Professional Documents
Culture Documents
Ament 2019
Ament 2019
2019; aop
Abstract: Let (X, 0) ⊂ (ℂn , 0) be an irreducible weighted homogeneous singularity curve and let f : (X, 0) →
(ℂ2 , 0) be a finite map germ, one-to-one and weighted homogeneous with the same weights of (X, 0). We show
that Ae -codim(X, f) = μ I (f), where the Ae -codimension Ae -codim(X, f) is the minimum number of parameters
in a versal deformation and μ I (f) is the image Milnor number, i.e. the number of vanishing cycles in the image
of a stabilization of f .
1 Introduction
Let α : (ℂ, S) → (ℂ2 , 0) be a map germ of finite Ae -codimension. Mond shows in [8] that the image Milnor
number μ I (α) of α is determined by the number of branches r(X, 0) of the curve and the number δ(X, 0) of
nodes appearing in a stable perturbation of α:
μ I (α) = δ(X, 0) − r(X, 0) + 1.
Mond also proves in [8] an inequality between the image Milnor number and the Ae -codimension:
Theorem 1.1 ([8]). Let α : (ℂ, S) → (ℂ2 , 0) be a map germ of finite Ae -codimension. Then
Ae - codim(α) ≤ μ I (α),
Inspired by this inequality, the first and second authors consider in [1] map germs f : (X, 0) → (ℂ2 , 0), where
(X, 0) is a plane curve. They define the image Milnor number of f and obtain a similar inequality:
Theorem 1.2 ([1]). Let (X, 0) be a plane curve and let f : (X, 0) → (ℂ2 , 0) be a finite map germ of degree 1 onto
its image (Y, 0). Then
Ae - codim(X, f) ≤ μ I (f),
with equality if and only if (Y, 0) is weighted homogeneous.
D. A. H. Ament, Departamento de Matemática, Universidade Federal de São Carlos, Caixa Postal 676, 13560-905, São Carlos, SP,
Brazil, email: daianehenrique@dm.ufscar.br
J. J. Nuño-Ballesteros, Departament de Matemàtiques, Universitat de València, Campus de Burjassot, 46100 Burjassot Spain,
email: Juan.Nuno@uv.es
*Corresponding author: J. N. Tomazella, Departamento de Matemática, Universidade Federal de São Carlos, Caixa Postal 676,
13560-905, São Carlos, SP, Brazil, email: tomazella@dm.ufscar.br
In this work, we consider map germs f : (X, 0) → (ℂ2 , 0), where (X, 0) is an isolated complete intersection
singularity (ICIS) of dimension one, and the Ae -codimension Ae -codim(X, f) in the sense of [9]. We define
the image Milnor number in this case as the number of vanishing cycles in the image of a stable perturbation
f s : X s → B ε , where B ε is a ball of radius ε centered at the origin in ℂ2 . Here, stable means that X s is smooth
and f s is stable in the usual sense.
We show that Ae -codim(X, f) = μ I (f), if (X, 0) is irreducible and both (X, 0) and f are weighted homoge-
neous (with the same weights). Furthermore, if (X, 0) is parametrized by a map α : (ℂ, 0) → (ℂn , 0), then we
have also the following equalities:
1
μ I (f) = δ(X, 0) + μ I (f ∘ α) and Ae - codim(f) = Ae - codim(f ∘ α) − Ae - codim(α).
n−1
Theorem 2.1. Let (X, 0) be an ICIS and assume that f : (X, 0) → (ℂp , 0) has finite singularity type. Then the
minimal number of parameters in a versal unfolding of f is equal to the number
where τ(X, 0) is the Tjurina number of (X, 0), that is, the minimal number of parameters in a versal deformation
of (X, 0).
As a corollary of this theorem, we have the following consequences. The second one is a generalization
of the Mather–Gaffney criterion; (see [14].
Corollary 2.2. Let (X, 0) be an ICIS and assume that f : (X, 0) → (ℂp , 0) has finite singularity type.
(1) f is stable if and only if X is smooth and f is stable in the usual sense.
(2) f is A -finite if and only if f has isolated instability (i.e. there is a representative f : X → B ε such that for
any y ∈ B ε \ {0} the multigerm of f at f −1 (y) ∩ S is stable).
Another consequence of the theorem is the existence of stabilizations, at least in the range of Mather’s
nice dimensions. Given an analytic map germ f : (X, 0) → (ℂp , 0), a stabilization is a 1-parameter unfolding
F : (X, 0) → (ℂ × ℂp ) with the property that for all s ≠ 0 small enough, the map f s : X s → B ε is stable, where
B ε is a ball of radius ε in ℂp . By Theorem 2.1, if f : (X, 0) → (ℂp , 0) is A -finite, a stabilization of f exists if
(r, p) are nice dimensions in Mather’s sense (where r = dim(X, 0)).
We consider germs f : (X, 0) → (ℂ2 , 0), where (X, 0) ⊂ (ℂn , 0) is a curve ICIS. Observe that by Corol-
lary 2.2, f is stable if and only if X is smooth and f is an immersion with only transverse double points, called
nodes. As a consequence, f is A -finite if and only if f is finite and has degree 1 onto its image. The degree of
f is denoted by deg(f) and is defined as the number of inverse images of a generic value.
In this case, the value w i is the weight of the variable x i and the value d i is the weighted degree of φ i .
A germ of an analytic variety (X, 0) ⊂ (ℂn , 0) is weighted homogeneous if it is defined by the zeros of
a weighted homogeneous map germ φ : (ℂn , 0) → (ℂr , 0). A map germ f : (X, 0) → (ℂp , 0) is weighted
homogeneous if it is the restriction of a weighted homogeneous map germ f ̃ : (ℂn , 0) → (ℂp , 0), and we say
that f is consistent with (X, 0) if f ̃ is weighted homogeneous with the same weights of (X, 0).
Greuel and Hamm obtain in [4] a formula for the Milnor number of an arbitrary weighted homogeneous
ICIS. We use a result from [11]: if (X, 0) is a weighted homogeneous curve ICIS of type (w1 , . . . , w n ; d1 ,
. . . , d n−1 ), then the Milnor number satisfies
d1 ⋅ ⋅ ⋅ d n−1 (d1 + ⋅ ⋅ ⋅ + d n−1 − w1 − ⋅ ⋅ ⋅ − w n )
μ(X, 0) = + 1.
w1 ⋅ ⋅ ⋅ w n
Moreover, if (X, 0) is irreducible, then it admits a parametrization by a monomial curve α(t) = (α1 t w1 , . . . ,
α n t w n ), for some α1 , . . . , α n ∈ ℂ. Let ψ : (X, 0) → ℂ be any finite function consistent with (X, 0). Then its
degree deg(ψ) is equal to deg(ψ ∘ α) = d, the weighted degree of ψ. By the Bezout theorem,
OX,0 d1 ⋅ ⋅ ⋅ d n−1 d
deg(ψ) = dimℂ = = d,
⟨ψ⟩ w1 ⋅ ⋅ ⋅ w n
hence d1 ⋅ ⋅ ⋅ d n−1 = w1 ⋅ ⋅ ⋅ w n . Thus the Milnor number in this case is given by
μ(X, 0) = d1 + ⋅ ⋅ ⋅ + d n−1 − w1 − ⋅ ⋅ ⋅ − w n + 1.
Wahl in [13] computes the generators of Θ X,0 when (X, 0) is a weighted homogeneous ICIS. In particular,
if we suppose (X, 0) = h−1 (0) a curve, where h = (h1 , . . . , h n−1 ), then Θ X,0 is generated by vector fields h i ∂x∂ j
with i = 1, . . . , n − 1 and j = 1, . . . , n, the Euler field ε = (w1 x1 , . . . , w n x n ) and H = (ξ1 , . . . , ξ n ), where
∂ ∂
∂x1 ⋅⋅⋅ ∂x n
∂h ∂h1
1 ⋅⋅⋅
∂x1 ∂x n
H = . .. ..
.. . .
∂h n−1 ∂h n−1
⋅⋅⋅
∂x1 ∂x n
∂h ̂ ∂h1 ∂h1 ∂h1 ∂h1 ̂1
∂h
∂x 1 ⋅⋅⋅ ∂x ⋅⋅⋅
1 ∂x2 ∂x n 1 ∂x n−1 ∂x n
= (−1)1+1 ... .. .. .. ∂ + ⋅ ⋅ ⋅ + (−1)n+1 .. .. .. .. ∂
. . . ∂x1 .
. . . ∂x n
∂h n−1 ∂h n−1 ∂h n−1
∂h n−1 ∂h n−1 ∂h n−1
∂x1 ∂x2 ⋅⋅⋅ ∂x n
∂x1 ⋅⋅⋅ ∂x n−1 ∂x n
∂ ∂
= ξ1 + ⋅ ⋅ ⋅ + ξn ,
∂x1 ∂x n
The local ring of (X, 0) is OX,0 = ℂ{t w1 , . . . , t w n }, which is a subring of O1 . We denote by Γ X the associated
semigroup. Since (X, 0) is irreducible, Γ X has a conductor c, which satisfies that c − 1 ∉ Γ X and if c ≤ n ∈ ℕ,
then n ∈ Γ X . Since (X, 0) is an ICIS, Γ X is symmetric, i.e. #ℕ \ Γ X = c/2. It follows from [3] that the semigroup
Γ X is symmetric if and only if for any x < c, either x ∈ Γ X or c − 1 − x ∈ Γ X .
The main invariant of (X, 0) is the delta invariant, which is in this case
ℂ{t}
δ X = dimℂ = #ℕ \ Γ X .
ℂ{t w1 , . . . , t w n }
Since (X, 0) is irreducible, the Milnor number μ(X, 0), which to simplify the notation we denote by μ X , is
μ X = 2δ X = c, by Milnor’s formula. For more details, see for instance [5].
Let f : (X, 0) → (ℂ2 , 0) be a finite map germ of degree 1 onto its image (Y, 0) such that f is consistent with
(X, 0). Recall that f is the restriction of a weighted homogeneous map germ f ̃ = (f1 , f2 ) : (ℂn , 0) → (ℂ2 , 0).
We denote by l1 , l2 the weighted degrees of f1 , f2 , respectively, which must be coprime. Then (Y, 0) is also
a weighted homogeneous irreducible plane curve with weights (l1 , l2 ) and admits a parametrization of the
form
f(α(t)) = (β1 t l1 , β2 t l2 ), β1 , β2 ∈ ℂ \ {0}.
We have OY,0 = ℂ{t l1 , t l2 } ⊂ OX,0 , Γ Y = ⟨l1 , l2 ⟩ ⊂ Γ X and δ Y = (l1 − 1)(l2 − 1)/2. Finally, we assume that (Y, 0)
l l
has defining equation g(u, v) = 0, where g ∈ ℂ{u, v} is weighted homogeneous. Thus g(u, v) = β21 u l2 − β12 v l1
is weighted homogeneous of type (l1 , l2 ; l1 l2 ). We can assume without loss of generality that l1 ≤ l2 .
We consider f = i ∘ f ̄, where f ̄ : (X, 0) → (Y, 0) is the restriction of f and i : (Y, 0) → (ℂ2 , 0) is the
inclusion map. We have
∂g ∂g
J g OX,0 = {a( ∘ f ) + b( ∘ f ); a, b ∈ OX,0 }
∂u ∂v
and
∂g ∂g
J g OY,0 = {(a ∘ f ̄)(( ∘ i) ∘ f ̄) + (b ∘ f ̄)(( ∘ i) ∘ f ̄); a, b ∈ OY,0 },
∂u ∂v
where J g is the Jacobian ideal of g, thus J g OY,0 ⊂ J g OX,0 .
Lemma 4.1. Let E : Θ(f) → J g OX,0 be the evaluation map given by E(ξ) = ξ(g). Then E is an epimorphism,
tf(Θ X,0 ) ⊂ ker(E) and E(wf(Θ2 )) ⊂ J g OY,0 . Consequently, the induced maps
∂ ∂
ξ = ζ ∘ f = (ã ∘ f) + (b̃ ∘ f )
∂u ∂v
and
∂g ∂g ∂g ∂g
ξ(g) = (ã ∘ f) ∘ f + (b̃ ∘ f ) ∘ f = ((ã ∘ i) ∘ f ̄)(( ∘ i) ∘ f ̄) + ((b̃ ∘ i) ∘ f ̄)(( ∘ i) ∘ f ̄),
∂u ∂v ∂u ∂v
with ã ∘ i, b̃ ∘ i ∈ OY,0 . Hence ξ(g) ∈ J g OY,0 and E(wf(Θ2 )) ⊂ J g OY,0 .
The conclusion follows by standard arguments of algebra; see for instance [7, p. 172]. 2
∂g l l −1 ∂g l l −1
∘ f = l2 β21 β12 t l1 (l2 −1) and ∘ f = −l1 β12 β21 t l2 (l1 −1) .
∂u ∂v
We also have
t d1 −w1 ∂h1 (α , . . . , α ) t d1 −w n ∂h
∂x n (α 1 , . . . , α n )
1
∂x1 1 n ⋅⋅⋅
.. .. ..
. . .
J j = .
t d n−1 −w1 ∂h n−1 (α , . . . , α ) ⋅⋅⋅ t d n−1 −w n ∂h n−1
(α , . . . , α )
1 n 1 n
l −w ∂f∂xj 1 ∂x n
∂f
t j 1 (α 1 , . . . , α n ) ⋅⋅⋅ t l j −w n ∂x n (α1 , . . . , α n )
j
∂x1
∂h i ∂f
We write ∂x k
(α1 , . . . , α n ) = γ i,k for i = 1, . . . , n − 1, k = 1, . . . , n and ∂xjk (α1 , . . . , α n ) = ζ j,k for j = 1, 2 and
k = 1, . . . , n.
Since h1 , . . . , h n−1 , f1 , f2 are weighted homogeneous with weights w1 , . . . , w n , they satisfy
∂h i ∂h i ∂f j ∂f j
d i h i = w1 x1 + ⋅ ⋅ ⋅ + wn xn , l j f j = w1 x1 + ⋅ ⋅ ⋅ + wn xn for i = 1, . . . , n − 1, j = 1, 2.
∂x1 ∂x n ∂x1 ∂x n
Thus
Thus
t d1 −w2 γ ⋅⋅⋅ t d1 −w n γ1,n
1,2
l β
J j = (−1)1+n
j j l j −w1
t .. .. ..
. . .
w1 α1
t d n−1 −w2 γ n−1,2 ⋅ ⋅ ⋅ t d n−1 −w n γ n−1,n
l j β j
= (−1)1+n t l j −w1 +d1 +⋅⋅⋅+d n−1 −w2 −⋅⋅⋅−w n B = l j β j t l j −w1 +d1 +⋅⋅⋅+d n−1 −w2 −⋅⋅⋅−w n A,
w1 α1
J1 = l1 β1 At l1 +μ X −1 and J2 = l2 β2 At l2 +μ X −1 .
Therefore
l −1 l −1
∂g l l −1 β 1 β12 μ Y −μ X
∘ f = l2 β21 β12 t l1 (l2 −1) = 2 t l2 β2 At l2 +μ X −1 = Cλ f J2
∂u A
l −1 l −1
∂g l l −1 β 2 β21 μ Y −μ X
and ∘ f = −l1 β12 β21 t l2 (l1 −1) = − 1 t l1 β1 At l1 +μ X −1 = −Cλ f J1 . 2
∂v A
Using the coordinates in the source and target, we can identify Θ(f) with O2X,0 = OX,0 ⊕ OX,0 . The module
Θ X,0 can be seen as a submodule of OnX,0 . With these identifications, the morphism tf : Θ X,0 → Θ(f) is the
restriction of the map tf : OnX,0 → O2X,0 , whose matrix in the canonical basis is the Jacobian matrix of f .
The following result describes the elements of ker(ẽv).
∂g ∂g l l −1 l l −1
ẽv([(a, b)]) = a ∘f +b ∘ f = al2 β21 β12 t l1 l2 −l1 − bl1 β12 β21 t l1 l2 −l2 .
∂u ∂v
Thus [(a, b)] ∈ ker(ẽv) if and only if (a, b) = c(l1 β1 t s1 , l2 β2 t s2 ) such that l1 l2 − l1 + s1 = l1 l2 − l2 + s2 for some
c ∈ OX,0 . It follows that ker(ẽv) is generated by monomial pairs of the form (l1 β1 t r , l2 β2 t r+k ), for some r ∈ ℕ
such that (t r , t r+k ) ∈ O2X,0 .
Since OX,0 = ℂ{t w1 , . . . , t w n }, OY,0 = ℂ{t l1 , t l2 } ⊂ OX,0 and f ̃ = (f1 , f2 ) is weighted homogeneous of type
(w1 , . . . , w n ; l1 , l2 ), we have
df ∘ ε = (l1 f1 , l2 f2 ) = (l1 β1 t l1 , l2 β2 t l1 +k ).
Recalling that
J1 = l1 β1 At l1 +μ X −1 and J2 = l2 β2 At l2 +μ X −1 ,
we obtain
df ∘ H = (−1)n+1 (l1 β1 At l1 +μ X −1 , l2 β2 At l1 +μ X −1+k ). 2
Theorem 4.4. Let s be the number of elements of the form (t r , t r+k ) in O2X,0 such that 0 ≤ r < μ X . Then
OX,0
dimℂ ker(ev) = l1 − δ X + s − 1 and dimℂ = l1 + δ X + s − 1.
⟨J1 , J2 ⟩
Proof. By Lemma 4.1 and Lemma 4.3,
We can identify the elements of ker(ev) with the elements of Γ k \ Γ Θ . Therefore, we compute the number of
elements in Γ k \ Γ Θ .
We observe that Γ k is equal to
and Γ Θ is
δ elements
X
⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞
{(l1 + p, l1 + p + k) : 0 ≤ p < μ X , p ∈ Γ X } ∪ ⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟
{(l1 + μ X − 1, l1 + μ X − 1 + k), (l1 + μ X , l1 + μ X + k), ⋅ ⋅ ⋅ } .
all elements
Then, we describe Γ k \ Γ Θ as
s elements l −1 elements
1
⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞ ⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞
{(r, r + k) : 0 ≤ r < μ X } ∪ {(μ X , μ X + k), . . . , (l1 + μ X − 2, l1 + μ X − 2 + k)}
\ {(l 1 + p, l 1 + p + k) : 0 ≤ p < μ X , p ∈ Γ X }
⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟
δ X elements
Therefore
dimℂ ker(ev) = s + l1 − 1 − δ X = l1 − δ X + s − 1.
of Γ X . We can identify the elements of OX /⟨J1 , J2 ⟩ with the elements of Γ X \ (Γ J1 ∪ Γ J2 ). Therefore, we compute
the number of elements in Γ X \ (Γ J1 ∪ Γ J2 ).
We need to determine the common elements in Γ J1 and Γ J2 . All elements after l1 + μ X − 1 + k + μ X are
common elements in Γ J1 and Γ J2 . We consider the subsets
Given r and r + k in Γ X such that 0 ≤ r < μ X , the elements common in A and B are the elements of the form
l1 + μ X − 1 + (r + k) = l1 + μ X − 1 + k + r.
Since s is the number of elements r ∈ Γ X with 0 ≤ r < μ X and r + k ∈ Γ X , the set A ∩ B contains s common
elements. Since A contains δ X +k+1 elements and B contains δ X elements, the union A∪ B contains μ X +k+1−s
elements.
Thus Γ X \ (Γ J1 ∪ Γ J2 ) is equivalent to
δ elements
X l1 +μ X +k elements
⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞ ⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞
{j ∈ Γ X ; 0 ≤ j < μ X } ∪ {μ X , . . . , l1 + μ X + μ X + k − 1}
\ {l 1 + μ X − 1 + i : 0 ≤ i ≤ μ X } ∪ {l 1 + μ X − 1 + k + j : 0 ≤ j < μ X } .
⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟
μ X +k+1−s elements
Therefore
OX,0
dimℂ = δ X + l1 + μ X + k − (μ X + k + 1 − s) = l1 + δ X + s − 1. 2
⟨J1 , J2 ⟩
Corollary 4.5. Let (X, 0) ⊂ (ℂn , 0) be an irreducible curve ICIS, weighted homogeneous and let f : (X, 0) →
(ℂ2 , 0) be a finite map germ of degree 1 onto its image (Y, 0) such that f is consistent with (X, 0). Then
OX,0 J g OX,0
Ae - codim(f) = dimℂ − μ X + dimℂ .
⟨J1 , J2 ⟩ J g OY,0
Proof. By Lemma 4.1,
J g OX,0
Ae - codim(f) = dimℂ ker(ev) + dimℂ .
J g OY,0
By Theorem 4.4,
OX,0
dimℂ − dimℂ ker(ev) = l1 + δ X + s − 1 − (l1 − δ X + s − 1) = μ X .
⟨J1 , J2 ⟩
Therefore
OX,0 J g OX,0
Ae - codim(f) = dimℂ − μ X + dimℂ . 2
⟨J1 , J2 ⟩ J g OY,0
Definition 5.1. The image Milnor number μ I (f) is the number of 1-spheres in Y s .
Analogous to the case where (X, 0) is a plane curve, μ I (f) is well-defined, that is, it is independent of the
stabilization and of the representative. To see this we consider the delta invariant of f which was introduced
in [12] for maps of degree 1 between curves.
Definition 5.2 ([12]). Let f : (X, 0) → (Y, 0) be a holomorphic map of degree 1 between curves (X, 0) and
(Y, 0). The delta invariant of f is
OX,0
δ(f) = dimℂ ∗ .
f OY,0
It follows from Buchweitz and Greuel [2] that
where S(Y s ) is the singular locus of Y s . But Y s has only node singularities, so the sum ∑y∈S(Y s ) μ(Y s , y) is the
number of nodes of Y s . By Mond and Pellikaan [10], the number of nodes of Y s is equal to
O2
dimℂ ,
F1 (f)
where F1 (f) is the first Fitting ideal of a presentation of OX,0 as an O2 -module via f ∗ . This number is equal to
δ(f) when (Y, 0) is a plane curve (see [12]), hence μ I (f) = μ(Y, 0) − δ(f).
Moreover the delta invariant satisfies δ(Y, 0) = δ(X, 0)+ δ(f) when (X, 0) and (Y, 0) are irreducible curves.
Therefore, when (X, 0) and (Y, 0) are irreducible curves then μ(Y, 0) − μ(X, 0) = 2δ(f), see [12], thus
μ(Y, 0) + μ(X, 0)
μ I (f) = μ(Y, 0) − δ(f) = μ(X, 0) + δ(f) = .
2
Now we consider again (X, 0) ⊂ (ℂn , 0), an irreducible curve ICIS and weighted homogeneous of type
(w1 , . . . , w n ; d1 , . . . , d n−1 ), and a finite map germ f : (X, 0) → (ℂ2 , 0) of degree 1 onto its image (Y, 0), such
that f is consistent with (X, 0). We take C the ideal generated by λ f in OX,0 . Then we have the following result.
Proposition 5.3. Let (X, 0) ⊂ (ℂn , 0) be an irreducible curve ICIS and weighted homogeneous of type (w1 , . . . ,
w n ; d1 , . . . , d n−1 ) and let f : (X, 0) → (ℂ2 , 0) be a finite map germ of degree 1 onto its image (Y, 0) and
consistent with (X, 0). Then
OX,0 OY,0
dimℂ = 2δ(f) and dimℂ = δ(f).
C C
OX,0 C
Furthermore, ⟨J1 ,J2 ⟩ is isomorphic to J g OX,0 .
Proof. By Lemma 4.2, λ f = t μ Y −μ X = t2δ(f) . Therefore, the weighted degree of λ f with the weights w1 , . . . , w n
is 2δ(f). Denote by (h1 , . . . , h n−1 ) : (ℂn , 0) → (ℂn−1 , 0) the map that generates (X, 0). By Bezout’s theorem
we obtain
OX,0 OX,0 O2 d1 ⋅ ⋅ ⋅ d n−1 2δ(f)
dimℂ = dimℂ = dimℂ = = 2δ(f).
C ⟨λ f ⟩ ⟨h1 , . . . , h n−1 , λ f ⟩ w1 ⋅ ⋅ ⋅ w n
To obtain the second equality, we first show that C ⊂ OY,0 . We recall that C is the ideal generated by
λ f = t2δ(f) in OX,0 . Then, we need to show that σt2δ(f) ∈ OY,0 for all σ ∈ OX,0 . To see this, we consider the
numerical semigroups Γ X and Γ Y associated the curves (X, 0) and (Y, 0) respectively. Recall that Γ X and Γ Y
are symmetric.
We show that if 2δ(f) ∈ Γ Y , then t2δ(f) ∈ OY,0 . In fact, since Γ Y is symmetric we have 2δ(f) ∈ Γ Y or
μ Y − 1 − 2δ(f) ∈ Γ Y . If μ Y − 1 − 2δ(f) ∈ Γ Y , then
μ Y − 1 − 2δ(f) = μ Y − 1 − μ Y + μ X = μ X − 1 ∈ Γ Y ⊂ Γ X .
μY − 1 − μY + μX − a = μX − 1 − a ∈ ΓY ⊂ ΓX .
To complete the proof, we observe that multiplication by λ f gives an isomorphism φ : OX,0 → C. More-
over, φ(⟨J1 , J2 ⟩) = J g OX,0 , hence φ induces an isomorphism
OX,0 C
→ . 2
⟨J1 , J2 ⟩ J g OX,0
Theorem 5.4. Let (X, 0) ⊂ (ℂn , 0) be an irreducible weighted homogeneous curve ICIS and let f : (X, 0) →
(ℂ2 , 0) be a finite map germ of degree 1 onto its image (Y, 0) and consistent with (X, 0). Then
Ae - codim(X, f) = μ I (f).
C OY,0 OY,0
Ae - codim(f) = dimℂ − μ X = dimℂ − dimℂ − μ X = μ Y − δ(f) − μ X = δ(f).
J g OY,0 J g OY,0 C
To prove the second part, we observe that f ∘ α : (ℂ, 0) → (Y, 0) is a parametrization of (Y, 0) ⊂ (ℂ2 , 0),
hence by [8] we have
Ae - codim(f ∘ α) = μ I (f ∘ α) = δ(Y, 0).
By [12], δ(Y, 0) = δ(X, 0) + δ(f), and by the definition of image Milnor number we obtain
μ I (f) = μ(X, 0) + δ(f) = μ(X, 0) + δ(Y, 0) − δ(X, 0) = δ(X, 0) + δ(Y, 0) = δ(X, 0) + μ I (f ∘ α).
We also have
Ae - codim(f) = δ(f) = δ(Y, 0) − δ(X, 0) = Ae - codim(f ∘ α) − δ(X, 0).
Example 5.5. We consider (X, 0) ⊂ (ℂ3 , 0), a weighted homogeneous curve that is defined by h(x, y, z) =
(x3 − y2 , xy − z) with parametrization α : (ℂ, 0) → (ℂ3 , 0) defined by α(t) = (t2 , t3 , t5 ), and the map f :
(X, 0) → (ℂ2 , 0) defined by f(x, y, z) = (x, y2 z); thus f is consistent with (X, 0). Then (Y, 0) is a weighted
homogeneous plane curve defined by g(u, v) = u11 − v2 with parametrization β : (ℂ, 0) → (ℂ2 , 0) defined by
β(t) = (t2 , t11 ).
In this case the Milnor number coincides with the conductor of the symmetric semigroup. We have Γ X =
⟨2, 3, 5⟩ and Γ Y = ⟨2, 11⟩, thus μ(X, 0) = 2 and μ(Y, 0) = 10. We observe that the Milnor number can also be
calculated in terms of weights and weighted degrees.
Thus
μ(Y, 0) + μ(X, 0)
μ I (f) = =6
2
and
Ae - codim(X, f) = μ I (f) = 6.
Theorem 5.4 also shows other formulas involving the image Milnor number and the Ae -codimension. We
know that Ae - codim(α) = 2δ(X, 0) = 2 (by [6]) and Ae - codim(f ∘α) = μ I (f ∘α) = δ(Y, 0) = 5 (by [8]). Therefore
μ I (f) = δ(X, 0) + μ I (f ∘ α) = 6
and
1
Ae - codim(f) = Ae - codim(f ∘ α) − Ae - codim(α) = 4.
2
Example 5.6. We consider (X, 0) ⊂ (ℂ3 , 0), a weighted homogeneous curve that is defined by h(x, y, z) =
(x5 − y2 , xy − z) with parametrization α : (ℂ, 0) → (ℂ3 , 0) defined by α(t) = (t2 , t5 , t7 ), and the map f :
(X, 0) → (ℂ2 , 0) defined by f(x, y, z) = (x2 , y + z). We observe that f is not consistent with (X, 0). Then, (Y, 0)
is a plane curve defined by g(u, v) = u7 − 2u6 + 4u3 v2 + u5 − v4 with parametrization β : (ℂ, 0) → (ℂ2 , 0)
defined by β(t) = (t4 , t5 + t7 ).
We have Γ X = ⟨2, 5, 7⟩, thus μ(X, 0) = 4 and τ(X, 0) = μ(X, 0) because (X, 0) is a weighted homogeneous
curve. We use the Singular software to calculate μ(Y, 0) = 12 and τ(Y, 0) = 11, consequently (Y, 0) is not
weighted homogeneous. Thus μ I (f) = (μ(Y, 0) + μ(X, 0))/2 = 8.
By [9] we have
Θ(γ)
Ae - codim(X, f) = dimℂ ,
tγ(Θ2 ) + γ∗ (Derlog D(G))
where G : ℂ4 → ℂ5 defined by G(x, y, z, s) = (x5 − y2 + sx, xy − z, x2 , y + z, s) is a stabilization of the map
(h, f ̃) : ℂ3 → ℂ4 given by (h, f ̃)(x, y, z) = (x5 − y2 , xy − z, x2 , y + z). We use the Singular software and obtain
Θ(γ)
Ae - codim(X, f) = dimℂ = 7.
tγ(Θ2 ) + γ∗ (Derlog Im(G))
Therefore Ae - codim(X, f) < μ I (f). We observe that the hypothesis that f be consistent with (X, 0) is necessary
to obtain an equality.
Funding: The first author has been supported by CAPES. The second author has been partially supported by
DGICYT Grant MTM2015–64013–P. The third author is partially supported by CNPq Grant 309086/2017-5 and
FAPESP Grant 2016/04740-7.
References
[1] D. A. H. Ament, J. J. Nuño Ballesteros, Mond’s conjecture for maps between curves. Math. Nachr. 290 (2017), 2845–2857.
[2] R.-O. Buchweitz, G.-M. Greuel, The Milnor number and deformations of complex curve singularities. Invent. Math. 58
(1980), 241–281. MR571575 Zbl 0458.32014
[3] R. Fröberg, C. Gottlieb, R. Häggkvist, On numerical semigroups. Semigroup Forum 35 (1987), 63–83.
MR880351 Zbl 0614.10046
[4] G.-M. Greuel, H. A. Hamm, Invarianten quasihomogener vollständiger Durchschnitte. Invent. Math. 49 (1978), 67–86.
MR511096 Zbl 0394.32006
[5] A. Hefez, M. E. Hernandes, Computational methods in the local theory of curves.
Instituto de Matemática Pura e Aplicada (IMPA), Rio de Janeiro 2001. MR1849596 Zbl 0994.14018
[6] M. E. Hernandes, M. E. Rodrigues Hernandes, M. A. S. Ruas, Ae -codimension of germs of analytic curves.
Manuscripta Math. 124 (2007), 237–246. MR2341787 Zbl 1131.32014