\]
and a Galois closure of \(f\) is a minimal Galois cover dominating \(f\). Equivalently, it is a least cover \(g\) with
\[
|(g,f)|=\deg f.
\]
Two algorithms, an iterative algorithm \(\mathds I\) and a divide-and-conquer algorithm \(\mathds R\), produce Galois closures in all the standard settings covered by the axioms [1507.04433].

These algorithms apply to finite separable field extensions, finite étale covers, finite covers of normal varieties, and therefore to finite dominant rational maps after normalization. They also yield degree bounds:
\[
\deg g\le (\deg f)!,
\]
and under the additional axiom (G5),
\[
\deg g\mid (\deg f)!.
\]
For finite covers of normal projective varieties, pullbacks are obtained from fiber products followed by normalization, and the categorical Galois closure of a rational map is the normalization of the target in the classical Galois closure field [1507.04433].

In higher-dimensional birational geometry, the same notion becomes restrictive. For a generically finite dominant rational map \(f:Y\dashrightarrow X\) between smooth projective varieties, the paper on Calabi–Yau and hyperkähler geometry defines \(f\) to be Galois exactly when the field extension \(C(X)\hookrightarrow C(Y)\) is Galois, and identifies the monodromy group with the Galois group of the Galois closure of \(C(Y)/C(X)\) [2507.14359]. In this setting, branch divisors and Galois closures reflect Hodge-theoretic and birational constraints.

For a hyper-Kähler manifold \(X\) with
\[
b_2(X)=23,\qquad \rho_X=1,
\]
any Calabi–Yau rational cover \(\phi:Y\dashrightarrow X\) satisfies
\[
\Bir(\phi)=\{\mathrm{id}\},
\]
so if \(\phi\) is Galois, it is an isomorphism [2507.14359]. In particular, a very general Fano variety of lines \(F_1(W)\) on a cubic fourfold admits no nontrivial Calabi–Yau Galois covers. The same work proves that the Voisin map on \(F_1(W)\) is not a Calabi–Yau Galois-like cover for very general \(W\), and not an abelian Galois-like cover for any smooth cubic fourfold [2507.14359]. A further structural statement is that if \(f:Y\dashrightarrow X\) is a Calabi–Yau Galois cover of a hyper-Kähler manifold and the Galois group acts by automorphisms on \(Y\), then the branch divisor \(\Bran(f)\) is \(q_X\)-exceptional [2507.14359].

Taken together, these viewpoints show that the Galois closure of a rational map is not a single construction but a unifying object. On \(\mathbb{P}^1\), it is organized by pullback, Hurwitz spaces, and the dichotomy between finite subgroups of \(\mathrm{PGL}_2(\mathbb{C})\) and all other groups. In semiconjugacy, it is constrained by genus \(0\) and \(1\) phenomena. In arithmetic dynamics, it becomes an inverse system of splitting fields of iterates and dynatomic polynomials. In explicit projective geometry, it can be realized as a complete intersection or classified from ramification data. In higher dimensions, especially for Calabi–Yau and hyper-Kähler manifolds, the Galois closure is tightly controlled by Hodge theory, monodromy, and birational rigidity.

Source: https://www.emergentmind.com/topics/galois-closure-of-rational-maps