Cheap Complex Limit Cycles

N. Goncharuk, Yu. Kudryashov

February 5, 2019
arXiv:1702.00897v1 [math.CV] 3 Feb 2017

Consider a holomorphic foliation with singularities of a 2-dimensional

complex manifold. In this article we prove a new sufficient condi-
tion for this foliation to have countably many homologically in-
dependent complex limit cycles. In particular, if all leaves of a
foliation are dense in the phase space, and it has a complex hy-
perbolic singular point, then it has infinitely many homologically
independent complex limit cycles.

1 Introduction
Recall that the second part of Hilbert’s 16th problem asks about the number
and location of limit cycles of a polynomial vector field in R2 . In 1950s,
I. Petrovski and E. Landis proposed [13, 14] to solve this problem by extending
the vector field to the complex domain. They conjectured that a complex
polynomial vector field in C2 has finitely many complex limit cycles satisfying
some additional assumptions. This conjecture turned out to be false. In [6]
Yu. Ilyashenko provided an example of a complex foliation that has infinitely
many complex limit cycles satisfying these assumptions. Then he discovered
[8] that a generic complex polynomial vector field in C2 of a given degree n ≥ 2
has infinitely many complex limit cycles. For a more detailed review see [7].
From now on, let F be a holomorphic foliation with singularities of a 2-
dimensional complex manifold M. Recall a few definitions.

Definition 1. Let L be a leaf of F . A non-trivial free homotopy class [γ],

γ : S 1 → L, is called a complex limit cycle if the holonomy along (any) its
representative is non-identical, and is called an identical cycle otherwise.

1
Definition 2. A set of complex limit cycles of a foliation F is called homo-
logically independent, if for any leaf L all the cycles located on this leaf are
linearly independent in H1 (L).

Definition 3. A singular point P of a vector field is called complex hyperbolic,

if its linearization has two non-zero eigenvalues, and their ratio λ is not real.

Poincaré Normalization Theorem (see, e.g., [9, Theorem 5.5, p. 62]) implies
that a vector field is linearizable in some neighborhood of its complex hy-
perbolic singular point. In particular, the corresponding foliation is given by
wdz = λzdw in appropriate local coordinates (z, w).

Main Theorem. Let F be a holomorphic foliation with singularities of a

complex 2-dimensional manifold M. Suppose that it has a complex hyperbolic
singularity at P . Let U ∋ P be an open bidisc in a linearization chart for F
near P .
Let L and L′ be the local separatrixes of P in U. Let L̂ be the leaf that
includes L. Suppose that L̂ returns to U not (or not only) along L′ , i.e. L̂∩U *
L∪L′ . Then F possesses a countable set of homologically independent complex
limit cycles.

Remark 1. The local leaves of all points of U have P in their closures, hence
the assumptions of the Main Theorem do not depend on a particular choice of
U.
This theorem is mainly motivated by the study of polynomial foliations
of C2 . We shall list corollaries of Main Theorem related to this study, and
review the state of the art in this field in section 2 below. The next corollary
immediately follows from the Main Theorem.

Corollary 1. Let F be a holomorphic foliation with singularities of a complex

2-dimensional manifold M. Suppose that all leaves of F are dense in M, and
F possesses a complex hyperbolic singular point. Then F possesses a countable
set of homologically independent complex limit cycles.

Proof. Let P be a complex hyperbolic singular point of F . Let U, L, L′ , L̂

be as in the Main Theorem. Since L̂ is dense in M, it visits the open set
U \ (L ∪ L′ ). Hence, the Main Theorem implies that F possesses infinitely
many homologically independent complex limit cycles.

2
2 Corollaries of the Main Theorem concerning
polynomial foliations of C2
2.1 Polynomial foliations of C2
Let An be the class of foliations of C2 given by polynomial vector fields of
degree at most n. A generic foliation F ∈ An can be extended to a foliation
of CP 2 ; the extension has n + 1 singular points aj , j = 1, . . . , n + 1, at the
infinite line L∞ , and L∞ \ { aj } is a leaf of the extension. We do not provide
formulas for the extension (see, e.g., [9, Section 25A]), because we do not use
them.
Denote by λj the ratio of the eigenvalues of aj , the one corresponding to L∞
is in the denominator. The pseudogroup of monodromy maps along curves in
the leaf L∞ provides a lot of information about the initial foliation F .
The following theorem was initially proved by Petrovski and Landis [14].
Later some gaps were sealed in [15], see also textbook [9, Theorem 25.56].

Theorem 1 ([9, 14, 15]). For n ≥ 2, a generic foliation F ∈ An has no

algebraic leaves besides the line at infinity. The exceptional set is included by
a real algebraic subset of codimension one.

The geometric properties of a generic foliation F ∈ An are very different

from those of a generic foliation of the real plane. We shall discuss some of
the properties below. For more details, see the survey [17].
In 1962, M. Khudai–Verenov [10] proved that the leaves of a generic foliation
F ∈ An , n ≥ 3, are dense in C2 . Namely, he proved that all leaves of a foliation
F ∈ An are dense, provided that F has no algebraic leaves, and λj generate a
dense subgroup in C. So, the exceptional set in this article has measure zero,
but is dense in An .
In 1978, Yu. Ilyashenko [8] proved that it is enough to require that λj gen-
erate a dense subgroup in C/(2πiZ), so his proof works for n = 2. In 1984,
A. Shcherbakov [16] proved a similar result for a thinner exceptional set, with
the main assumption being the unsolvability of the monodromy group at in-
finity. Also, in 1994 I. Nakai provided another proof of the same fact.

Theorem 2 ([8, 10, 12, 16]). For n ≥ 2, all leaves of a generic foliation of
class An are dense in C2 . The exceptional set is included by a nowhere dense
real analytic subset of codimension 1.

As we mentioned above, the study of limit cycles of polynomial foliations

F ∈ An was motivated by Petrovskii and Landis attempt to solve the second

3
part of Hilbert’s 16th problem. In 1978, Yu. Ilyashenko proved [8] that a
generic foliation F ∈ An , n ≥ 3, has infinitely many homologically independent
complex limit cycles. As in his version of Theorem 2, the main assumption
was that the subgroup of C/2πiZ generated by λj is dense. This result was
reinvented in 1995 [4].
In 1984, A. Shcherbakov [16] announced a version of Ilyashenko’s theorem
with the main assumption replaced by unsolvability of the monodromy group
at infinity. The proof was published in 1998 [18], and works for n ≥ 3.
In the meantime, several authors [1, 12, 19] provided proofs of the fact that
an unsolvable subgroup of Aut(C, 0) possesses infinitely many attracting fixed
points accumulating to the origin. This fact is closely related to the study of
limit cycles for the following reason.
Remark 2. Let T be a cross-section for a foliation F of a two-dimensional
complex manifold M. Let γ : [0, 1] → M be a leafwise path with endpoints
in T . Suppose that the holonomy map Mγ : (T, γ(0)) → (T, γ(1)) extends to
a holonomy map Mγ : T → T . Suppose that Mγ has an isolated fixed point
p ∈ T . Then the curve “implementing” the equality Mγ (p) = p is a complex
limit cycle with multiplier Mγ′ (p).
Recently we provided [5] another proof of the main theorem of [18]. Our
proof works for n = 2, and avoids lengthy estimates of integrals.
Theorem 3 ([5, 8, 18]). For n ≥ 2, a generic foliation of class An possesses in-
finitely many homologically independent complex limit cycles. The exceptional
set is included by a nowhere dense real analytic submanifold of codimension
two.
Corollary 1 at page 2 shows that a slightly weaker version of Theorem 3,
with codimension one instead of two, follows from Theorems 1 and 2.
The following corollary of Main Theorem together with Theorem 1 implies
a version of Theorem 3 with another exceptional set (thinner than the one
provided by Corollary 1) of real codimension one.
Corollary 2. Let F be a polynomial foliation of class An . Suppose that F
has n + 1 distinct complex hyperbolic points at the infinite line, and F has no
algebraic leaves. Then it satisfies the assumptions of Main Theorem.
Proof. Let P be a complex hyperbolic singular point of F on the infinite line.
Consider the separatrix of P transversal to the infinite line. This leaf is not
algebraic, thus it accumulates to any non-singular point of the infinite line, see
[9, Lemma 28.10]. In particular, it returns to an arbitrary neighborhood of P .
Thus F satisfies the assumptions of Main Theorem.

4
2.2 Analytic foliations of C2
Let Aω be the set of analytic vector fields in C2 with the topology of compact
convergence.
The topology of the leaves of foliations given by generic vector fields v ∈ Aω
was investigated in [2, 3] by T. Firsova and T. Golenishcheva–Kutuzova.
Theorem 4 ([2]). There exists a residual set U F ⊂ Aω such that for v ∈ U F ,
at most countably many leaves of the foliation defined by v are topological
cylinders, and all other leaves are topological discs.
In [3], T. Kutuzova proved that one can drop “at most” from the statement
of this theorem, i.e., a generic foliation v ∈ Aω possesses countably many leaves
homeomorphic to cylinders.
The following corollary improves Kutuzova’s theorem.
Corollary 3 (cf. [3]). There exists an open dense subset U LC ⊂ Aω such
that for each v ∈ U LC , the foliation defined by v possesses infinitely many
homologically independent complex limit cycles.
This corollary together with Theorem 4 implies Kutuzova’s theorem. In-
deed, for v ∈ U LC , the corresponding foliation possesses infinitely many ho-
mologically independent complex limit cycles. On the other hand, for v ∈ U F ,
all leaves are either topological discs, or topological cylinders. Thus for v from
the residual set U LC ∩ U F , these limit cycles are located at distinct leaves,
and these leaves are topological cylinders. In particular, the corresponding
foliation has infinitely many leaves homeomorphic to cylinders.
Proof of Corollary 3. Let U LC be the set of vector fields v ∈ Aω satisfying the
assumptions of the Main Theorem. Let us prove that it is open and dense in
Aω .

U LC is open in Aω Consider a foliation F that satisfies the assumptions

of the Main Theorem. Let P , U, L, L′ , L̂ be as in the Main Theorem. Let
β : [0, 1] → L̂ be a path joining a point of L to a point of (L̂ ∩ U) \ (L ∪ L′ ).
Let K be the closure of an ε-neighborhood of U ∪ β([0, 1]). Clearly, K is a
compact set, and a foliation F ′ close to F on K satisfies the assumptions of
the Main Theorem. Thus U LC is open in Aω .

U LC is dense in Aω Note that for all n, U LC includes a dense subset of An ,

hence U LC ⊃ An , and the union of all An is dense in Aω . Therefore, U LC is
dense in Aω as well.

5
2.3 Polynomial foliations of CP 2
Let Bn be the class of foliations of CP 2 given by a polynomial vector field of
degree at most n in each affine chart. One can show that Bn ⊂ An ⊂ Bn+1 .
A generic foliation of class Bn , n > 2, has no algebraic leaves [9, Theorem
25.18], so we cannot use a monodromy pseudogroup to investigate it. For this
reason, we know much less about generic foliations of class Bn , than about those
of class An . In particular, it is not known if the leaves of a generic foliation
F ∈ Bn are dense in CP 2 , or if a generic foliation F ∈ Bn has infinitely many
homologically independent complex limit cycles.
For some properties, it is known that foliations from some non-empty open
subset of Bn , n > 2, have this property. In particular, it is known for the
density of the leaves [11], and for existence of infinitely many homologically
indepent complex limit cycles [5].
The following corollary provides an alternative proof of the latter fact.

Corollary 4. For n > 2, foliations from a non-empty open subset of Bn satisfy

the assumptions of the Main Theorem.

Proof. Since An−1 ⊂ Bn , the set of foliations satisfying the assumptions of the
Main Theorem is not empty for n > 2. Similarly to the first paragraph of the
proof of Corollary 3, this set is open in Bn .
Alternative proof. Due to the main theorem of [11], there exists an open subset
UnD ⊂ Bn such that all leaves of foliations F ∈ UnD are dense in Bn , so we can
apply Corollary 1 directly.

3 A hint for the homological independence of

complex limit cycles
Our approach includes a very simple idea which yields a great simplification
in the proofs concerning independence of complex limit cycles.
Both in [8, 18], the authors used the following proposition to establish the
independence of complex limit cycles.

Proposition 1. Consider a tuple of cycles cn , n = 1, . . . , N on a surface L

of real dimension 2. Suppose that these cycles are simple (i.e., have no self-
intersections) and pairwise disjoint. If [cn ] ∈ H1 (L), n = 1, . . . , N are linearly
dependent, then there exists a tuple αn ∈ { −1, 0, 1 }, n = 1, . . . , N, such that
P N
n=1 αn [cn ] = 0 ∈ H1 (L).

6
Corollary 5. Let F be an analytic foliation with singularities of the complex
plane C2 with coordinates (x, y). Let (cj ) be a sequence of leafwise cycles of F
such that

1. all cycles cj are simple and pairwise disjoint;

2. the sequence Z
Ij = x dy − y dx
cj

satisfies |Ij | > |I1 | + · · · + |Ij−1 |.

Then these cycles are homologically independent.

Proof.PSuppose that there exists PN a tuple αn ∈ { −1, 0, 1 }, n = 1, . . . , N, such

that N α
n=1 n n [c ] = 0. Then n=1 αn In = 0 which contradicts the inequality
|Ij | > |I1 | + · · · + |Ij−1 |.
The proofs in [8, 18] thus rely on the large estimates on integrals Ii . In [5],
we suggested to use the following corollary instead:

Corollary 6 (See [5, Lemma 9]). Let F be an analytic foliation with singu-
larities of a two-dimensional closed complex manifold M. Let cj be complex
limit cycles of F such that

1. all cycles cj are simple and pairwise disjoint;

2. their multipliers µj = µ(cj ) satisfy |µj | < 1 and 0 < |µj | < |µ1 · · · µj−1|.

Then these cycles are homologically independent.

Proof.PSuppose that there exists Q a tuple αn ∈ { −1, 0, 1 }, n = 1, . . . , N, such

N αn
that n=1 αn [cn ] = 0. Then (µn ) = 1 which contradicts the inequalities
|µj | < 1 and 0 < |µj | < |µ1 · · · µj−1|.
The calculations of the multipliers of complex limit cycles are much simpler
than the estimates on Ii , especially if the complex limit cycles are constructed
as fixed points of monodromy maps, see Remark 2 above.

7
4 Proof of the Main Theorem
Let M, F , P , U, L, L′ , L̂ be as in the Main Theorem. Recall that F is given by
zdw = λwdz in U, and L, L′ are given by w = 0, z = 0, respectively. We shall
identify U with the corresponding coordinate space. After shrinking U and
rescaling the coordinates, we may and will assume that U = { |z| < 1, |w| < 1 },
and F is given by zdw = λwdz in some neighborhood of U. Without loss of
generality, we may and will assume that ℑλ > 0, otherwise we pass to the
complex conjugate coordinates.
Due to assumptions of the Main Theorem, L̂ ∩ U * L ∪ L′ . Let q be a point
of (L̂ ∩ U ) \ (L ∪ L′ ). Let O be its projection to L. Note that O, q ∈ L̂. Join
O to q by a leafwise path β : [0, 1] → L̂.

Lemma 1 (Technical lemma). We can shrink U, rescale z, w, and choose a

smooth path β : [0, 1] → L̂ without self-intersections so that

• q = β(1) = (1, w0 ), |w0 | < 1;

• β(t) ∈
/ U for 0 < t < 1.

Proof. Take a leafwise path β̃ : [0, 1] → L̂ joining (1, 0) ∈ L to a point of

(L̂ ∩ ∂U) \ (L ∪ L′ ). Without loss of generality, we may and will assume that
β̃(0) = (1, 0), and β̃(t) ∈/ U for 0 < t < 1. Indeed, otherwise we cut β̃ to its
first return to U \(L∪L′ ), and deform it to avoid intersections with L∪L′ ⊂ U.
Put β̃(1) = (z̃, w̃), then |z̃| = 1 (“vertical” component of ∂U) or |w̃| = 1
(“horizontal” component of ∂U).

Case I: |z̃| = 1, |w̃| < 1 After rotation of z-coordinate, we may and will as-
sume that z̃ = 1. We modify an initial segment of β̃ so that it starts at
(1, 0), and immediately leaves U . This yields β with desired properties.

Case II: |w̃| = 1 The leaf L̂ near β̃(1) = (z̃, w̃) is given by z = z̃et , w = w̃eλt ,
t ∈ (C, 0). Let us append to β̃ a curve of the form βκ = (z̃eτ κ , w̃eλτ κ ),
τ ∈ [0, 1].
Since λ ∈/ R, we can choose κ so that ℜκ < 0 and ℜ(λκ) < 0, hence
both |z̃eτ κ | and |w̃eλτ κ | decrease as τ increases. Let us shrink U to
U ′ = { |z| ≤ |eκ z̃|, |w| ≤ 1 }, and scale the coordinates appropriately.
Then β̃∪βκ and U ′ satisfy assumptions of Case I This case is thus reduced
to Case I above.

8
 Consider the cross-section T = { (z, w) | z = 1 } ⊂ U. Let α : [0, 1] → L be
the unit circle α(t) = (e2πit , 0) making one turn around P in L \ { P }. The
holonomy map Mα : T → T along α is given by Mα (w) = νw, where ν = e2πiλ .
Recall that ℑλ > 0, hence |ν| = e−2πℑλ < 1, so Mα contracts in T .
Let us choose a small disc D, O ∈ D ⊂ T , such that

• the holonomy map Mβ : D → T along β is well-defined and univalent in

D;

• for w, w ′ ∈ D we have

Mβ (w)
|ν| < < |ν|−1 ; (1)
Mβ (w ′)

We shall shrink D once more later.

Put Mn = Mαn ◦ Mβ = ν n Mβ . Fix N so that for n > N we have Mn (D) ⊂
D, and Mn contracts in D. Let pn be the unique fixed point of Mn in D,
Mn (pn ) = pn .
Let cn be the complex limit cycle corresponding to pn , see Remark 2. This
cycle consists of two parts: a leafwise path γβ [pn ] from pn to Mβ (pn ), and the
spiral γαn [Mβ (pn )] from Mβ (pn ) to pn = ν n Mβ (pn ), where γαn [w] is given by

γαn [w](t) = e2πit , we2πiλt , t ∈ [0, n].

(2)

Let us prove that some subsequence (cnk ) satisfies the assumptions of Corollary 6.

Lemma 2. In the settings introduced above, for D small enough, the cycles
cn , n > N are simple and pairwise disjoint.

Proof. Let us shrink D so that γβ [w], w ∈ D, are simple, pairwise disjoint,

and γβ [w](t) ∈ / U for 0 < t < 1. We can satisfy the last requirement because
β(t) ∈/ U for 0 < t < 1, see Lemma 1.
M (p ) M (p )
For m 6= n we have βpn n = ν n 6= ν m = βpmm , hence pn 6= pm . Therefore,
the paths γβ [pn ], n > N, are simple and pairwise disjoint. Recall that cn =
γβ [pn ] ∪ γαn [Mβ (pn )]. The bidisc U includes all the spirals (2), so the spirals
γαn [Mβ (pn )] cannot intersect the paths γβ [w]|(0,1) . Thus it is enough to prove
that the spirals γαn [Mβ (pn )] are simple pairwise disjoint paths.
Consider two pairs (n, t), t ∈ [0, n], and (m, s), s ∈ [0, m], such that the
corresponding points of the spirals γαn [Mβ (pn )], γαm [Mβ (pm )] coincide,

e2πit , e2πiλt Mβ (pn ) = e2πis , e2πiλs Mβ (pm ) .

9
Since e2πit = e2πis , the difference k = t − s is integer. Then Mβ (pm ) =
M (p )
e2πiλ(t−s) Mβ (pn ) = ν k Mβ (pn ), thus Mββ (pmn ) = ν k . Due to (1), this is possible
only for k = 0, hence t = s, and Mβ (pm ) = Mβ (pn ). Recall that Mβ is univalent
in D, therefore pm = pn . Since pm 6= pn for m 6= n, we have m = n. Finally,
(n, t) = (m, s), so the spirals γαn [Mβ (pn )] are simple and pairwise disjoint.
Let µn be the multiplier of cn . Since |µn | = |ν|n × |Mβ′ (pn )| ≤ |ν|n ×
maxw∈D |Mβ′ (w)| → 0, we can choose a sequence (nk ) such that |µn0 | < 1, and
|µnk+1 | < |µn0 × · · · × µnk |. Finally, Corollary 6 implies that the cycles cnk are
homologically independent. This completes the proof of the Main Theorem.

References
[1] M. Belliart, I. Liousse, and F. Loray. “Sur l’existence de points fixes at-
tractifs pour les sous-groupes de Aut(C, 0)”. fr. In: C.R. Acad. Sci. Paris.
Série I 324 (1997), pp. 443–446. doi: 10.1016/S0764-4442(97)80083-3.
[2] T. Firsova. “Topology of analytic foliations in C2 . The Kupka-Smale
property”. In: Proceedings of the Steklov Institute of Mathematics 254
(2006), pp. 152–168. doi: 10.1134/S0081543806030060.
[3] T. I. Golenishcheva–Kutuzova. “A generic analytic foliation in C2 has in-
finitely many cylindrical leaves”. In: Proceedings of the Steklov Institute
of Mathematics 254 (2006), pp. 180–183. doi: 10.1134/S0081543806030084.
[4] X. Gómez-Mont and B. Wirtz. “On fixed points of conformal pseu-
dogroups”. In: Boletim da Sociedade Brasileira de Matemática - Bul-
letin/Brazilian Mathematical Society 26.2 (1995), pp. 201–209. issn: 1678-
7714. doi: 10.1007/BF01236994.
[5] N. Goncharuk and Y. Kudryashov. “Bounded limit cycles of polynomial
foliations of C2 ”. In: Bulletin of the Brazilian Mathematical Society, New
Series (2016), pp. 1–21. issn: 1678-7714. doi: 10.1007/s00574-016-0005-9.
arXiv: 1504.03313.
[6] Yu. S. Ilyashenko. “An Example of Equations dwdz
= QPnn(z,w)
(z,w)
Having a
Countable Number of Limit Cycles and Arbitrarily Large Petrovskiı̆-
Landis Genus”. In: Mathematics of the USSR-Sbornik 9.3 (1969), p. 365.
doi: 10.1070/SM1969v009n03ABEH001288. url: http://stacks.iop.org/0025-5734/9/i=
[7] Yu. S. Ilyashenko. “Centenial history of Hilbert’s 16th problem”. In:
Bull. Amer. Math. Soc. (N.S.) 39.3 (2002), 301–354 (electronic). doi:
10.1090/S0273-0979-02-00946-1.

10
 [8] Yu. S. Ilyashenko. “Topology of phase portraits of analytic differential
equations on a complex projective plane”. In: Trudy Sem. Petrovsk. 4
(1978), pp. 83–136.
[9] Yu. S. Ilyashenko and S. Yakovenko. Lectures on Analytic Differential
Equations. Vol. 86. Graduate Studies in Mathematics. American Math-
ematical Society, 2008. isbn: 9780821836675.
[10] M. G. Khudaı̆-Verenov. “A property of the solutions of a differential
equation. (Russian)”. Russian. In: Mat. Sbornik 56(98).3 (1962), pp. 301–
308.
[11] B. Mjuller. “On the density of solutions of an equation in CP 2 ”. In: Mat.
Sbornik 27.3 (1975), pp. 325–338. doi: 10.1070/SM1975v027n03ABEH002517.
[12] I. Nakai. “Separatrices for non-solvable dynamics on (C, 0)”. In: Ann.
Inst. Fourier 44.2 (1994), pp. 569–599. doi: 10.5802/aif.1410.
[13] I. G. Petrovskiı̆ and E. M. Landis. “On the number of limit cycles of the
equation dy/dx = P (x, y)/Q(x, y), where P and Q are polynomials”. In:
Mat. Sb. N.S. 43 (85 1957), pp. 149–168.
[14] I. G. Petrovskiı̆ and E. M. Landis. “On the number of limit cycles of the
equation dy/dx = P (x, y)/Q(x, y), where P and Q are polynomials of
2nd degree”. In: Mat. Sb. N.S. 37 (79 1955), pp. 209–250.
[15] I. G. Petrowsky. Selected works. Vol. 5.II: differential equations and prob-
ability theory. Classics of Soviet Mathematics. Amsterdam: Gordon and
Breach Publishers, 1996.
[16] A. Shcherbakov. “Topological and analytical conjugacy of non-commutative
groups of germs of conformal mappings”. Russian. In: Trudy Sem. Petro-
vskogo 10 (1984), pp. 170–196.
[17] A. A. Shcherbakov. “Dynamics of local groups of conformal mappings
and generic properties of differential equations on C2 ”. In: Proceedings
of the Steklov Institute of Mathematics 254.1 (2006), pp. 103–120. issn:
1531-8605. doi: 10.1134/S0081543806030047.
[18] A. Shcherbakov, E. Rosales–González, and L. Ortiz–Bobadilla. In: J. Dy-
nam. Control Systems 4 (1998), pp. 539–581. doi: 10.1023/A:1021819201777.
[19] B. Wirtz. “Fixed points and entropy for non solvable dynamics on (C, 0)”.
In: Boletim da Sociedade Brasileira de Matemática - Bulletin/Brazilian
Mathematical Society 29.1 (1998), pp. 53–74. issn: 1678-7714. doi: 10.1007/BF01245868.

11