THE POLAR CURVE OF A FOLIATION ON P2

ROGÉRIO S. MOL

Departamento de Matemática
Universidade Federal de Minas Gerais
Av. Antônio Carlos, 6627 C.P. 702
30123-970 - Belo Horizonte - MG
BRAZIL
[email protected]

Abstract. We study some properties of the polar curve P F l associated

to a singular holomorphic foliation F on the complex projective plane
P2 . We prove that, for a generic center l ∈ P2 , the curve P F
l is irreducible
and its singular points are exactly the singular points of F with vanishing
linear part. We also obtain upper bounds for the algebraic multiplicities
of the singularities of F and for its number of radial singularities.

Résumé. On étudie dans cet article quelques propriétés de la courbe

polaire P Fl associée à un feuilletage holomorphe singulier F dans le
plan projectif complexe P2 . On démontre que, pour un centre l ∈ P2
générique, la courbe P Fl est irréductible et ses points singuliers sont
précisément les points singuliers de F avec partie linéaire nulle. On ob-
tient aussi des bornes supérieurs pour la multiplicité algébrique des sin-
gularités de F et pour son nombre de singularités radiales.

1. Basic definitions
A foliation of degree d ≥ 0 on the complex projective plane P2 = P2C is a
non trivial morphism of vector bundles Φ : H ⊗(1−d) → T P2 , where H stands
for the hyperplane bundle. Two such maps define the same foliation if one
is multiple of the other by a non zero complex number. We denote by F
the foliation and the bundle T F := H ⊗(1−d) is called the tangent bundle
of F. The singular set of F, denoted by Sing(F), is formed by the points
of P2 over which Φ fails to be injective. Equivalently, if ΘP2 = O(T P2 )
is the tangent sheaf of P2 and H = O(H) is the sheaf of sections of H,
defining F is equivalent to giving a morphism of locally free analytic sheaves
H⊗(1−d) → ΘP2 , two such morphisms defining the same foliation if and only
they are multiple from each other by a non zero complex number. Thus, the
space of foliations of degree d, denoted by Fol(d), can be identified with the

1Work partially supported by CAPES, FAPEMIG and Pronex/FAPERJ

1
2 ROGÉRIO S. MOL

projectivization of H 0 (P2 , Hom(H⊗(1−d) , ΘP2 )) ' H 0 (P2 , Θ2P (d−1)). Shortly

Fol(d) = P(H 0 (P2 , Θ2P (d − 1))).
Geometrically, the map Φ that defines F establishes, for every p ∈ P2 outside
Sing(F), a direction in Tp P2 which will be denoted by Tp F. The distribution
of tangent directions p 7→ Tp F, where p ∈ P2 \ Sing(F), defines, on this set,
a non singular holomorphic foliation. This distribution of directions corre-
sponds, in affine coordinates (x, y) ∈ P2 , to the one defined by a polynomial
vector field of the form
∂ ∂
(1) v = P (x, y) + Q(x, y) ,
∂x ∂y
To be more precise, this vector field can be written in the form
 
∂ ∂ ∂ e y) ∂ ,
(2) v = G(x, y) x +y + Pe(x, y) + Q(x,
∂x ∂y ∂x ∂y
where G(x, y), when non-zero, is a homogeneous polynomial of degree d,
while Pe(x, y) and Q(x,
e y) have degrees at most d. The line at infinity is
invariant by F if and only if G(x, y) ≡ 0, in which case at least one of the
degrees of Pe(x, y) or Q(x,
e y) is actually d. This writing for v shows that,
considered as a meromorphic vector field on P2 , it has a pole of order d − 1
in the line at infinity L∞ . The degree d is interpreted geometrically as the
number of tangencies, with multiplicities counted, between F and a line
L ⊂ P2 non invariant by F. The degree of a foliation F will be denoted by
deg(F).
It is evident that in the affine coordinates considered, Sing(F) is the set
of common zeroes of P (x, y) and Q(x, y). If this set has a component of
codimension one, it means that P (x, y) and Q(x, y) have a common factor.
So, by canceling this factor, we can always suppose that Sing(F) is a finite
set of points (evidently, in this case, the actual degree of F is less than the
number d we started with). Furthermore, by a proper choice of the affine
plane, we can always suppose that the line at infinity L∞ is not invariant
by F and that it contains no points in Sing(F).
Alternatively, we can represent a foliation F on P2 in homogeneous coor-
dinates (X : Y : Z) ∈ P2 . Suppose that the affine plane considered above
is Z = 1, so that x = X/Z and y = Y /Z. Let η = −Q(x, y)dx + P (x, y)dy
be the dual form of v. By taking the natural projection π : C3 \ {0} → P2 ,
we have that π ∗ η is a meromorphic 1-form in C3 with a pole of order d + 1
along Z = 0. By canceling this pole, we get a polynomial 1-form
(3) ω = A(X, Y, Z)dX + B(X, Y, Z)dY + C(X, Y, Z)dZ,
where A, B and C are homogeneous polynomials of degree d + 1. Evidently,
the rays from the origin of C3 must be contained in the leaves of the codi-
mension one foliation defined by ω. This property is expressed as the Euler
condition
(4) XA(X, Y, Z) + Y B(X, Y, Z) + ZC(X, Y, Z) = 0.
 THE POLAR CURVE OF A FOLIATION ON P2 3

We remark that, in this case, Sing(F) is the set of common zeroes of A, B

and C and the hypothesis that Sing(F) has codimension at least two is
equivalent to the fact that A, B and C have no common factor. In the
present text we will make use of both representations for F, in affine and in
homogeneous coordinates.
We now proceed to the definition of polar curve of a foliation. Let us
denote by TpP F the the line through p with direction Tp F. We call it the
projective tangent line of F at p ∈ P2 \ Sing(F). The polar locus of F with
center at l ∈ P2 is the closure of the set of points p ∈ P2 \ Sing(F) such that
TpP F passes through l:
PF 2
l = {p ∈ P \ Sing(F); l ∈ Tp F}.
P

In affine coordinates (x, y) ∈ P2 as above, taking l = (x0 , y0 ), P F

l has as
equation
(5) (y − y0 )P (x, y) − (x − x0 )Q(x, y) = 0.
In homogeneous coordinates (X : Y : Z) ∈ P2 , when F is induced by a
1-form as in (3), the polar locus with center l = (α : β : γ) is given by the
equation
(6) αA(X, Y, Z) + βB(X, Y, Z) + γC(X, Y, Z) = 0.
Anyone of the two previous equations shows easily that the center l always
belongs to the polar curve P Fl .
Recall that the radial foliation with center l ∈ P2 is the one given in affine
coordinates (x, y) ∈ P2 by the vector field
∂ ∂
v = (x − x0 ) + (y − y0 ) ,
∂x ∂y
where l = (x0 , y0 ). An easy calculation shows that in homogeneous coor-
dinates (X : Y : Z) ∈ P2 the radial foliation centered at l = (α : β : γ) is
given by the 1-form
η = (βZ − γY )dX + (γX − αZ)dY + (αY − βX)dZ.
We have the following:
Proposition 1. The polar locus of a foliation F on P2 with center l ∈ P2
is the whole P2 if and only if F is the radial foliation with center l. In all
other cases the polar locus is a curve of degree d + 1, where d is the degree
of F.
Proof. Let us suppose that αA(X, Y, Z) + βB(X, Y, Z) + γC(X, Y, Z) ≡ 0.
We assume that γ = 1, so that
(7) C = −(αA + βB).
Putting this in Euler’s relation (4) we get
XA + Y B − Z(αA + βB) = 0 ⇔ (X − αZ)A = −(Y − βZ)B.
4 ROGÉRIO S. MOL

If d0 = deg(A) = deg(B) = d + 1 ≥ 2, we would find a common factor

for A and B. By (7), this would also be a factor for C, which contradicts
the fact that the points in Sing(F) are isolated. Thus, d0 = 1 and, modulo
multiplication by a non-zero constant, we have

A = −(Y − βZ) , B = X − αZ,

and, by Euler’s relation, C = αY − βX. We conclude that F is induced by

the 1-form

ω = −(Y − βZ)dX + (X − αZ)dY + (αY − βX)dZ,

that is, F is the radial foliation with center (α : β : 1) ∈ P2 . 

By varying the center l = (α : β : γ) of the polar curve, we produce a linear

system in P2 generated by the divisors A = 0, B = 0 and C = 0. We remark
that, if d = deg(F) ≥ 1, this linear system has dimension two. In fact, if this
dimension were less than two, then one of the polynomials A, B or C would
be written as a linear combination of the others, say C = αA + βB with
α, β ∈ C. We are then in the situation of the proof of the above proposition,
and a contradiction is reached. Thus, for a foliation F of degree at least 1
we have a net of polar curves, the so-called polar net. Its base points are
exactly the singular points of F. The polar net of a foliation was thoroughly
studied in [4], where it is used to prove that, for d ≥ 2, the scheme Sing(F)
determines F.
Polar curves are a particular case of more general objects, polar vari-
eties associated to distributions in projective manifolds. These objects were
studied in [10] and a short description is given below. First, a singular
holomorphic distribution F of dimension r on a complex manifold M of di-
mension m is a coherent analytic subsheaf T of rank r of the tangent sheaf
ΘM = O(T M ) of M . We call T the tangent sheaf of F. The singular set of
F, denoted Sing(F), is defined as the singular set of the sheaf ΘM /T , that
is, the set of points where the stalks are not free modules over OM . When
T is involutive, that is, when its stalks are invariant by the Lie bracket,

[Tp , Tp ] ⊂ Tp ∀ p ∈ M,

T actually determines a (non singular) foliation on M \ Sing(F). This is a

consequence of Frobenius theorem. However, integrability is not necessary
for the forthcoming definition.
Let us now suppose that the ambient manifold M is projective, say
M ⊂ Pn = PnC . If F is a distribution with tangent sheaf T defined on
M , then for each p ∈ M \ Sing(F), there is a unique r-dimensional plane
TpP F ⊂ Pn passing through p with direction determined by the stalk Tp in
Tp M ⊂ Tp Pn . We fix a flag of linear subspaces on Pn

(8) D : Lr+1 ⊂ Lr ⊂ · · · ⊂ L2 ⊂ Pn ,
 THE POLAR CURVE OF A FOLIATION ON P2 5

where codimLj = j. For k = 1, . . . , r, the k-th polar locus of F with respect

to D is defined as
PkF = Cl{p ∈ M \ Sing(F); dim(TpP F ∩ Lr−k+2 ) ≥ k − 1},
where Cl stands for the closure in M . We remark that p ∈ M \ Sing(F)
belongs to P F
k if and only if the subspaces of C
n+1 corresponding to T P F and
p
to Lr−k+2 fail to span Cn+1 . For a generic choice of D and for k = 1, . . . , r,
PFk is empty
 or is a reduced analytic variety of pure codimension k whose
F
class P k ∈ Am−k (M ) is independent of the flag, Am−k (M ) standing for
the Chow group of M of complex dimension m − k. When F is a foliation
on P2 , then the polar curve with center l is the P1F defined above taking
L2 = {l} as the element of codimension two in the flag.

2. Some properties of the polar curve

From this point to the end of the text we will fix our attention on foliations
on P2 with isolated singularities. We will start by a definition:

Definition 1. Let F and Fe be foliations on P2 of the same degree d. We say

that Fe is a radial modification of F with center at l ∈ P2 if, in homogeneous
coordinates (X : Y : Z) ∈ P2 , there are polynomial 1-forms ω and ω̃ that
induce F and Fe and a homogeneous polynomial Φ(X, Y, Z) of degree d such
that
ω̃ = ω + Φη,
where η = (βZ − γY )dX + (γX − αZ)dY + (αY − βX)dZ is the 1-form that
induces the radial foliation centered at l = (α : β : γ).
The notion of radial modification defines an equivalence relation in the
space of foliations on P2 and polar curves classify the equivalence classes
under this relation as shown in the following proposition:

Proposition 2. Let F and Fe be foliations on P2 of the same degree d ≥ 1.

Denote by P F F 2
l and P l the polar loci of F and F centered at l ∈ P . Then
e e
PF F
l = P l if and only if F is a radial modification of F with center at l.
e e

Proof. Let ω and ω̃ be 1-forms defining F and F. e We first suppose that

F F
P l = P l . By multiplying ω and ω̃ by non zero constants, we may suppose
e

that that P F F
l and P l have the same equation. We keep the notation ω and
e

ω̃ for these new 1-forms. Thus, the foliation induced by ν = ω̃ − ω has P2

as polar locus with center l. By proposition 1, this foliation must be the
radial foliation with center l, which is induced by the 1-form η as above. In
order to have compatible degrees, we find Φ(X, Y, Z) homogeneous of degree
d such that ν = ω̃ − ω = Φη. Thus ω̃ = ω + Φη, which gives the result. The
proof for the converse is straightforward. 
6 ROGÉRIO S. MOL

Let us now fix affine coordinates (x, y) ∈ P2 . Suppose that l = (0, 0) is the
center of the polar curve P Fl , which has F (x, y) = yP (x, y) − xQ(x, y) = 0
as equation. Remark that P F l is invariant by F if and only if the tangent
lines to P Fl and to F coincide wherever both are defined. It follows from
F
the very definition of P l that this is equivalent to the fact that the tangent
lines to P F
l pass through l = (0, 0). This is true if and only if

r(F (x, y)) = xFx (x, y) + yFy (x, y) ≡ 0

on P Fl , where r = x∂/∂x + y∂/∂y is the radial vector field centered at
l = (0, 0). Notice that when F (x, y) is homogeneous of degree d, by Euler’s
formula we have
r(F (x, y)) = dF (x, y),
which evidently vanishes on P F
l . This motivates the following definition:

Definition 2. We say that a foliation F on P2 is homogeneous with center at

l ∈ P2 if F is induced in affine coordinates (x, y) ∈ P2 centered at l = (0, 0)
by a polynomial vector field
∂ ∂
v = P (x, y) + Q(x, y)
∂x ∂y
such that P (x, y) and Q(x, y) are homogeneous polynomials of the same
degree.
Let F be a foliation as in the above definition, where P and Q are homo-
geneous polynomials of degree d0 . In view of expression (2), if G 6≡ 0, we
would have deg(F) = d0 − 1 and P̃ = Q̃ = 0, so that F would be the radial
foliation centered at l = (0, 0) having G = 0 as a curve of singularities. So,
excluding the possibility of non isolated singularities, we must have G ≡ 0,
giving P = P̃ , Q = Q̃ and deg(F) = d0 . We remark that, in this case, the
line at infinity is F-invariant.
If F is a homogeneous foliation centered at l = (0, 0), then its polar curve
PFl is defined in affine coordinates as the set of zeroes of a homogeneous
polynomial and, as we have seen above, it is F-invariant. More generally,
we have:
Proposition 3. Let F be a foliation on P2 . Suppose that its polar curve
PF 2 F
l centered at l ∈ P is reduced. Then P l is F-invariant if and only if F
is a radial modification of a homogeneous foliation centered at l.
Proof. We have already proved the “if” part. Let us then suppose that P F l
is invariant by F. We fix affine coordinates (x, y) ∈ P2 for which l = (0, 0).
We have
xFx (x, y) + yFy (x, y) ≡ 0
on P F F
l . Since P l is reduced with equation F (x, y) = 0, it follows that
F (x, y) divides xFx (x, y) + yFy (x, y). Since the degree of the latter is less
 THE POLAR CURVE OF A FOLIATION ON P2 7

than or equal to the degree of the former, we conclude that both polynomials
have the same degree and, thus, there exists a non-zero constant α such that
αF (x, y) = xFx (x, y) + yFy (x, y).
But this happens if and only if F (x, y) is homogeneous, in which case α is
the degree of F (x, y). Let us now look at the expression
F (x, y) = yP (x, y) − xQ(x, y).
We decompose P (x, y) = P0 (x, y) + P1 (x, y) in the following way: P0 (x, y)
involves all monomials which vanish when we produce F (x, y) and P1 (x, y)
comprises the ones which do not vanish. We do the same for Q(x, y), getting
Q(x, y) = Q0 (x, y) + Q1 (x, y). On the one hand, we have
F (x, y) = yP1 (x, y) − xQ1 (x, y)
and, since F (x, y) is homogeneous, all monomials in P1 (x, y) and in Q1 (x, y)
have the same degree, so that the foliation induced by the vector field
∂ ∂
P1 (x, y) + Q1 (x, y)
∂x ∂y
is homogeneous. On the other hand, we have
yP0 (x, y) − xQ0 (x, y) ≡ 0,
so that there exists a polynomial Φ(x, y) such that P0 (x, y) = xΦ(x, y) and
Q0 (x, y) = yΦ(x, y). Therefore we conclude that
 
∂ ∂ ∂ ∂ ∂ ∂
P (x, y) +Q(x, y) = P1 (x, y) +Q1 (x, y) +Φ(x, y) x +y ,
∂x ∂y ∂x ∂y ∂x ∂y
meaning that F is a radial modification of a homogeneous foliation. 

3. Singularities of the polar curve

We recall that as l runs over P2 , the curves P F l describe a linear system
of dimension two with base Sing(F). By Bertini’s theorem (see [9]), the
generic element of a linear system is smooth outside the base locus. So, for
generic l ∈ P2 , the singular set of P F
l is contained in Sing(F).
Suppose that F is induced in affine coordinates by a vector field as in (1),
so that its polar curve centered at l = (x0 , y0 ) is given by the equation
F (x, y) = (y − y0 )P (x, y) − (x − x0 )Q(x, y) = 0.
For future reference, we calculate the derivatives of F with respect to x and
y:

Fx (x, y) = (y − y0 )Px (x, y) − (x − x0 )Qx (x, y) − Q(x, y)
(9)
F (x, y) = P (x, y) + (y − y )P (x, y) − (x − x )Q (x, y)
y 0 y 0 y

These equations give promptly that the center l is a regular point for P F
l
whenever it lies outside Sing(F).
8 ROGÉRIO S. MOL

Fix now p ∈ Sing(F). We may suppose that p = (0, 0), and thus
P (0, 0) = Q(0, 0) = 0. Let k = mp (F) be the algebraic multiplicity of F
at p, that is, the smallest of the orders of P and Q. We write
P (x, y) = Pk (x, y) + P̃ (x, y) and Q(x, y) = Qk (x, y) + Q̃(x, y),
where Pk and Qk comprise all monomials of degree k, while P̃ and Q̃ as-
semble those of higher degree. We thus have
F (x, y) = −y0 Pk (x, y) + x0 Qk (x, y) + S(x, y),
where all monomials in S(x, y) have degree k + 1 or greater. We con-
clude that, for generic l = (x0 , y0 ), the algebraic multiplicity of P F
l at p
is k = mp (F). Summarizing this discussion:
Proposition 4. For generic l ∈ P2 it holds Sing(P F
l ) ⊂ Sing(F). Further-
F
more if p ∈ Sing(P l ), then
mp (P F
l ) = mp (F) ,
where mp denotes the algebraic multiplicity.
Keeping the above notation, we say that p ∈ Sing(F) is a non-degenerate
singularity if det Dv(0, 0) 6= 0. According to the above proposition, if p is a
non-degenerate singularity for F, then p is a regular point for P F
l . Recalling
that foliations having only non-degenerate singularities are generic in the
space of foliation of degree d, we have
Corollary 1. For generic F ∈ Fol(d), the polar curve P F
l centered at a
generic point l ∈ P2 is non-singular.

4. Radial singularities
Let us denote by Sing1,dic (F) ⊂ Sing(F) the set of dicritical singulari-
ties of algebraic multiplicity 1. We remark that Sing1,dic (F) contains the
set of radial singularities, consisting of those which become non singular af-
ter a single blow-up. Let us suppose that p = (0, 0) ∈ Sing1,dic (F). The
polynomial vector field inducing F has the form
∂ ∂
v = (x + P̃ (x, y)) + (y + Q̃(x, y)) ,
∂x ∂y
where P̃ and Q̃ refer to the terms of degree 2 or greater (P̃ = Q̃ in the case
of radial singularities).
In this case, the polar curve P F
l centered at l = (x0 , y0 ) is defined by the
equation
F (x, y) = (y − y0 )(x + P̃ (x, y)) − (x − x0 )(y + Q̃(x, y)) = 0.
We have

Fx (x, y) = (y − y0 )(1 + P̃x (x, y)) − (x − x0 )Q̃x (x, y) − (y + Q̃(x, y))
F (x, y) = (x + P̃ (x, y)) + (y − y )P̃ (x, y) − (x − x )(1 + Q̃ (x, y))
y 0 y 0 y
 THE POLAR CURVE OF A FOLIATION ON P2 9

so that Fx (0, 0) = −y0 and Fy (0, 0) = x0 . This implies that the tangent
line to P F
l at p = (0, 0) passes through l = (x0 , y0 ). Thus, Sing1,dic (F) is
contained in the set of tangencies between P F l and the pencil of lines with
base l. The number of these tangencies, with multiplicities counted, is the
degree of P̌lF , the projective dual of P F
l (see [12]). Noticing that l is in this
set of tangencies but may be chosen outside Sing(F), we have the following:
Proposition 5. Let F be a foliation on P2 . Then
X
(10) #Sing1,dic (F) ≤ deg(P̌lF )−1 = d(d+1)− (m ep −1)−1,
e p +µ
p∈Sing(PlF )

where me p and µep refer to the algebraic multiplicity and the Milnor number
F
of P l at p. In particular, this gives un upper bound for the number of radial
singularities of F.
We remark that the upper bound in (10) is written in terms of algebraic
multiplicities and Milnor numbers of the polar curve P F l and it works for
any center l ∈ P2 outside Sing(F) and not specifically in the generic case.
Nevertheless, by taking P F l as the generic polar curve, we can establish a
bound in terms of data of the foliation F. We proceed as follows. Let
us denote by mp and µp the algebraic multiplicity and the Milnor number
of F at p ∈ Sing(F). We know that, in this case, m e p = mp . In what
concerns Milnor numbers, it is easy to see from equations (9) that, for generic
l ∈ P2 , the vector field Fx (x, y)∂/∂x+Fy (x, y)∂/∂y has algebraic multiplicity
mp − 1 at p = (0, 0) ∈ Sing(F). Thus µ ep ≥ (mp − 1)2 for every p ∈ Sing(F)
(see [3]), and we get that the term m ep +µep − 1 in the sum in (10) is greater
than or equal to mp + (mp − 1)2 − 1 = mp (mp − 1). Substituting in (10), we
have
X
(11) #Sing1,dic (F) ≤ d(d + 1) − mp (mp − 1) − 1.
p∈Sing(F )

5. Irreducibility of the generic polar

The generic element of a linear system of curves in the projective plane,
or, more generally, of projective divisors, need not be irreducible. How-
ever, projective linear systems whose generic element is irreducible are well
characterized by the Bertini-Krull theorem (see [1] and theorem 37 in [11]):
Theorem 1. Let K be an algebraically closed field of characteristic 0. Let
r ≥ 1 and P0 , P1 , . . . , Pr be non-zero distinct polynomials, not all of them
constant, relatively prime in K[x], where x = (x1 , . . . , xn ) are independent
variables. Then the set of λ = (λ0 : λ1 : . . . : λr ) ∈ Pr = PrC such that
F = F (x, λ) = P0 + λ1 P1 + · · · + λr Pr
is reducible in K[x] is a Zariski dense set in Pr if and only if the following
condition holds: there exist relatively prime polynomials Φ, Ψ ∈ K[x], with
10 ROGÉRIO S. MOL

degx (F ) > max{deg(Φ), deg(Ψ)}, and r + 1 polynomials hi (u, v) ∈ K[u, v]

homogeneous of degree s, for some s > 1, such that
s
X
Pi = hi (Φ(x), Ψ(x)) = aik Φ(x)k Ψ(x)s−k , i = 0, . . . , r.
k=0
This means that, setting
r
X
H(u, v, λ) = λi hi (u, v),
i=0

we have F (x, λ) = H(Φ(x), Ψ(x), λ).

We remark that the polar net of a foliation on P2 is not an arbitrary
net, since its generators satisfy the Euler condition. This fact along with
Bertini-Krull theorem allow us to prove the following:
Proposition 6. The generic polar curve of a foliation on P2 is irreducible
Proof. We work in homogeneous coordinates. Suppose that the foliation F
is induced by the 1-form
ω = A(X, Y, Z)dX + B(X, Y, Z)dY + C(X, Y, Z)dZ,
where A, B and C are homogeneous polynomials of degree d+1 = deg(F)+1,
satisfying the Euler condition XA+Y B +ZC ≡ 0. Suppose that the generic
element of the polar net of F,
αA(X, Y, Z) + βB(X, Y, Z) + γC(X, Y, Z) = 0 ; (α : β : γ) ∈ P2 ,
is reducible. Then, by the Bertini-Krull theorem, there exists relatively
prime polynomials Φ, Ψ ∈ C[X, Y, Z], with max{deg(Φ), deg(Ψ)} < d + 1,
and homogeneous polynomials h0 (u, v), h1 (u, v), h2 (u, v) ∈ C[u, v] of degree
s such that
A = h0 (Φ, Ψ) , B = h1 (Φ, Ψ) , C = h2 (Φ, Ψ).
Let us write
s
X
hi (u, v) = aik uk v s−k , i = 0, 1, 2.
k=0
Since XA + Y B + ZC ≡ 0, putting all terms in Φs in the same side, we get
s−1
X
s
(Xa0s + Y a1s + Za2s )Φ = − (Xa0k + Y a1k + Za2k )Φk Ψs−k
k=0
We first remark that Xa0s +Y a1s +Za2s is non zero for, otherwise, we would
have a0s = a1s = a2s = 0 and Ψ would be a common factor for A, B and C.
Thus, since Φ and Ψ are relatively prime, the above expression gives that Ψ
divides Xa0s +Y a1s +Za2s . But this is possible if and only if deg(Ψ) = 1 and
Ψ = λ(Xa0s +Y a1s +Za2s ) for some λ ∈ C\{0}. By arguing in a symmetric
manner, we also have that deg(Φ) = 1 and Φ = ν(Xa00 + Y a10 + Za20 ) for
 THE POLAR CURVE OF A FOLIATION ON P2 11

some ν ∈ C \ {0}. To show that this leads to a contradiction, let us make

the following linear change of coordinates:

u = Φ(X, Y, Z) = ν(Xa00 + Y a10 + Za20 )




v = Ψ(X, Y, Z) = λ(Xa0s + Y a1s + Za2s )




w = Λ(X, Y, Z) = Xa + Y b + Zc

where Λ is any linear form such that Φ, Ψ and Λ are linearly independent.
If (u, v, w) = F (X, Y, Z) denotes this linear change of coordinates, then F
is induced in the new coordinates by a 1-form of the kind
ω̃ = F ∗ ω = Ã(u, v)du + B̃(u, v)dv + C̃(u, v)dw,
where Ã, B̃, C̃ are homogeneous polynomial of degree d + 1, in the variables
u and v only. Euler condition gives −wC̃(u, v) = uÃ(u, v) + v B̃(u, v), which
implies that C̃(u, v) = 0. This is absurd, since, when d = deg(F) ≥ 1, Ã, B̃
and C̃ are linearly independent, as already proved in the text. 

6. The genus of the polar curve

We summarize here the properties satisfied by the polar curve P F
l for generic
l ∈ P2 :
(i) P F
l is irreducible;
(ii) the degrees satisfy d(P F
l ) = d(F) + 1;
(iii) Sing(P F
l ) ⊂ Sing(F);
(iv) mp (F) = mp (P F l ) ∀ p ∈ Sing(F).
We remark that the genus, since it is calculated by a finite combinatorial
process, is constant for the generic element of the polar net of F. Property
(i) allows us to apply Noether’s formula for the genus of a plane curve (see
[2]), yielding
(d(P F F
l ) − 1)(d(P l ) − 2) 1X
g = − mp (P F F
l )(mp (P l ) − 1)
2 2
p∈S̃

d(d − 1) 1 X
= − mp (P F F
l )(mp (P l ) − 1) ,
2 2
p∈S̃

where d = deg(F) and S̃ denotes all points infinitely near Sing(P F l ). In

particular, when mp (F) = 1 for all points in Sing(F), we have that the
the generic polar curve is smooth and its genus is g = d(d − 1)/2, where
d = deg(F).
Now, we consider the fact that g ≥ 0 and, in Noether’s formula, we neglect
the points in S̃ which appear in the first blow up and after. By taking into
account properties (iii) and (iv), we get the following bound for the algebraic
multiplicities of the singularities of F in terms of d = deg(F):
12 ROGÉRIO S. MOL

Proposition 7. For a foliation F of degree d ≥ 1 on P2 it holds

X
mp (mp − 1) ≤ d(d − 1) ,
p∈Sing(F )

where mp denotes the algebraic multiplicity of p ∈ Sing(F).

We say that two foliations F and F̃ defined on surfaces M and M̃ are
topologically equivalent if there is a homeomorphism h : M → M̃ mapping
Sing(F) into Sing(F̃) and taking leaves of F into leaves of F̃. Local topo-
logical equivalences preserve Milnor number (see [3]). Thus, if p ∈ M is
non-degenerate, which is equivalent to the fact that its Milnor number is
one, then h(p) is also non-degenerate.
Two foliations on P2 which are topological equivalent have the same de-
gree. In fact, the first Chern class of the tangent bundle of a foliation is
invariant by topological equivalences (see [8]). If F is a foliation on P2 ,
then its tangent bundle T F is H ⊗(d−1) , where d = deg(F). Another way to
see the topological invariance of the degree is by considering Baum-Bott’s
theorem, which, for a foliation F of degree d on P2 , reads
X
µp = d2 + d + 1,
p∈Sing(F )

where µp denotes the Milnor number at p ∈ Sing(F) (see, for instance,

proposition 1.1 in [4]). The topological invariance of d is then a consequence
of the topological invariance of Milnor numbers.
The above considerations give the following:
Proposition 8. Let F be a foliation on P2 of degree d ≥ 1 with only
non-degenerate singularities. Then the genus of its generic polar curve is
g = d(d − 1)/2. Furthermore, this genus is a topological invariant, meaning
that if F̃ is a foliation on P2 topologically equivalent to F, then the genus of
the generic polar curve of F̃ is the same.
This result leads to the following question:

Question: Is the genus of the generic polar curve of a foliation F on P2 a

topological invariant of the foliation?

The difficulty for answering this question is the fact that, in general, the
desingularization data of the generic polar is not read in the desingulariza-
tion data of the foliation. N. Corral has extensively worked on this subject,
comparing, for a singularity p of a foliation F on P2 which is a generalized
curve (see [3] for the definition), the desingularization of F, which is equiv-
alent to the desingularization of the set of its separatrices at p, with the
desingularization of the generic polar curve (see [5], [7] and [6]).

Acknowledgements. I am grateful to M. Soares and to P. Sad, for their

helpful comments, to the Institut de Mathématiques de Luminy (France)
and to Universidad de Valladolid (Spain), for hospitality.
 THE POLAR CURVE OF A FOLIATION ON P2 13

References
[1] Arnaud Bodin, Pierre Dèbes, and Salah Najib. Irreducibility of hypersurfaces. Comm.
Algebra, 37(6):1884–1900, 2009.
[2] Egbert Brieskorn and Horst Knörrer. Plane algebraic curves. Birkhäuser Verlag,
Basel, 1986.
[3] César Camacho, Alcides Lins Neto, and Paulo Sad. Topological invariants and
equidesingularization for holomorphic vector fields. J. Differential Geom., 20(1):143–
174, 1984.
[4] Antonio Campillo and Jorge Olivares. Polarity with respect to a foliation and Cayley-
Bacharach theorems. J. Reine Angew. Math., 534:95–118, 2001.
[5] Nuria Corral. Sur la topologie des courbes polaires de certains feuilletages singuliers.
Ann. Inst. Fourier (Grenoble), 53(3):787–814, 2003.
[6] Nuria Corral. Infinitesimal adjunction and polar curves. Bull. Braz. Math. Soc. (N.S.),
40(2):181–224, 2009.
[7] Nuria Corral. Polar pencil of curves and foliations. Astérisque, (323):161–179, 2009.
[8] X. Gómez-Mont, J. Seade, and A. Verjovsky. The index of a holomorphic flow with
an isolated singularity. Math. Ann., 291(4):737–751, 1991.
[9] Phillip Griffiths and Joseph Harris. Principles of algebraic geometry. Wiley Classics
Library. John Wiley & Sons Inc., New York, 1994.
[10] Rogério S. Mol. Classes polaires associées aux distributions holomorphes de sous-
espaces tangents. Bull. Braz. Math. Soc. (N.S.), 37(1):29–48, 2006.
[11] A. Schinzel. Polynomials with special regard to reducibility, volume 77 of Encyclopedia
of Mathematics and its Applications. Cambridge University Press, Cambridge, 2000.
[12] C. T. C. Wall. Singular points of plane curves, volume 63 of London Mathematical
Society Student Texts. Cambridge University Press, Cambridge, 2004.