You are on page 1of 15

TORIC RESOLUTIONS OF STRONGLY MIXED WEIGHTED HOMOGENEOUS


POLYNOMIAL GERMS OF TYPE J10

SACHIKO SAITO

In Commemoration of Professor Goo Ishikawa’s all years of hard work


arXiv:2402.15660v1 [math.AG] 24 Feb 2024

Abstract. We consider toric resolutions of some strongly mixed weighted homogeneous polynomials of

type J10 . We show that the strongly mixed weighted homogeneous polynomial f := f2,2,1,2,1,4 (k = 3)
(see §3) has no mixed critical points on C∗2 (Lemma 14), and moreover, show that the strict transform Ṽ
of the mixed hypersurface singularity V := f −1 (0) via the toric modification π̂ : X → C2 , where we set
f := f2,2,1,2,1,4 (k = 3), is not only a real analytic manifold outside of Ṽ ∩ π̂ −1 (0) but also a real analytic
manifold as a germ of Ṽ ∩ π̂ −1 (0) (Theorem 15).

1. Introduction
The famous Hilbert’s 16th problem (see [1], [18], [2] for example) asks the topology of nonsingular real
algebraic curves of degree 6 on the real projective plane RP 2 and nonsingular real algebraic surfaces of
degree 4 in the real projective space RP 3 . D.A. Gudkov completed the isotopic classification (the 56 isotopy
types) of nonsingular real algebraic curves of degree 6 on RP 2 in 1971 (cf. [1]). Especially he proved that
the number of the isotopy types of nonsingular real algebraic M -curves of degree 6 on RP 2 is three. See
Figure 1 below:

Figure 1. The isotopy types of nonsingular real algebraic M -curves of degree 6 on RP 2

In the early 1970s, V. A. Rokhlin ([10], [11], [12]) reproved this Gudkov’s result by differential topology,
which includes Smith theory on group actions, and some index theorems on 4-dimensional manifolds.
From complex geometric point of view, double coverings of CP 2 branched along nonsingular algebraic
sextic curves and nonsingular algebraic quartic surfaces in CP 3 are K3 surfaces. V. M. Kharlamov ([3])
started applying the K3 surface theory and the deformation theory to Hilbert’s 16th problem and obtained
many remarkable results. Subsequently to Kharlamov’s approaches, V.V. Nikulin ([4]) reproved Gudkov’s
isotopic classification of nonsingular real sextic curves by using the moduli spaces (period domains, or
Hilbert schemes) of real projective K3 surfaces and his lattice (integral symmetric bilinear forms) theory
in 1979.

2020 Mathematics Subject Classification. 14P05, 32S45.


Key words and phrases. J10 -singularity, mixed weighted homogeneous polynomial, Newton non-degenerate, toric modifica-
tion, toric resolution.
1
In the 1980s, O. Ya. Viro ([15]) reconstructed Gudkov’s 56 isotopy types of nonsingular real algebraic
curves of degree 6 by his patchworking method (see [17] for the details), where he used the real weighted

homogeneous polynomials of type J10 with two variables:
f (z1 , z2 ) := (z2 − z12 )(z2 − 2z12 )(z2 − kz12 ),
where k is a real number and k > 2, and their nonsingular perturbations. He got all the nonsingular

perturbations of the weighted homogeneous polynomials of type J10 ([16]). Roughly speaking, he patch-

worked two appropriate nonsingular perturbations of some polynomials of type J10 , and projectified the
patchworked polynomial of degree 6. Nonsingular perturbations whose zero sets (in R2 ) have many con-
nected components are very useful for constructions of nonsingular real algebraic M -curves of degree 6 on

RP 2 . Thus, the weighted homogeneous polynomials of type J10 were found to be very important for the
constructions of nonsingular real algebraic curves of degree 6 on RP 2 .

In this paper we consider mixed weighted homogeneous polynomials of type J10 , especially, strongly

mixed weighted homogeneous polynomials of type J10 . It would be expected that there exist some relations
between the topology of the real parts of (toric) resolutions of mixed weighted homogeneous polynomials
with real coefficients and that of nonsingular perturbations of real weighted homogeneous polynomials. In
this paper we eventually show that the strongly mixed weighted homogeneous polynomial (the polynomial
(3.3) in §3):
f := f2,2,1,2,1,4 (k = 3)
has no mixed critical points on C∗ 2 (Lemma 14), and moreover, show that the strict transform Ṽ of
the mixed hypersurface singularity V := f −1 (0) via the toric modification π̂ : X → C2 , where we set
f := f2,2,1,2,1,4 (k = 3), is not only a real analytic manifold outside of Ṽ ∩ π̂ −1 (0) but also a real analytic
manifold as a germ of Ṽ ∩ π̂ −1 (0) (Theorem 15).

2. Basic terminology about mixed polynomials


Here we recall basic terminology about mixed polynomials.

2.1. Mixed functions and their radial Newton polyhedrons. Let U be a neighborhood of 0 in
Cn with Ū = U , where z̄ stands for the complex conjugate (z¯1 , . . . , z¯n ) of z = (z1 , . . . , zn ) ∈ Cn . Let
F (z, w) be a complex valued holomorphic function on U × U with complex 2n variables. We assume that
F (0, 0) = 0 We define f : U → C by
f (z, z̄) := F (z, z̄).
We call such a f (or f (z, z̄)) a mixed analytic function (or mixed function) on U . Let F (z, w) =
ν µ
P
ν,µ cν,µ z w be the Taylor expansion of F at (0, 0), where ν = (ν1 , . . . , νn ), µ = (µ1 , . . . , µn ), νi ≥
0, µj ≥ 0, z ν := z1ν1 · · · znνn , wµ := w1µ1 · · · wnµn . Then we have
X
f (z, z̄) = cν,µ z ν z̄ µ . (2.1)
ν,µ

Note that the coefficients of the Taylor expansion (2.1) of a mixed function f are unique. We call f (z, z̄)
a mixed polynomial if the number of monomials cν,µ z ν z̄ µ , cν,µ ̸= 0 is finite.
Definition 1. Let f (z, z̄) be a mixed function on U (⊂ Cn ). We say a = (a1 , . . . , an ) ∈ U is a mixed
critical point (or a mixed singular point) of f if the rank of the differential map
(df )a : Ta Cn → Tf (a) C ∼
= Tf (a) R2
is less than 2. We say a ∈ U is a mixed regular point of f if it is not a mixed critical point of f .
2
We set K+n := {(x , . . . , x ) ∈ K n | x ≥ 0 for every i}, where K = R or Z.
1 n i X
n
For the germ (f, 0) (at 0 ∈ C ) of a mixed function f (z, z̄) = cν,µ z ν z̄ µ ,
ν,µ

Γ+ (f )
[
is defined to be the convex hull of the set (ν + µ) + Rn+ . We call Γ+ (f ) the (radial) Newton polyhedron
cν,µ ̸=0
of (f, 0).
For a “weight vector” P = t (p1 , . . . , pn ) (̸= 0) ∈ Zn+ , we define d(P ) to be the minimum value of the
Xn
linear function P : Γ+ (f ) → R, P (ξ) := pj ξj , where ξ = (ξ1 , . . . , ξn ) ∈ Γ+ (f ). We set
j=1

∆(P ) := {ξ ∈ Γ+ (f ) | P (ξ) = d(P )},


which we call a face of Γ+ (f ). Note that ∆(P ) ̸= ∅ by its definition.
We say a weight vector P = t (p1 , . . . , pn ) ∈ Zn+ is strictly positive (P ≫ 0) if pi > 0 for every i (=
1, . . . , n). Note that a face ∆ of Γ+ (f ) is compact if and only if ∆ = ∆(P ) for some strictly positive weight
vector P . For a compact face ∆(P ), we define
X
f∆(P ) (z) (or fP (z) ) := cν,µ z ν z̄ µ ,
µ+ν∈∆(P )

which we call a face function (or face polynomial) of a mixed function germ (f, 0) ([9], p.78).
Definition 2 ([9], p.79). Let (f, 0) be the germ of a mixed function f at 0 ∈ Cn . For I ⊂ {1, . . . , n}, f I
denotes the restriction of f on the coordinate subspace CI := {z | zj = 0, j ∈ / I}. A mixed function germ
(f, 0) is called convenient if f I ̸≡ 0 for every I ⊂ {1, 2, · · · , n} with I ̸= ∅.

2.2. Radially and polar weighted homogeneous polynomials.


Definition 3 ([9], p.182; see also [8]).
• A mixed polynomial f (z, z̄) = ν,µ cν,µ z ν z̄ µ is called radially weighted homogeneous if there exists
P

a weight vector P = t (p1 , . . . , pn ) (̸= 0) ∈ Zn+ and a positive integer dr (> 0) such that
n
X
cν,µ ̸= 0 =⇒ P (ν + µ) = pi (νi + µi ) = dr .
i=1
We call dr the radial degree of fP , and define rdegP f := dr .
• A mixed polynomial f (z, z̄) = ν,µ cν,µ z ν z̄ µ is called polar weighted homogeneous if there exists a
weight vector Q = t (q1 , . . . , qn ) (̸= 0) ∈ Zn and an integer dp (> 0, 0 or < 0) such that
n
X
cν,µ ̸= 0 =⇒ Q(ν − µ) = qi (νi − µi ) = dp .
i=1
We call dp the polar degree of f , and define pdegQ f := dp .
Every face function f∆(P ) (z), where P is strictly positive, of a mixed function germ (f, 0) is a radially
weighted homogeneous polynomial of positive radial degree d(P ) with respect to the weight vector P .
Example 4 ([9], Example 9.17). f (z, z̄) := z12 z¯1 − z2 z¯2 2 is radially weighted homogeneous with respect to
P = t (1, 1) and polar weighted homogeneous with respect to Q = t (1, −1).

For radially and polar weighted homogeneous polynomials, we have the following basic facts:
3
Lemma 5 (cf.[6], [7]). Let f (z, z̄) be a mixed polynomial. We have the following.
• Let P = t (p1 , . . . , pn ) (̸= 0) ∈ Zn+ be a weight vector and dr be a positive integer. For a positive
real number t and z ∈ Cn , we define
t ◦ z := (tp1 z1 , . . . , tpn zn ).
Then, f (t◦z) = tdr f (z) for every t > 0 and every z ∈ Cn if and only if “cν,µ ̸= 0 ⇒ P (ν +µ) = dr ”
holds.
• Let Q = t (q1 , . . . , qn ) (̸= 0) ∈ Rn be a weight vector and dp be an integer. For a real number θ and
z ∈ Cn , we define
θ ◦ z := (eiq1 θ z1 , . . . , eiqn θ zn ).
Then, f (θ ◦ z) = eidp θ f (z) for every θ ∈ R and every z ∈ Cn if and only if “cν,µ ̸= 0 ⇒ Q(ν − µ) =
dp ” holds. 2
Proposition 6 (Euler equalities, [9]). Let f (z, z) = ν,µ cν,µ z ν z µ be a mixed polynomial.
P

(R): If f (z, z) is a radially weighted homogeneous polynomial of radial degree dr (> 0) with respect to a
weight vector P = t(p1 , . . . , pn ), then we have
n  
X ∂f ∂f
p j zj + zj = dr f (z, z). (2.2)
∂zj ∂zj
j=1

(P): If f (z, z) is a polar weighted homogeneous polynomial of polar degree dp with respect to a weight
vector Q = t(q1 , . . . , qn ), then we have
n  
X ∂f ∂f
qj z j − zj = dp f (z, z). 2 (2.3)
∂zj ∂zj
j=1

Definition 7 (mixed weighted homogeneous polynomial, [9], pp.182–184). We say a mixed polynomial
f (z, z̄) is a mixed weighted homogeneous polynomial if it is both radially and polar weighted homogeneous.
Here, the corresponding weight vectors P and Q are possibly different. We say a mixed weighted ho-
mogeneous polynomial f (z, z̄) is a strongly mixed weighted homogeneous polynomial if f is radially and
polar weighted homogeneous with respect to the same weight vector P . Furthermore, a mixed weighted
homogeneous polynomial f is called a strongly polar positive mixed weighted homogeneous polynomial with
respect to a weight vector P if f is radially and polar weighted homogeneous with respect to the same
weight vector P and pdegP f > 0.
Definition 8 (cf.[9],Definition 9.18; [8],p.174). Let (f, 0) be a mixed function germ at 0 ∈ Cn . The germ
(f, 0) of a mixed function f (z, z̄) at 0 ∈ Cn is defined to be of strongly polar positive mixed weighted
homogeneous face type if for every compact face 1 ∆, the face function f∆ (z, z̄) is a strongly polar positive
mixed weighted homogeneous polynomial (Definition 7) with respect to some strictly positive weight vector
P with ∆ = ∆(P ).

1In [8], a convenient (Definition 2) mixed function germ (f (z, z̄), 0) is defined to be of strongly polar positive mixed weighted
homogeneous face type if the face function f∆ (z, z̄) is a strongly polar positive mixed weighted homogeneous polynomial for
every (n − 1)-dimensional face. Besed on this definition, Proposition 10 of [8] asserts that the face function f∆(P ) is also a
strongly polar positive mixed weighted homogeneous polynomial with respect to P for every weight vector P when (f, 0) is a
convenient mixed function germ of strongly polar positive mixed weighted homogeneous face type in the sense of [8].
4
2.3. Newton non-degeneracy and Strong Newton non-degeneracy.
Definition 9 ([7], p.6, Definition 3; [9], p.80 and pp.181–182). Let (f, 0) be the germ of a mixed function
f at 0 ∈ Cn .
(1) We say (f, 0) is Newton non-degenerate over a compact face ∆ if 0 is not a mixed critical value of
−1
the face function f∆ : C∗ n → C. (In particular, if f∆ (0) ∩ C∗ n = ∅, then 0 is not a mixed critical
∗ n
value of the face function f∆ : C → C.)
(2) Let ∆ be a compact face with dim ∆ ≥ 1. We say (f, 0) is strongly Newton non-degenerate over ∆
if the face function f∆ : C∗ n → C has no mixed critical points and f∆ : C∗ n → C is surjective onto
C.
(3) Let ∆ be a compact face with dim ∆ = 0, that is, ∆ is a vertex of Γ+ (f ). We say (f, 0) is strongly
Newton non-degenerate over ∆ if the face function f∆ : C∗ n → C has no mixed critical points.
Definition 10 ([9], p.80 and p.182). We say the germ (f, 0) of a mixed function f at 0 ∈ Cn is Newton non-
degenerate (respectively, strongly Newton non-degenerate) if (f, 0) is Newton non-degenerate (respectively,
strongly Newton non-degenerate) over every compact face ∆.
Example 11 ([9], 8.3.1). The germ of the mixed homogeneous polynomial ρ(z, z) = nj=1 zj zj = nj=1 |zj |2
P P
of degree 2 at 0 ∈ Cn is Newton non-degenerate, but not strongly Newton non-degenerate. Actually, for
every compact face ∆, we have ρ∆ −1 (0) ∩ C∗n = ∅. However, since ρ∆ (Cn ) ⊂ R, every point in C∗n is a
mixed critical point. Hence, ρ is not strongly Newton non-degenerate.
Note that if f −1 (0) ∩ C∗ n ̸= ∅, then f : C∗ n → C is surjective. Namely, in this case, Newton non-
degeneracy over a compact face ∆(P ) implies strong Newton non-degeneracy over ∆(P ). For more precise
statements, see Proposition 12 below:
Proposition 12 (Remark 4 in [7]; and also [13], [14]).
• Let f (z) be a holomorphic weighted homogeneous polynomial of positive degree with respect to a
strictly positive weight vector P . (Then, f = f∆(P ) .)
(i) Suppose that f is Newton non-degenerate over ∆(P ), namely, 0 is not a critical value of
f∆(P ) = f : C∗ n → C. Then f∆(P ) = f : C∗ n → C has no critical point. Hence, with (ii) below, f
is strongly Newton non-degenerate over ∆(P ).
(ii) Suppose that dim ∆(P ) ≥ 1, namely, f = f∆(P ) has at least two monomials. Then f∆(P ) =
f : C∗ n → C is surjective.
• Let f (z, z̄) be a mixed weighted homogeneous polynomial (Definition 7) with respect to a radial
weight vector P (≫ 0) and a polar weight vector Q.
(iii) Suppose that (f, 0) is Newton non-degenerate over a compact face ∆(P ) and the face func-
tion f∆(P ) = f is a polar weighted homogeneous polynomial of non-zero polar degree with respect to
the polar weight vector Q. Then f : C∗ n → C has no mixed critical points.
(iv) In addition to (iii), we assume that f −1 (0) ∩ C∗ n ̸= ∅. Then f : C∗ n → C is surjective.
Hence, with (iii), in this case, Newton non-degeneracy over a compact face ∆(P ) implies strong
Newton non-degeneracy over ∆(P ). 2


3. Mixed weighted homogeneous polynomials of type J10

We now consider the weighted homogeneous polynomials of type J10 :
f (z1 , z2 ) := (z2 − z12 )(z2 − 2z12 )(z2 − kz12 )
(3.1)
= z23 − (k + 3)z12 z22 + (3k + 2)z14 z2 − 2kz16 ,
5
where we assume that k ∈ R and k > 2. Note that these polynomials are weighted homogeneous with
respect to the weight vector
P := t (1, 2)
of degree 6 with real coefficients.
Based on the weighted homogeneous polynomials (3.1), let us consider the radially weighted homogeneous
polynomials with respect to the radial weight vector P = t (1, 2) of radial degree 6:
fa,b,c,d,e,f := z2a z̄23−a − (k + 3)z1b z̄12−b z2c z̄22−c + (3k + 2)z1d z̄14−d z2e z̄21−e − 2kz1f z̄16−f , (3.2)
where a, b, c, d, e, f are integers with 0 ≤ a ≤ 3, 0 ≤ b, c ≤ 2, 0 ≤ d ≤ 4, 0 ≤ e ≤ 1 and 0 ≤ f ≤ 6.
Obviously fa,b,c,d,e,f are convenient (Definition 2).
3.1. The radial Newton polyhedron, the dual Newton diagram and the regular simplicial cone
subdivision. The radial Newton polyhedron of the mixed function germ fa,b,c,d,e,f at 0 is as follows:

ν2 + µ2
6
4

3 s
HHH
H
s
H
H
HH
2 H
H
HH
HH
HH
sH
H
H
HH
1 H
H
HH
HHHH
Hs
HH
HH
0 HH - ν1 + µ1
0 1 2 3 4 5 6 7

Figure 2. The radial Newton polyhedron of fa,b,c,d,e,f

The dual Newton diagram Γ∗ (f ) of the mixed function germ f := fa,b,c,d,e,f at 0 is as in the Figure 3,
where we set E1 = t (1, 0), P = t (1, 2), E2 = t (0, 1).

E2

E1

Figure 3. The dual Newton diagram Γ∗ (f )

Adding the vertex S := t (1, 1) to Γ∗ (f ), we obtain a regular simplicial cone subdivision (see [9], §5.6;
or [5] for the definition) Σ∗ of NR+ which is admissible for Γ∗ (f ).
6
E2

P S

E1 S P E2

E1

Figure 4. The regular simplicial subdivision Σ∗ which is admissible for Γ∗ (f ).

Definition 13 ([9], p.98). We set Ej := t (0, . . . , 0, 1, 0, . . . , 0) (∈ Zn ) for every j (= 1, . . . , n), and


| {z }
j

EJ := Cone(Ej1 , . . . , Ejk ),
where J := {j1 , . . . , jk } (⫋ {1, . . . , n}). Let (f, 0) be the germ of a mixed function f at 0 ∈ Cn . For
I ⊂ {1, . . . , n}, f I denotes the restriction of f on the coordinate subspace CI := {z | zj = 0, j ∈
/ I}. Let
Σ∗ be a regular simplicial cone subdivision which is admissible for Γ∗ (f ). We say Σ∗ is convenient if for
every I with f I ̸≡ 0, the cone EI c is contained in Σ∗ .
Then our Σ∗ (Figure 4) is convenient in the sense of Definition 13, namely, the cone EI c is contained in
Σ∗ for every I with f I ̸≡ 0.

We moreover set T := t (1, 3). Then, all the compact faces of the radial Newton polyhedron of the germ
(fa,b,c,d,e,f , 0) are
∆(P ), ∆(S), and ∆(T ),
and hence, the face functions of (f := fa,b,c,d,e,f , 0) are
fP = fa,b,c,d,e,f , fS = z2a z̄23−a and fT = −2kz1f z̄16−f .


3.2. Strongly mixed weighted homogeneous polynomials of type J10 . The radially weighted ho-
mogeneous polynomials fa,b,c,d,e,f are strongly mixed weighted homogeneous with respect to P (Definition
7) if and only if (a, b, c, d, e, f) are in the following 5 cases:
a b c d e f polar degree
I 3 2 2 4 1 6 6 holomorphic case
II 2 2 1 4 0 4 2
III 2 0 2 4 0 4 2
IV 2 2 1 2 1 4 2
V 2 0 2 2 1 4 2
In the cases II ∼ V (a = 2, f = 4), the radial degrees of f := f2,b,c,d,e,4 are 6, and the polar degrees of
those are 2. For their 0-dimensional face functions fS = z22 z̄2 and fT = −2kz14 z̄12 , we have
rdegS fS = 1 · 0 + 1 · (2 + 1) = 3, pdegS fS = 1 · 0 + 1 · (2 − 1) = 1,
rdegT fT = 1 · (4 + 2) + 3 · 0 = 6, and pdegT fT = 1 · (4 − 2) + 3 · 0 = 2.
Hence, remark that the germs (f2,b,c,d,e,4 , 0) (in the cases II ∼ V) are of strongly polar positive mixed
weighted homogeneous face type (Definition 8).
7
3.3. Strong Newton non-degeneracy. We now argue about the Newton degeneracy of the strongly
mixed weighted homogeneous polynomials f2,b,c,d,e,4 in the cases II ∼ V. Especially let us investigate the
case IV with k = 3:
f2,2,1,2,1,4 (k = 3) = z22 z̄2 − 6z12 z2 z̄2 + 11z12 z̄12 z2 − 6z14 z̄12 . (3.3)
Lemma 14. We have the following.
(1) f := f2,2,1,2,1,4 : (k = 3) has no mixed critical points on C∗ 2 .
(2) f : C∗ 2 → C is surjective.
(3) The 0-dimensional face functions fS , fT of (f, 0) also have has no mixed critical points on C∗ 2 .
Thus, the germ (f, 0) is strongly Newton non-degenerate (Definition 10).
Proof. We first prove the assertion (1). We have
∂f
= −12z1 z2 z̄2 + 22z1 z̄12 z2 − 24z13 z̄12 ,
∂z1
∂f
= 22z12 z̄1 z2 − 12z14 z̄1
∂ z̄1
= 2z12 z̄1 (11z2 − 6z12 ),
∂f (3.4)
= 2z2 z̄2 − 6z12 z̄2 + 11z12 z̄12
∂z2
= 2|z2 |2 + z12 (11z̄12 − 6z̄2 ),
∂f
= z22 − 6z12 z2
∂ z̄2
= z2 (z2 − 6z12 ).
By Lemma 5, we have f (tz1 , t2 z2 ) = r6 e2iθ f (z1 , z2 ) for every t = reiθ (∈ C∗ ), r > 0, θ ∈ R since dr = 6
(radial degree) and dp = 2 (polar degree). Hence, if f (z1 , z2 ) = 0, then we have
f (tz1 , t2 z2 ) = 0 (3.5)
for all t ∈ C∗ . By Proposition 6, we have
6+2 ∂f ∂f
f (z1 , z2 ) = z1 + 2z2
2 ∂z1 ∂z2
and
6−2 ∂f ∂f
f (z1 , z2 ) = z̄1 + 2z̄2 .
2 ∂ z̄1 ∂ z̄2
Suppose that a = (a1 , a2 ) ∈ C∗ 2 is a mixed critical point of f . Recall that the following two conditions
are equivalent (Proposition 1 in [6]):
(1) a = (a1 , a2 ) (∈ C2 ) is a mixed critical point of f .
(2) There exists a complex number α with |α| = 1 which satisfies
   
∂f ∂f ∂f ∂f
(a), (a) = α (a), (a) .
z1 z2 ∂ z¯1 ∂ z¯2
We have
∂f ∂f
4f (a1 , a2 ) = ā1 ∂z 1
(a1 , a2 ) + 2ā2 ∂z 2
(a1 , a2 )
= ā1 α ∂∂f ∂f
z¯1 (a1 , a2 ) + 2ā2 α ∂ z¯2 (a1 , a2 )

= α(ā1 ∂∂f ∂f
z¯1 (a1 , a2 ) + 2ā2 ∂ z¯2 (a1 , a2 ))
and
∂f ∂f
2f (a1 , a2 ) = ā1 (a1 , a2 ) + 2ā2 (a1 , a2 ).
∂ z̄1 ∂ z̄2
8
Thus we have 4f (a1 , a2 ) = 2αf (a1 , a2 ) and 4|f (a1 , a2 )| = 2|α||f (a1 , a2 )| = 2|f (a1 , a2 )|. Hence, it is con-
cluded that f (a1 , a2 ) = 0. Thus we see that every mixed critical point of f is a zero of that.
Now we show that f has no mixed critical points on C∗ 2 . It is sufficient to prove that f has no mixed
critical points in f −1 (0) ∩ C∗ 2 , namely, f is Newton non-degenerate over ∆(P ).
∂f
Suppose that (a1 , a2 ) ∈ C∗ 2 and f (a1 , a2 ) = 0. If | ∂z 1
̸ | ∂∂f
(a1 , a2 )| = ∂f
z¯1 (a1 , a2 )| or | ∂z2 (a1 , a2 )| =
̸
∂f ∂f
| ∂ z¯2 (a1 , a2 )|, then (a1 , a2 ) is not a mixed critical point. Thus we may moreover suppose that | ∂z1 (a1 , a2 )| =
| ∂∂f ∂f ∂f
z¯1 (a1 , a2 )| and | ∂z2 (a1 , a2 )| = | ∂ z¯2 (a1 , a2 )|. Then we have

∂f ∂f ∂f ∂f
(a1 , a2 ) = α1 (a1 , a2 ) and (a1 , a2 ) = α2 (a1 , a2 ) (3.6)
∂z1 ∂ z¯1 ∂z2 ∂ z¯2
for some α1 , α2 ∈ C with |α1 | = |α2 | = 1. On the other hand, by (3.5), we have f (ta1 , t2 a2 ) = 0 for all
t ∈ C∗ . We set c(t) = (c1 (t), c2 (t)) := (ta1 , t2 a2 ). Then, by the chain rules for Wirtinger derivative, we
have
d P2 ∂f dcj P2 ∂f dcj
dt (f ◦ c)(t) = j=1 ( ∂zj ◦ c)(t) dt (t) + j=1 ( ∂ z̄j ◦ c)(t) dt (t)
P2 ∂f dcj
= j=1 ( ∂zj ◦ c)(t) dt (t).
Hence, we have
∂f dc1 ∂f dc2
(ta1 , t2 a2 ) (t) + (ta1 , t2 a2 ) (t) = 0.
∂z1 dt ∂z2 dt
Setting t = 1, we have
∂f ∂f
a1 (a1 , a2 ) + 2a2 (a1 , a2 ) = 0.
∂z1 ∂z2
By (3.6), we have
∂f ∂f
ā1 α1 (a1 , a2 ) + 2ā2 α2 (a1 , a2 ) = 0. (3.7)
∂ z¯1 ∂ z¯2
We now show that such (a1 , a2 ) is not a mixed critical point. It is sufficient to prove that α1 ̸= α2 . Suppose
that α1 = α2 . We have
∂f ∂f
ā1 (a1 , a2 ) + 2ā2 (a1 , a2 ) = 0, (3.8)
∂ z¯1 ∂ z¯2
that is,
ā1 a21 ā1 (11a2 − 6a21 ) + ā2 a2 (a2 − 6a21 ) = 0.
We have
|a1 |4 (11a2 − 6a21 ) + |a2 |2 (a2 − 6a21 ) = 0, (3.9)
11|a1 | a2 − 6|a1 |4 a21 + |a2 |2 a2 − 6|a2 |2 a21 =
4
0,
(11|a1 |4 + |a2 |2 )a2 = 6(|a1 |4 + |a2 |2 )a21 ,
and
6(|a1 |4 + |a2 |2 ) 2
a2 = a . (3.10)
11|a1 |4 + |a2 |2 1
This means that a2 and a21 have the same direction in the complex plane C. Note that by (3.9), we also
have
|a2 |2
11a2 − 6a21 = − (a2 − 6a21 ) (3.11)
|a1 |4
and
−10|a1 |4
a2 − 6a21 = a2 . (3.12)
|a1 |4 + |a2 |2
9
By the calculations (3.4) of partial derivatives, we have

∂f −10|a1 |4 2
(a1 , a2 ) = a2 (a2 − 6a21 ) = a .
∂ z̄2 |a1 |4 + |a2 |2 2
On the other hand, we have
∂f
∂z2 (a1 , a2 ) = 2a2 ā2 − 6a21 ā2 + 11a21 ā21
= a2 ā2 + a2 ā2 − 6a21 ā2 + 11a21 ā21
= a2 ā2 + (a2 − 6a21 )ā2 + 11a21 ā21
−10|a1 |4
= a2 ā2 + a ā + 11a21 ā21
|a1 |4 +|a2 |2 2 2
4
= |a2 |2 + |a−10|a 4
1|
1 | +|a2 |
2
2 |a2 | + 11|a1 |
4

−10|a1 |4 +|a1 |4 +|a2 |2


= |a1 |4 +|a2 |2
|a2 |2 + 11|a1 |4
−9|a1 |4 +|a2 |2
= |a1 |4 +|a2 |2
|a2 |2 + 11|a1 |4
−9|a1 |4 |a2 |2 +|a2 |4 +11|a1 |8 +11|a1 |4 |a2 |2
= |a1 |4 +|a2 |2
2|a1 |4 |a2 |2 +|a2 |4 +11|a1 |8
= |a1 |4 +|a2 |2
> 0.

By the latter equality of (3.6), we have

2|a1 |4 |a2 |2 + |a2 |4 + 11|a1 |8 −10|a1 |4


= α2 a22 .
|a1 |4 + |a2 |2 |a1 |4 + |a2 |2
a2
Since α2 a22 is a negative real number, we have α2 |a22|2 = −1. Namely, we have

|a2 |2
α1 = α2 = − .
a22
Then, on the other hand, we have

∂f
(a1 , a2 ) = −12ā1 ā2 a2 + 22ā1 a21 ā2 − 24ā31 a21
∂z1
and
α1 ∂∂f 2 2
z̄1 (a1 , a2 ) = 2α1 a1 ā1 (11a2 − 6a1 )
2
= −2 |aa22| a21 ā1 (11a2 − 6a21 )
2
= −2 āa22 a21 ā1 (11a2 − 6a21 ).
Thus, by the first equality of (3.6), we have
ā2 2
−12ā1 ā2 a2 + 22ā1 a21 ā2 − 24ā31 a21 = −2 a ā1 (11a2 − 6a21 ),
a2 1

ā2 2
6ā1 ā2 a2 − 11ā1 a21 ā2 + 12ā31 a21 = a ā1 (11a2 − 6a21 ),
a2 1
and
ā2 2
6ā2 a2 − 11a21 ā2 + 12ā21 a21 = a (11a2 − 6a21 ). (3.13)
a2 1
10
The equality (3.13) with (3.11) and (3.12) yields
|a2 | 2
6ā2 a2 − 11a21 ā2 + 12ā21 a21 = − āa22 a21 |a1|
2
4 (a2 − 6a1 )

ā2 2 |a2 |2 10|a1 |4


= a2 a1 |a1 |4 |a1 |4 +|a2 |2 a2
10|a2 |2
= a2 ā .
|a1 |4 +|a2 |2 1 2
Thus we have
10|a2 |2
 
4 2
12|a1 | + 6|a2 | = + 11 a21 ā2 .
|a1 |4 + |a2 |2
Using (3.10), we have
 
10|a2 |2 +11|a1 |4 +11|a2 |2 6(|a1 |4 +|a2 |2 )
12|a1 |4 + 6|a2 |2 = |a1 |4 +|a2 |2 11|a1 |4 +|a2 |2
a21 ā21
 
11|a1 |4 +21|a2 |2 6(|a1 |4 +|a2 |2 )
= |a1 |4 +|a2 |2 11|a1 |4 +|a2 |2
|a1 |4 (3.14)
4 2
1 | +21|a2 |
= 6 11|a
11|a1 |4 +|a2 |2
|a1 |4 ,
namely,
11|a1 |4 +21|a2 |2
2|a1 |4 + |a2 |2 = 11|a1 |4 +|a2 |2
|a1 |4
 2

20|a2 |
= 1+ 11|a1 |4 +|a2 |2
|a1 |4
20|a1 |4 |a2 |2
= |a1 |4 + 11|a1 |4 +|a2 |2
.
Hence, we have
20|a1 |4 |a2 |2
|a1 |4 + |a2 |2 = ,
11|a1 |4 + |a2 |2
(11|a1 |4 + |a2 |2 )(|a1 |4 + |a2 |2 ) = 20|a1 |4 |a2 |2 ,
11|a1 |8 + 12|a1 |4 |a2 |2 + |a2 |4 = 20|a1 |4 |a2 |2 ,
|a2 |4 − 8|a1 |4 |a2 |2 + 11|a1 |8 = 0,
and √ √
(|a2 |2 − (4 − 5)|a1 |4 )(|a2 |2 − (4 + 5)|a1 |4 ) = 0.
Hence, we have √
|a2 |2 = (4 ± 5)|a1 |4 ,
that is,

q
|a2 | = 4 ± 5 |a1 |2 .
Since a2 and a21 have the same direction in C (see (3.10)), we have

q
a2 = 4 ± 5 a21 .
Recall that (a1 , a2 ) is a zero of f . We have
0 = f (a1 , a2 ) = a22 ā2 − 6a21 a2 ā2 + 11a21 ā21 a2 − 6a41 ā21
= a2 |a2 |2 − 6a21 |a2 |2 + 11a2 |a1 |4 − 6a21 |a1 |4
= (a2 − 6a21 )|a2 |2 + (11a2 − 6a21 )|a1 |4
= (ξ± − 6)a21 |a2 |2 + (11ξ± − 6)a21 |a1 |4
= (ξ± − 6)ξ± 2 a2 |a |4 + (11ξ − 6)a2 |a |4 ,
1 1 ± 1 1
p √
where we set ξ± := 4 ± 5. Then we have
2
(ξ± − 6)ξ± + (11ξ± − 6) = 0,
11
3 2
ξ± − 6ξ± + 11ξ± − 6 = 0,
and
(ξ± − 1)(ξ± − 2)(ξ± − 3) = 0.
Finally we have
ξ± = 1, 2 or 3.
p √
This assertion contradicts to ξ± = 4± 5. We finally have α1 ̸= α2 , and see that (a1 , a2 ) is not a mixed
critical point.
We next prove the assertion (2), i.e., the surjectivity of f : C∗ 2 → C. For the mixed polynomial (3.2), if
z2 is a real variable, then we have
fa,b,c,d,e,f (1, z2 ) = z23 − (k + 3)z22 + (3k + 2)z2 − 2k.
Since k > 2, the equation fa,b,c,d,e,f (1, z2 ) = 0 has a real solution z2 (̸= 0). Hence we have
−1
fa,b,c,d,e,f (0) ∩ C∗ 2 ̸= ∅.
Recall that fa,b,c,d,e,f are strongly mixed weighted homogeneous if (a, b, c, d, e, f) is one of the 5 cases in Table
3.2. By the above results and (iii),(iv) of Proposition 12, we conclude that f2,2,1,2,1,4 (k = 3) : C∗ 2 → C is
surjective. (Hence, f2,2,1,2,1,4 (k = 3) is strongly Newton non-degenerate over the 1-dimensional face ∆(P ),
where P = t (1, 2). )

Finally, we prove the assertion (3). Let us consider the 0-dimensional face functions fS = z2a z̄23−a and
fT = −2kz1f z̄16−f of fa,b,c,d,e,f in the Table 3.2. If a = 2, then we have
∂fS ∂fS
= 2z2 z̄2 , = z22 ,
∂z2 ∂ z̄2
and hence, fS : C∗ 2 → C has no mixed critical points. If f = 4, then we have
∂fT ∂fT
= −8kz13 z̄12 , = −4kz14 z̄1 ,
∂z1 ∂ z̄1
and hence, fT : C∗ 2 → C has no mixed critical points. Hence, the 0-dimensional face functions fS and fT
of f := f2,b,c,d,e,4 are strongly Newton non-degenerate. This completes the proof of Lemma 14. □

3.4. Toric modifications associated with the regular simplicial cone subdivision Σ∗ . Now let
us consider the toric modification π̂ : X → C2 associated with the regular simplicial cone subdivision Σ∗
(Figure 4). Note that
π̂ −1 (0) = Ê(S) ∪ Ê(P ).
All 2-dimensional cones of Σ∗ are as follows (up to permutations of vertices):
     
1 1 1 1 1 0
σ1 := Cone(E1 , S) = , σ2 := Cone(S, P ) = , σ3 := Cone(P, E2 ) = . (3.15)
0 1 1 2 2 1

We now show that the Assumption (*) in Theorem 32 of [13] is satisfied for the mixed polynomial germ
(f2,2,1,2,1,4 (k = 3), 0) and our regular simplicial cone subdivision Σ∗ .

(I) We first set  


1 1
σ1′ := Cone(S, E1 ) = .
1 0
12
On the toric chart C2σ′ , the toric modification is written as π̂σ1′ (u1 , u2 ) = (u1 u2 , u1 ). The toric chart C2σ′
1 1
intersects the exceptional divisor Ê(S) only. If u0σ′ ∈ Ṽ ∩ π̂ −1 (0) ∩ C2σ′ , then u0σ′ ∈ Ṽ ∩ Ê(S). We have
1 1 1
fS = z22 z̄2 . Here we set
r1 := rdegS fS = 1 · 0 + 1 · (2 + 1) = 3, and p1 := pdegS fS = 1 · 0 + 1 · (2 − 1) = 1.
Then we have
r1 + p1 3+1 r1 − p1 3−1
= = 2, and = = 1,
2 2 2 2
and
(π̂σ∗ ′ f )(u1 , u2 ) = u21 ū1 − 6u31 ū1 u22 + 11u31 ū21 u22 ū22 − 6u41 ū21 u42 ū22
1
= u21 ū1 (1 − 6u1 u22 + 11u1 ū1 u22 ū22 − 6u21 ū1 u42 ū22 ).
We set
fe(u1 , u2 ) = 1 − 6u1 u22 + 11u1 ū1 u22 ū22 − 6u21 ū1 u42 ū22
and the strict transform Ṽ of V to X in the toric chart C2σ′ is given by
1

Ṽ = {fe(u1 , u2 ) = 0}.
If u0σ′ ∈ Ṽ ∩ π̂ −1 (0) ∩ C2σ′ , then u0σ′ ∈ Ṽ ∩ Ê(S). Hence, u0σ′ = (0, u2 ) for some u2 ∈ C. However, we
1 1 1 1
have f (0, u2 ) = 1 ̸= 0. Thus we have Ṽ ∩ Ê(S) = ∅ on the toric chart C2σ′ . Thus, the Assumption (*) in
1
Theorem 32 of [13] is satisfied for Σ∗ on the toric chart C2σ′ .
1
 
1 1
(II) We next consider the cone σ2 := Cone(S, P ) = and the toric chart C2σ2 . Both S and P are
1 2
strictly positive, and C2σ2 intersects the exceptional divisors Ê(S) and Ê(P ). If u0σ2 ∈ Ṽ ∩ π̂ −1 (0) ∩ C2σ2 ,
then u0σ2 ∈ Ṽ ∩ Ê(S) or u0σ2 ∈ Ṽ ∩ Ê(P ). If u0σ2 ∈ Ê(S) ∩ Ê(P ), then u0σ2 = (0, 0). If u0σ2 ∈ Ê(S) and
u0σ2 ̸∈ Ê(P ), then u0σ2 = (0, u2 ) for some u2 (̸= 0) ∈ C. If u0σ2 ̸∈ Ê(S) and u0σ2 ∈ Ê(P ), then u0σ2 = (u1 , 0)
for some u1 (̸= 0) ∈ C. Then u0σ2 is written as (0, u2 ) for some u2 (̸= 0) ∈ C in the toric chart C2σ′ , where
2
we set  
1 1
σ2′ := Cone(P, S) = .
2 1
Thus, the Assumption (*) in Theorem 32 of [13] is satisfied for Σ∗ on the toric chart C2σ2 .
 
1 0
(III) We finally consider the cone σ3 := Cone(P, E2 ) = and the toric chart C2σ3 . On the toric
2 1
chart C2σ3 , the toric modification is written as π̂σ3 (u1 , u2 ) = (u1 , u21 u2 ). The toric chart C2σ3 intersects the
exceptional divisor Ê(P ) only. If u0σ3 ∈ Ṽ ∩ π̂ −1 (0) ∩ C2σ3 , then u0σ3 ∈ Ṽ ∩ Ê(P ). Here fP = f and recall
that
r1 := rdegP f = 6, and p1 := pdegS fS = 2.
Then we have
r1 + p1 6+2 r1 − p1 6−2
= = 4, and = = 2,
2 2 2 2
and
(π̂σ∗3 f )(u1 , u2 ) = u41 ū21 u22 ū2 − 6u41 ū21 u2 ū2 + 11u41 ū21 u2 − 6u41 ū21
= u41 ū2 (u22 ū2 − 6u2 ū2 + 11u2 − 6).
We set
fe(u1 , u2 ) = u22 ū2 − 6u2 ū2 + 11u2 − 6,
then the strict transform Ṽ of V to X in the toric chart C2σ3 is given by
Ṽ = {fe(u1 , u2 ) = 0}.
13
If u0σ3 ∈ Ṽ ∩ π̂ −1 (0) ∩ C2σ3 , then u0σ3 ∈ Ṽ ∩ Ê(P ). Hence, u0σ3 = (0, u2 ) for some u2 ∈ C. Since
fe(0, 0) = −6 ̸= 0, we see that u2 ̸= 0. Thus, the Assumption (*) in Theorem 32 of [13] is satisfied for Σ∗
on the toric chart C2σ3 .

By the above arguments (I)∼(III), the Assumption (*) in Theorem 32 of [13] is satisfied for the mixed
polynomial germ (f2,2,1,2,1,4 (k = 3), 0) and Σ∗ . Hence, by Theorem 32, it is concluded that the strict
transform Ṽ of V := f −1 (0) via the toric modification π̂ : X → C2 is a real analytic manifold outside of
Ṽ ∩ π̂ −1 (0), and a topological manifold as a germ at Ṽ ∩ π̂ −1 (0).

Moreover, we have the following theorem:


Theorem 15. We set f := f2,2,1,2,1,4 (k = 3). The strict transform Ṽ of V := f −1 (0) via the toric
modification π̂ : X → C2 is a real analytic manifold as a germ at Ṽ ∩ π̂ −1 (0).
Proof. Recall the definition of L(Σ∗ ) in Theorem 32 of [13]. It is sufficient to prove that L(Σ∗ ) = ∅. The
cones of Σ∗ whose vertices are all strictly positive are
τ1 := Cone(S), τ2 := Cone(P ), σ2 := Cone(S, P ).
For τ1 , we have {(ν, µ) | cν,µ ̸= 0, ν + µ ̸∈ ∆(S)} = {((2, 1), (0, 1)), ((2, 1), (2, 0)), ((4, 0), (2, 0))} ̸= ∅.
Recall that rdegS fS = 3. Hence, we have
Λ(τ1 ) = min{S(ν + µ) − 3 | (ν, µ) = ((2, 1), (0, 1)), ((2, 1), (2, 0)), ((4, 0), (2, 0))}
= min{t (1, 1)(ν + µ) − 3 | ν + µ = (2, 2), (4, 1), (6, 0)}
= min{4 − 3, 5 − 3, 6 − 3} = 1.
For τ2 , we have {(ν, µ) | cν,µ ̸= 0, ν + µ ̸∈ ∆(P )} = ∅. For σ2 , we have {(ν, µ, S) | cν,µ ̸= 0, ν + µ ̸∈
∆(S)} ∪ {(ν, µ, P ) | cν,µ ̸= 0, ν + µ ̸∈ ∆(P )} = {((2, 1), (0, 1), S), ((2, 1), (2, 0), S), ((4, 0), (2, 0), S)} ̸= ∅.
Hence, we have Λ(σ2 ) = min{S(ν + µ) − 3 | (ν, µ) = ((2, 1), (0, 1)), ((2, 1), (2, 0)), ((4, 0), (2, 0))} = 1.
On the other hand, we see that Ṽ (τ1 )∗ ∩ C2σ1 = ∅, Ṽ (τ1 )∗ ∩ C2σ2 = ∅, and moreover, Ṽ (σ2 )∗ ∩ C2σ2 = ∅.
Thus it is concluded that L(Σ∗ ) = ∅. This completes the proof of Theorem 15. □

At the end of this paper, we present the following problem:


Problem 16.
• Does each f2,2,1,2,1,4 (k ̸= 3 and k > 2) have some mixed critical points on C∗ 2 ?
• More generally, does each f2,b,c,d,e,4 in the cases II ∼ V (a = 2, f = 4) of Table 3.2 have some
mixed critical points on C∗ 2 ?
We need another useful criteria for mixed critical points like Proposition 1 of [6].

References
[1] D. A. Gudkov, The topology of real projective algebraic varieties, Usp. Mat. Nauk, 29–4 (1974), 3–79, English transl.,
Russ. Math. Surveys, 29–4 (1974), 1–79.
[2] G. Ishikawa, S. Saito and T. Fukui (the second part), I. Shimada and H. Tokunaga (the first part), Algebraic curves
and singularities (in Japanese), Kyoritsu shuppan, 2001.
[3] V. M. Kharlamov, The topological type of nonsingular surfaces in RP 3 of degree four, Funct. Anal. Appl. 10 (1976),
295–305.
[4] V. V. Nikulin, Integral symmetric bilinear forms and some of their geometric applications, Izv. Akad. Nauk SSSR Ser
Mat. 43-1 (1979), 111–177, English transl., Math. USSR Izv., 14-1 (1980), 103–167.
[5] M. Oka, Non-degenerate complete intersection singularity, Hermann, Paris, 1997.
[6] M. Oka, Topology of polar weighted homogeneous hypersurfaces, Kodai Math. J. 31 (2008), 163–182.
[7] M. Oka, Non-degenerate mixed functions, Kodai Math. J. 33 (2010), 1–62.
[8] M. Oka, Mixed functions of strongly polar weighted homogeneous face type, Singularities in Geometry and Topology
2011, Advanced Study in Pure Math. 66 (2015), 173–202.
14
[9] M. Oka, Introduction to Complex and Mixed hypersurface singularities (in Japanese), Maruzen, Tokyo, 2018.
[10] V. A. Rokhlin, Proof of Gudkov conjecture, Funkts. Anal. Prilozhen., 6–2 (1972), 62–64. English transl., Funct. Anal.
Appl., 6 (1972), 62–64.
[11] V. A. Rokhlin, Congruences modulo 16 in Hilbert’s sixteenth problem I, Funkts. Anal. Prilozhen., 6–4 (1972), English
transl., Funct. Anal. Appl., 6 (1972), 301–306.
[12] V. A. Rokhlin, Congruences modulo 16 in Hilbert’s sixteenth problem II, Funkts. Anal. Prilozhen., 7–2 (1973), 91–92.
English transl., Funct. Anal. Appl., 7 (1973), 163–164.
[13] S. Saito and K. Takashimizu, Resolutions of Newton non-degenerate mixed polynomials of strongly polar non-negative
mixed weighted homogeneous face type, Kodai Math. J. 44 (2021), 457–491.
[14] S. Saito and K. Takashimizu, A note on Newton non-degeneracy of mixed weighted homogeneous polynomials,
arXiv:2107.08691, preprint.
[15] O. Ya. Viro, Gluing of plane algebraic curves and constructions of curves of degree 6 and 7, Lecture Notes in Math.,
1060 (1984), 187–200.
[16] O. Ya. Viro, Real algebraic plane curves: Constructions with controlled topology, Algebra and analysis, 1–5 (1989);
English transl. Leningrad Math. J., 1–5 (1990), 1059–1134.
[17] O. Ya. Viro, Patchworking real algebraic varieties, arXiv:math/0611382 [math.AG], 2006.
[18] G. Wilson, Hilbert’s sixteenth problem, Topology, 17 (1978), 53–73.

Department of Mathematics Education, Asahikawa Campus, Hokkaido University of Education, Asahikawa


070-8621, Hokkaido, Japan
Email address: saito.sachiko@a.hokkyodai.ac.jp

15

You might also like