Bull. Korean Math. Soc. 46 (2009), No. 1, pp.

171182
EMBEDDINGS OF LINE IN THE PLANE AND
ABHYANKAR-MOH EPIMORPHISM THEOREM
Dosang Joe and Hyungju Park
Abstract. In this paper, we consider the parameter space of the rational
plane curves with uni-branched singularity. We show that such a param-
eter space is decomposable into irreducible components which are ratio-
nal varieties. Rational parametrizations of the irreducible components
are given in a constructive way, by a repeated use of Abhyankar-Moh
Epimorphism Theorem. We compute an enumerative invariant of this
parameter space, and include explicit computational examples to recover
some classically-known invariants.
1. Introduction
Let C be a curve in the complex ane plane A
2
C
. We call C an un-
parametrized ane embedding of line of degree d if it is isomorphic to A
1
C
and
its projective closure C is a curve of degree d in P
2
C
. We call such a curve C
a smooth rational ane plane curve of degree d. Denote by /(d) the space of
all unparametrized ane embeddings of line of degree d. Note that the notion
of the degree of C is well-dened with a xed ane structure on A
2
C
since any
ane automorphism can be extended to an automorphism of P
2
C
.
Many results are available on the local properties of the parameter space
/(d). For example, one knows that it is a union of smooth irreducible varieties
[9]. Our main interest in this paper is in studying some of its global proper-
ties. As a result, we show that /(d) is a disjoint union of smooth irreducible
rational varieties. In fact, we give explicit rational parametrizations of all of
its irreducible components, and compute the dimensions of the components. In
particular, we show that the (maximal) dimension of /(d) is d + 2 for d 2,
and that there are exactly (
d(d+1)
2
1) distinct unparametrized embeddings of
line (i.e., embeddings of line up to reparametrization) passing through general
d + 2 points in A
2
C
for d 2.
Received May 20, 2008.
2000 Mathematics Subject Classication. 14N10, 14Q05, 14R10.
Key words and phrases. embedding of ane line, Abhyankar-Moh epimorphism theorem,
enumerative problems, Puiseux pairs.
The rst author was supported by the Faculty Research Fund of Konkuk University in
2004.
c 2009 The Korean Mathematical Society
171
172 DOSANG JOE AND HYUNGJU PARK
The main ingredient of the proof is the Abhyankar-Moh Epimorphism The-
orem [1]. To explain this connection more explicitly, let us start by recalling a
basic denition.
Denition 1. A polynomial map = (f, g) : C
1
C
2
is called a parametrized
ane embedding of line, or simply an embedding of line, if C[t] = C[f(t), g(t)].
Note that the above condition amounts to the condition that the image
curve in C
2
is isomorphic to C
1
under the polynomial map. The Abyhankar-
Moh Epimorphism Theorem provides a necessary condition on the degrees of
f and g for (f, g) to be an embedding of line.
Theorem 1 ([1]). Suppose f, g are in C[t] with m := deg g n := deg f. If
C[f(t), g(t)] = C[t], then m divides n.
Among the curves parametrically dened by = (f, g) : C
1
C
2
with
m dividing n, there are smooth curves and non-smooth curves. An obvious
question arises now:
How many such curves are smooth?
Since /(d) can be viewed as the space of all embeddings of line of degree
d up to reparametrization, this question naturally prompts one to study the
parameter space /(d). In fact, this was the initial motivation for the authors
to study the parameter space /(d).
We also study the parameter space /(d) of unicuspidal rational curves
with maximal tangent [4]. /(d) can be considered as a sub-variety of /(d),
consisting of the curves whose singular points lie on the line at innity which
is the maximal tangent line. Here, the maximal tangent line means the line
which intersect the curve only at the singular point.
The variety /(d) has the main component /
d
which has exactly the largest
dimension among the irreducible components. The explicit parametrization of
a Zariski open subset /

d
of the main component /
d
is given by monomial terms
with one exception in the homogeneous coordinates of PH
0
(P
2
, O(d)). So, in a
sense the main component is almost a toric variety. Using this parametrization,
we compute the degree of the variety /(d). Two proofs are given: one by using
a projection and Kouchinirenko theorem on the number of solutions of a system
of equations [16, 7], the other one by a direct method.
The same parametrization can be used for degree counting for the parameter
space of rational projective plane curves with unique irreducible singular points.
As an illustration of our technique, we attempt to compute the degree of the
parameter space /(3) of cuspidal cubic curves in P
2
C
. An explicit computation
with the computer algebra package Singular [10] shows that this degree is 24,
coinciding with a classical result [18, 15, 2].
After writing up this paper, we learned that the degree of /(d) had been
independently obtained in a recent paper [13], as an example. The approaches,
however, appear to be substantially dierent. It is to be noted that our method
EMBEDDINGS OF LINE IN THE PLANE 173
produces not only the degree of the parameter space /(d) but also an explicit
parametrization of /(d) itself.
2. Decomposition of embeddings of line
Let us start by recalling basic terminologies.
Denition 2 (c.f. [17]). A polynomial map : C
2
C
2
is called an elemen-
tary polynomial map of C
2
if (x, y) is (x, y + f(x)) or (x + g(y), y) for some
polynomial f(x) C[x] or g(y) C[y].
The Abyhyankar-Moh can be used to reduce the degrees of by applying
an elementary polynomial map.
Lemma 1. Let = (f, g) : C
1
C
2
be an embedding of line in C
2
where
f(t) = a
0
t
n
+a
1
t
n1
+ +a
n
g(t) = b
0
t
m
+b
1
t
m1
+ +b
m
.
Suppose deg f = n = k deg g = km for a positive integer k. Then there is a
unique elementary polynomial map (x, y) = (x c(y), y) such that (t) =
(h(t), g(t)), c(0) = 0, and deg h is strictly less than m. Furthermore, if f and g
are viewed as polynomials in Z[a
0
, . . . , a
n
, b
0
, . . . , b
m
][t], then c is a polynomial
in Z[a
0
, . . . , a
n
, b
0
, . . . , b
n
][
1
b
0
][t].
Proof. There are unique c
0
, c
1
, . . . , c
k1
C such that the degree of
h(t) := f(t) (c
0
g(t)
k
+c
1
g(t)
k1
+ +c
k1
g(t))
is less than the degree of g. This is possible due to a successive application of
Theorem 1. Let c
0
= a
0
/b
k
0
, and suppose c
0
, c
1
, . . . , c
l1
are dened then c
l
is
determined as
c
l
= (a
lm
c
0
b
k,nlm
c
1
b
k1,nlm
c
l1
b
kl+1,nlm
)/b
kl
0
,
where g(t)
p
= (

m
i=0
b
i
t
mi
)
p
=

n
j=0
b
p,j
t
j
.
Denition 3. (1) Denote by Mor(C
1
, C
2
) the locus of all embeddings of
parametrized ane line and dene
(n, ) = (f, g) [ deg f = n, deg g < n.
Similarly, dene
(n, m) = (f, g) [ deg f = n, deg g = m
and

(n) = (f, g) [ deg f = n, deg g n.

(2) Let C
k
= C

C
k1
for k > 1 and C
k
1
,...,k
q
=

q
i=1
C
k
i
for k
i
> 1.
174 DOSANG JOE AND HYUNGJU PARK
Note that (n, ) =

m|n, m<n
R(n, m). Lemma 1 will be used to decompose
(n, ) further.
Dene, for k > 1,

k,m
: C
k
(m, ) (mk, )
(c, g, h) (h +
k1

i=0
c
i
g
ki
, g).
Given ordered positive integers k
1
, . . . , k
q
with

q
i=1
k
i
= n and k
i
> 1, by
iteratively composing s we obtain a map

(k
1
,...,k
q
)
: C
k
1
(C
k
2
(C
k
q
(1, )) ) (n, ),
where
(k
1
,...,k
q
)
:=
k
1
,k
2
k
q

k
q1
,k
q

k
q
,1
.
Denition 4. Denote the image of this composition map by

k
where k =
(k
1
, . . . , k
q
).
Lemma 2.

k
is a locally closed subset of C
n+1
C
m+1
under
k
, where
n =

q
i=1
k
i
, m =

q
i=2
k
i
. Furthermore it is isomorphic to C
k
(1, ) as a
variety.
Proof. Consider

k
: C
k
(1, ) C
n+1
C
m+1
and its converse

k
: U C
k
(1, )
in some open neighborhood U of

k
in C
n+1
C
m+1
. Then
k

k
is the
identity. However
k

k
() = if and only if

k
. This shows that

k
is
a closed subset of U.
Lemma 3. Let k > 1. Endow (l, ) with the induced Zariski topology as a
subspace of C
l+1
C
l
. Then the polynomial map : C
k
(m, ) (mk, )
sending (c, g, h) (h +

k1
i=0
c
i
g
ki
, g) is a closed map.
Proof. Let n = mk. Consider the sequences (c
(j)
, g
(j)
, h
(j)
) such that the
limit (f
()
, g
()
) of (c
(j)
, g
(j)
, h
(j)
) exists in (m, ). Then the sequence
g
(j)
has the limit g
()
= b
0
t
m
+ + b
m
in C
m+1
. Since the rst component
of (c
(j)
, h
(j)
, g
(j)
)) has degree mk > 1, g
()
must not be zero.
Case I: Suppose that b
0
= 0. We will show that this cannot happen. Since
the coecient of t
mk
in c
(j)
0
(g
(j)
)
k
converges to a nonzero number, [c
(j)
0
[ goes to
. By a change of coordinate t, we may assume that g
()
(t) = b
ml
t
l
+O(t
l+1
)
where n > l 1 and b
ml
,= 0. Now consider the coecients of t
mk
, t
mk1
, . . .,
EMBEDDINGS OF LINE IN THE PLANE 175
t
m(k1)+l
in h
(j)
+

c
(j)
i
(g
(j)
)
ki
, which are
c
(j)
0
(b
(j)
0
)
k
,
c
(j)
0
(b
(j)
0
)
k1
b
(j)
1
,
c
(j)
0
(k(b
(j)
0
)
k1
b
(j)
2
+
k(k 1)
2
(b
(j)
0
)
k2
(b
(j)
1
)
2
),
c
(j)
0
(k(b
(j)
0
)
k1
b
(j)
3
+k(k 1)(b
(j)
0
)
k2
b
(j)
1
b
(j)
2
+
k(k 1)(k 2)
3!
(b
(j)
0
)
k3
(b
(j)
1
)
3
),
.
.
.
c
(j)
0
(k(b
(j)
0
)
k1
b
(j)
ml
+(b
(j)
0
, . . . , b
(j)
ml+1
)),
where is a homogeneous polynomial of degree k. It shows that [b
(j)
0
[ decreases
as fast as 1/[c
(j)
0
[
1
k
, [b
(j)
1
[ decreases as fast as (or faster than) 1/[c
(j)
0
[
1
k
, . . .,
[b
(j)
ml
[ decreases as fast as (or faster than) 1/[c
(j)
0
[
1
k
. This is a contradiction to
b
(j)
ml
b
ml
,= 0.
Case II: Suppose that b
0
,= 0. Let f
()
= a
0
t
n
+ + a
n
with n = mk.
Then c
(j)
0
goes to a
0
/b
k
0
,= 0, and c
l
goes to
(a
lm
c
0
b
k,nlm
c
1
b
k1,nlm
c
l1
b
kl+1,nlm
)/b
kl
0
,
where (

m
i=0
b
i
t
mi
)
p
=

n
j=0
b
p,j
t
j
. Thus, c
(j)
i
has the (bounded) limit for all
i and also the sequence h
(j)
has the limit in C
m
.
Since the Zariski closure and the closure by strong topology of a constructible
subset in a variety coincide, the above arguments prove the lemma.
Remark 1. (1) (n, n)

= C

(n, ) by sending (f, g) to (

b
0
a
0
, (f, g
b
0
a
0
f)),
where f =

n
i=0
a
i
t
ni
and g =

n
i=0
b
i
t
ni
. Here the equivalence symbol

=
means that there are regular maps between open subsets containing
n,n
and
C

(n, ) in C
n+1
C
n+1
such that those maps are inverses to each other
when restricted to (n, n) and C

(n, ). Later we will show that (n, n)

and (n, ) are subvarieties (but not closed ones).
(2) Similarly

(n)

= C (n, ) by sending (f, g) to (
b
0
a
0
, (f, g
b
0
a
0
f))
where g =

n
i=0
b
i
t
ni
. Note that b
0
here could possibly be zero.
In what follows, we will call the maximal dimension (or simply, dimension)
of a variety X to be the maximum among the dimensions of the irreducible
components of X.
Theorem 2. (1) The parameter space

(k
1
,...,k
q
)
in C
n+1
C
m+1
is isomorphic
to

(k
1
,...,k
q
)

=
q

i=1
C
k
i
C
3
of dimension (

k
i
) + 3.
176 DOSANG JOE AND HYUNGJU PARK
(2) (n, )

k,
Q
k
i
=n

k
.
(3)

(n)

= C

k,
Q
k
i
=n,

of the maximal dimension n + 4 if n 2

or 4 if n = 1.
Proof. The proof of statement (1) is proven in Lemma 2. The second statement
follows from Lemmas 2 and 3. For the proof of statement (3). Consider
: (C

C
n
) C
n+1
C (C

C
n
) C
n
sending (a, b) to (b
0
/a
0
, a, b
b
0
a
0
(a
1
, . . . , a
n
)). It has the inverse (c, a, d) =
(a, d +ca).
Note that the maximal dimension of (n, m) is (n/m) + m + 3, and the
maximal dimension of (n, n) is n + 4, and the maximal dimension of (n, )
is n + 3.
Example 1.
(8, 8) = C

R(8, )
= C

(C
4
C
2

C
2
C
2
C
2

C
2
C
4

C
8
) (1, ),
(8, 4) = (C
2
C
2
C
2

C
2
C
4
) (1, ),

(8) = C (C
4
C
2

C
2
C
2
C
2

C
2
C
4

C
8
) (1, ).
Note that Mor(C
1
, C
2
) has an action by A(C
2
) = GL
2
(C)C
2
induced
from its action of ane automorphisms of the target C
2
. Since A(C
2
) is a
connected algebraic group and

is a disjoint union of components of types,

we obtain the following denition.
Denition 5. An element of

k
is called a parametrized ane embedding of
line of type k. In general we dene the type of a closed parametrized ane
embedding of the line A
1
C
to the ane plane A
2
C
after choosing appropriate
ane coordinates of A
1
C
and A
2
C
.
3. Applications
3.1. Parameter spaces of (unparametrized ane) embeddings of line
Given (t) : C
1
C
2
, consider its unique projectivization :
(t) : C
1
C
2

: P
1
P
2
.
EMBEDDINGS OF LINE IN THE PLANE 177
In terms of homogeneous coordinates of P
1
, the projectivization can be writ-
ten as
([t : s])
= [a
0
t
n
+a
1
t
n1
s + +a
n
s
n
: b
0
t
m
s
nm
+b
1
t
m1
s
nm+1
+ +b
m
s
n
: s
n
],
where (t) = (a
0
t
n
+a
1
t
n1
+ +a
n
, b
0
t
m
+b
1
t
m1
+ +b
m
).
This rational projective curve of degree n meets the line at innity exactly
at one point, which is [a
0
: b
0
: 0] if n = m, [1 : 0 : 0] if n > m. It is obvious
that this point is the unique singular point of the given curve, if there is any.
Moreover the line l

at innity is a unique tangent line at this singular point

of the curve. The intersection multiplicity C l

is n. Let us denote that

cuspidal singularity of plane curves dened to be irreducible singular point.
Also unicuspidal curve meant to be a curve having only one singular point.
Denition 6. Denote by /(n) P(H
0
(P
2
, O(n)) the space of unicuspidal
rational curves of degree n with maximal tangent line.
Example 2 ([8]). There is a rational plane curve with one cusp and the unique
tangent line at the cuspidal point.
In P
2
, consider the plane curve C dened by the equation
(x
2
yz)
2
zy
3
= 0.
By a parametrization, we see that it is a rational curve having only one singular
point (0, 0, 1) with a unique tangent line:
x = t(t
2
1), y = (t
2
1)
2
, z = 1.
However the curve C (0, 0, 1) is not isomorphic to C: Otherwise, there is a
regular map : P
1
1 C P
1
. Now considering the degree of the
extension : P
1
P
1
, we obtain a contradiction.
Example 3. There is a rational unicuspidal quintic not having maximal tan-
gent line [6, 20]. The equation is the following:
(y x
2
)(y x
2
2xy) +y
5
= 0.
The ane plane curve given by the parametrization (t) : C
1
C
2
has the
dening equation
h(x, y) := Res
t
(f(t) x, g(t) y) PH
0
(P
2
, O
P
2(n)),
where Res
t
(f(t) x, g(t) y) is the resultant of f(t) x and g(t) y with
respect to t (c.f. Theorem 0.4 of [19]). Dene /(n, m) by
/(n, m) = h(x, y) = Res
t
(f(t) x, g(t) y)[(f, g) (n, m),
which is the locus of the curves corresponding to (n, m). We also dene /

k
to be the locus corresponding to

k
. The locus /(n, m) is viewed as a subset
of PH
0
(P
2
, O
P
2(n)). Let /
k
be the subset of /(n) consisting of those that
can be transformed to elements of /

k
by an automorphism of P
2
.
178 DOSANG JOE AND HYUNGJU PARK
Theorem 3. Each component /
k
is a rational variety of dimension 4 +

k
i
for n 3.
Proof. Note that there is a bration : /
k
F(3) where F(3) is a ag
manifold and assigns to a curve the ag of its singular point and the tangent
line at the singular point. The bration has a rational section by the following
construction. Near the ag
((0, 1, 0)), (0, 1, 0), (1, 0, 0)))
of a point in F(3), there is an ane chart of (, , ) C
3
parametrizing ags
(, 1, ), (, 1, ), (1, 0, ))). Pick any f(x, y, z) c
k
, then f(x, y, zx+
( )y) denes a rational section. Thus /
k
is a rational variety.
Corollary 1. /(n) =

k
/
k
is a quasi projective subvariety, and each com-
ponent /
k
is a rational variety of dimension 2 +

k
i
for n 2.
Example 4. Note that /(1) is P
2
minus one point. In general /
n
is an ane
bundle over P
1
with ber C

C
n
for n 2.
3.2. The degree of A(n)
Lemma 4. Let n be an nonnegative integer. The lattice volume (c.f. [7]) of
the convex hull of the exponents of
1, y
k
0
, x
1
, x
1
y
k
1
, x
2
, x
2
y
k
2
, . . . , x
n
, x
n
y
k
n
in Z
n+1
is

n
i=0
k
i
if k
i
are nonnegative integers not vanishing simultaneously.
Proof. When n = 0, it is clear. Assume that n > 0. By the Kouchnirenko
theorem [16], it is equal to the number N of common zeros of generic n + 2
equations

= 0, where runs over all exponents in the set of monomials

above.
Note that the equations can be rewritten as Ax = 0, where
x = (1, x
1
, . . . , x
n
)
transpose
and the entries a
ij
of the matrix A satises a
ij
C + Cy
k
j
if n j 0. By
taking the determinant of A, we conclude that the number N is less than or
equal to the degree of the determinant of A in y. Thus N

n
i=1
k
i
. However
we can achieve the upper bound by simply taking A as the maximal cyclic form

y
k
0
0 0 0 1
1 y
k
1
0 0 0
0 1 y
k
2
0
.
.
.
.
.
.
0 1 y
k
n

Theorem 4. The degree of (the maximal dimension component of) /(n) is
n(n+1)
2
1.
EMBEDDINGS OF LINE IN THE PLANE 179
Proof. Since we have
C /

n,1
=

i=0
a
i
(y x)
ni
x P(H
0
(P
2
, O(n)) [ a C
n
, C

,
it is enough to nd the degree of the closure of the image
: (C

)
n+2
P
(
n+2
2
)1
dened by sending (a
0
, . . . , a
n
, ) to (a
0
, a
0
, . . ., a
0

n
, a
1
, a
1
, . . ., a
1

n1
, . . .,
a
n2
, a
n2
, a
n2

2
, a
n1
, a
n1
1, a
n
).
Let be the composition

A of the ane transformation

A(z
0
, . . . , z
N
, z
N+1
) (z
0
, . . . , z
N
z
N+1
, z
N+1
)
of P
N+1
and the projection of P
N+1
to P
N
centered at (0, . . . , 0, 1) along the
subspace P(C
N
0). Consider
: (C

)
n+2
P
N
sending (a
0
, . . . , a
n
, ) to (a
0
, a
0
, . . ., a
0

n
, a
1
, a
1
, . . ., a
1

n1
, . . ., a
n2
,
a
n2
, a
n2

2
, a
n1
, a
n1
, a
n
, 1).
Note that is dened by monomials and = . Now applying Lemma 4,
one notes that the degree of the closure of Im is
n(n+1)
2
. Since this variety is
dened by the obvious quadratic equations, it is easy to check that the center
(0, . . . , 0, 1, 0, 1) is a nonsingular point, which, together with the injectivity of
, implies that the degree of Im is one less than the degree of Im.
Another Proof of Theorem 4. Denote the general hyperplane P(H
0
(P
2
, O
P
2(n))
by H
i
. To compute the intersection number
H
1
H
n+2
/(n),
we may assume that all the intersection points occur in
C /

n,1
=

i=0
a
i
(y x)
ni
x P(H
0
(P
2
, O(n))[a C
n
, C

.
The equations can be expressed by a matrix equation Ab = 0 where A has
entries a
ij
that are polynomials in with degree k
j
and j = 0, 1, . . . , n + 1.
Note that the rst column entries a
i0
are constants, and b is the transpose of
(1, b
1
, . . . , b
n+1
) := (1, a
n
, a
n1
, . . . , a
0
).
There is one more constraint that the constant term a
i0
equals the negative of
the leading coecient of a
i1
. By considering the determinant of A, it is clear
that there are at most (

n+1
i=0
k
i
) 1 solutions. However one can achieve the
180 DOSANG JOE AND HYUNGJU PARK
upper bound by taking the specialization of A to

1
k
1
0 0 1
0
k
1
1
0 0 1
0 1
k
2
0
.
.
.
.
.
.
0 1
k
n+1

.

The above theorem combined with Kleimans transversality theorem [14, 5]
applied to (P
2
)
n+2
with the group action (GL
2
(C))
n+2
implies the following.
Corollary 2. Let n 2. There are exactly (
n(n+1)
2
1) distinct plane curves
in /(n), passing through general dim/(n) points in P
2
.
3.3. The degree of M(3)
Corollary 3 ([15, 2]). Let /(3) be the space of cuspidal cubic curves. Then
the deg /(3) = 24.
Let (f(t), g(t)) be a smooth polynomial parametrization of a cubic curve.
Then its general form must be (a
0
t
3
+ a
1
t
2
+ a
2
t + a
3
, t). And their dening
equations are parametrized as follows:
a
0
y
3
+a
1
y
2
+a
2
y +a
3
= x a
0
y
3
+a
1
y
2
z +a
2
yz
2
+a
3
z
3
= xz
2
.
The dening equations near singular point [1 : 0 : 0] are
a
0
y
3
+a
1
y
2
z +a
2
yz
2
+a
3
z
3
z
2
= f(y, z).
Now we consider PGL(3, C) action on them. The local action changes the
singular point and the unique singular tangent:
f(y, z) f(y +b
1
, y +z +b
2
).
Hence we have local parametrization of cuspidal cubics as follows:
a
0
(y +b
1
)
3
+a
1
(y +b
1
)
2
(y +z +b
2
) +a
2
(y +b
1
)(y +z +b
2
)
2
+a
3
(y +z +b
2
)
3
(y +z +b
2
)
2
= 0.
We can use deg lex order y > z with descending order.
(a
0
+a
1
+a
2

2
+a
3

3
)y
3
+(a
1
+ 2a
2
+ 3a
3

2
)y
2
z
+(a
2
+ 3a
3
)yz
2
+a
3
z
3
+(3a
0
b
1
+a
1
b
2
+ 2a
1
b
1
+ 2a
2
b
2
+a
2
b
1

2
+ 3a
3
b
2

2
)y
2
+(2a
1
b
1
+ 2a
2
b
2
+ 2a
2
b
1
+ 6a
3
b
2
2)yz
+(a
2
b
1
+ 3a
3
b
2
1)z
2
+(3a
0
b
2
1
+a
1
b
2
1
+ 2a
1
b
1
b
2
+a
2
b
2
2
+ 2a
2
b
1
b
2
+ 3a
3
b
2
2
2b
2
)y
+(a
1
b
2
1
+ 2a
2
b
1
b
2
+ 3a
3
b
2
2
2b
2
)z
EMBEDDINGS OF LINE IN THE PLANE 181
+(a
0
b
3
1
+a
1
b
2
1
b
2
+a
2
b
1
b
2
2
+a
3
b
3
2
b
2
2
).
Dene polynomials P
i,j
(a
0
, a
1
, a
2
, a
3
, b
1
, b
2
, ) by
f(y +b
1
, y +z +b
2
) =

i+j3
P
i,j
(a
0
, a
1
, a
2
, a
3
, b
1
, b
2
, )y
i
z
j
.
Hence the projective closure, Im(), of the image of the map
: C

C
6
C
9
CP
9
is the parameter space of cuspidal cubic curves in CP
2
. It is an explicit rational
parametrization of the parameter space of cuspidal cubic curves.
An explicit computation with the computer algebra package Singular [10]
shows that the degree of Im() is 24. Moreover another numerical invariant
which counts the number of cuspidal cubics with a xed singular point can be
computed by computing the degree of the subvariety Im((b
1
= b
2
= 0)). A
computation shows that this number is 2. We also have checked the degree of
the subvariety Im((b
1
= 0). This number counts the cuspidal cubics whose
singular points lie on a xed line. Our computation shows it is 12.
Remark 2. (1) These numerical invariants have been known for years [18,
15, 2, 11, 12], going all the way back to a very old Enriques formula.
(2) After writing up this paper, we learned that the degree of /(d) had
been independently obtained in a recent paper [13] as an example. The
approaches, however, appear to be substantially dierent.
(3) Note that our method produces not only the degree of the parameter
space /(d) but also an explicit parametrization of /(d) itself.
Acknowledgements. The authors are grateful to Ziv Ran for helpful com-
ments. Dosang Joe thanks the sta at Korea Institute for Advanced Study,
for providing him with pleasant research atmosphere during his visits as an
associate member. The second author was supported by KRF Grant #2005-
070-C00004.
References
[1] S. Abhyankar and T. Moh, Embeddings of the line in the plane, J. Reine Angew. Math.
276 (1975), 148166.
[2] P. Alu, The enumerative geometry of plane cubics. II. Nodal and cuspidal cubics,
Math. Ann. 289 (1991), no. 4, 543572.
[3] E. Brieskorn and H. Knorrer, Plane Algebraic Curves, Birkhauser Verlag, Basel, 1986.
[4] H. Flenner and M. Zaidenberg, On a class of rational cuspidal plane curves, Manuscripta
Math. 89 (1996), no. 4, 439459.
[5] W. Fulton, Intersection Theory, Springer-Verlag, Berlin, 1984.
[6] M. H. Gizatullin and V. I. Danilov, Automorphisms of ane surfaces. I, Izv. Akad.
Nauk SSSR Ser. Mat. 39 (1975), no. 3, 523565, 703.
[7] I. M. Gelfand, M. M. Kapranov, and A. V. Zelevinsky, Discriminants, Resultants
and Multidimensional Determinants, Mathematics: Theory & Applications. Birkhauser,
1994.
182 DOSANG JOE AND HYUNGJU PARK
[8] C. G. Gibson, Elementary Geometry of Algebraic Curves: An Undergraduate Introduc-
tion, Cambridge University Press, Cambridge, 1998.
[9] G.-M. Greuel, C. Lossen, and E. Shustin, Introduction to Singularities and Deforma-
tions, Springer, 2007.
[10] G.-M. Greuel, G. Pster, and H. Sch onemann, Singular 3.0. A Computer Algebra Sys-
tem for Polynomial Computations, Centre for Computer Algebra, University of Kaiser-
slautern, 2005. http://www.singular.uni-kl.de.
[11] M. Kazaryan, Multisingularities, cobordisms, and enumerative geometry, Russian Math.
Surveys 58 (2003), no. 4, 665724
[12] , Characteristic Classes in Singularity Theory, Doctoral Dissertation (habilita-
tion thesis), Steklov Math. Inst., 2003.
[13] D. Kerner, Enumeration of unisingular algebraic curves, Math. AG/0407358, 2004.
[14] S. Kleiman, The transversality of a general translate, Compositio Math. 28 (1974),
287297.
[15] S. Kleiman and R. Speiser, Enumerative geometry of cuspidal plane cubics, Proceedings
of the 1984 Vancouver conference in algebraic geometry, 227268, CMS Conf. Proc., 6,
Amer. Math. Soc., Providence, RI, 1986.
[16] A. G. Kouchnirenko, Poly`edres de newton et nombres de milnor, Invent. Math. 32
(1976), no. 1, 131.
[17] A. van den Essen, Polynomial Automorphisms, Birkhauser, Basel, 2000.
[18] I. Vainsencher, Counting divisors with prescribed singularities, Trans. Amer. Math. Soc.
267 (1981), no. 2, 399422.
[19] A. van den Essen and J. Yu, The D-resultant, singularities and the degree of unfaith-
fulness, Proc. Amer. Math. Soc. 125 (1997), no. 3, 689695.
[20] H. Yoshihara, On plane rational curves, Proc. Japan Acad. Ser. A Math. Sci. 55 (1979),
no. 4, 152155.
Dosang Joe
Department of Mathematics Education
Konkuk University
Seoul 143-701, Korea
E-mail address: [email protected]
Hyungju Park
School of Computational Sciences
Korea Institute for Advanced Study
Seoul 130-722, Korea
E-mail address: [email protected]