
The monoid of components is well-defined and commutative ([Seg23, Proposition 3.3.8], [Seg23,
Proposition 3.3.11]). Components of degree nare named this way because they correspond bijectively
to connected components of the Hurwitz space classifying marked G-covers of the projective line
branched at n· |ξ|points, among which n·ξ(γ)have their monodromy elements in each class γ∈D.
This connection is explained more carefully in [Seg23, Subsection 3.3.2]. The identity element of
the monoid CompP1(C)(G, D, ξ)is the orbit of the empty tuple, which corresponds to the connected
component containing only the trivial G-cover (with no branch points).
Definition 2.5. A nontrivial element of CompP1(C)(G, D, ξ)is a non-factorizable component2if it
does not equal any product of two nontrivial components.
A simple pigeonhole argument (carried out in [Seg23, Lemma 3.4.17]) shows that there are finitely
many non-factorizable components. Therefore, the monoid of components is a finitely generated
commutative graded monoid.
Remark 2.6.The non-factorizable components do not necessarily all have the same degree; see [Seg23,
Remark 3.4.20] for a counterexample.
2.2. The ring of components
2.2.1. Definition. We now define the ring of components:
Definition 2.7. The ring of components Ris the graded k-algebra k[CompP1(C)(G, D, ξ)] obtained
as the monoid ring (over k) of the monoid of components. The irrelevant ideal ϖis the (maximal)
ideal of Rgenerated by components of positive degree.
The ring Rof Definition 2.7 corresponds to RP1(C)(G, D, ξ)in the notation of [Seg23, Defini-
tion 3.4.12]. The properties mentioned in Subsection 2.1 imply that the ring Ris a commutative
graded k-algebra of finite type, generated by the non-factorizable components.
2.2.2. Variety of components. We now define our main object of study, the variety of components:
Definition 2.8. The variety of components is the set Spec Rof prime ideals p⊊R, equipped with
the Zariski topology. If Iis an ideal of R, we denote by V(I)the closed subset of Spec Rconsisting of
all prime ideals containing I.
2.2.3. Affine embedding. Assume that kis algebraically closed, and let Spm Rbe the subset of
Spec Rconsisting of closed points, i.e., of maximal ideals m⊊R. Then, the set Spm Rcan be
identified with the set of morphisms of k-algebras from Rto k(identifying a maximal ideal mwith
the projection R↠R/m≃k), or equivalently with the set of k-points of the scheme Spec R. Let Σ
be the finite set of non-factorizable components. Then, we can identify Spm Rwith a classical variety
by embedding it in kΣas follows: a point (xm)m∈Σbelongs to Spm Rif and only if the equality
xm1···xmu=xm′
1···xm′
vholds whenever the equality m1···mu=m′
1···m′
vholds in the monoid of
components. Note that, as the monoid of components is commutative and finitely generated, Dickson’s
lemma implies that it is presented by finitely many equalities of that type.
Remark 2.9.The ring Ris a graded k-algebra, and thus it may be more natural to consider the projec-
tive variety that it defines (i.e., the set of homogeneous ideals which are maximal among those properly
contained in ϖ) instead of the affine variety Spm R. However, since non-factorizable components need
not all have the same degree (cf. Remark 2.6), the space in which the variety naturally embeds is
a “weighted” projective space, namely the set of orbits of kΣ\ {0}under the action of k×for which
a scalar λ∈k×acts on a point zby multiplying its coordinate zm(associated to a non-factorizable
2Non-factorizable components are simply the irreducible elements of the monoid CompP1(C)(G, D, ξ), but we avoid
using the ambiguous term “irreducible component”.
4