On some convex combinations of biholomorphic mappings in several complex variables

Abstract

In this paper, our interest is devoted to study the convex combinations of the form \((1-\lambda)f+\lambda g\) where \(\lambda \in(0,1)\) of biholomorphic mappings on the Euclidean unit ball \({\mathbb B}^n\) the case of several complex variables. Starting from a result proved by S. Trimble [26] and then extended by P.N. Chichra and R. Singh [3, Th. 2] which says that if f is starlike such that \(Re[f'(z)]>0\) then \((1-\lambda)z+\lambda f(z)\) is also starlike, we are interested to extend this result to higher dimensions.

In the first part of the paper, we construct starlike convex combinations using the identity mapping on \({\mathbb B}^n\) and some particular starlike mappings on \({\mathbb B}^n\).

In the second part of the paper, we define the class \(L_{\lambda}^∗B^n\) and prove results involving convex combinations of normalized locally biholomorphic mappings and Loewner chains. Finally, we propose a conjecture that generalize the result proved by Chichra and Singh.

Authors

Eduard Stefan Grigoriciuc
Babes-Bolyai” University, Faculty of Mathematics and Computer Science, Cluj-Napoca, Romania
”Tiberiu Popoviciu” Institute of Numerical Analysis, Romanian Academy, Cluj-Napoca, Romania

Keywords

Biholomorphic mappings; Convex sums; Starlike mappings; Herglotz vector field; Loewner chains

Paper coordinates

E.S. Grigoriciuc, On some convex combinations of biholomorphic mappings in several complex variables, Filomat 36 (16) (2022), 5503-5519;
https://doi.org/10.2298/FIL226503G

PDF

About this paper

Journal

Filomat

Publisher Name

Faculty of Sciences and Mathematics, University of Nis, Serbia

Print ISSN

2406-0933

Online ISSN

google scholar link

[1] S.D. Bernardi, Convex and Starlike Univalent Functions, Trans. Amer. Math. Soc. 135 (1969), 429–446.
[2] D.M. Campbell, A survey of properties of the convex combination of univalent functions, Rocky Mountain J. Math. 5 (1975), 475–492.
[3] P. Chichra, R. Singh, Convex sum of univalent functions, J. Austral. Math. Soc. 14 (1972), 503–507.
[4] P.L. Duren, Univalent Functions, Springer-Verlag, New York, 1983.
[5] M. Elin, S. Reich, D. Shoikhet, Complex Dynamical Systems and the Geometry of Domains in Banach Spaces, Dissertationes Math. (Rozprawy Mat.) 427 (2004), 62 pp.
[6] I. Graham, H. Hamada, G. Kohr, Parametric Representation of Univalent Mappings in Several Complex Variables, Canad. J. Math. 54 (2002), 324–351.
[7] I. Graham, H. Hamada, G. Kohr, Radius problems for holomorphic mappings on the unit ball in Cn, Math. Nachr. 279 (2006), 1474–1490.
[8] I. Graham, H. Hamada, G. Kohr, M. Kohr, Extreme points, support points and the Loewner variation in several complex variables, Sci. China Math. 55(7) (2012), 1353–1366.
[9] I. Graham, G. Kohr, Geometric Function Theory in One and Higher Dimensions, Marcel Dekker Inc., New York, 2003.
[10] I. Graham, G. Kohr, M. Kohr, Loewner Chains and the Roper-Suffridge Extension Operator, J. Math. Anal. Appl. 247 (2000), 448–465.
[11] I. Graham, G. Kohr, M. Kohr, Loewner chains and parametric representation in several complex variables, J. Math. Anal. Appl. 281 (2003), 425–438.
[12] D.J. Hallenbeck, T.H. MacGregor, Linear Problems and Convexity Techniques In Geometric Function Theory, Pitman, Boston, 1984.
[13] H. Hamada, G. Kohr, Loewner chains and quasiconformal extension of holomorphic mappings, Ann. Polon. Math. 81 (2003), 85–100.
[14] H. Hamada, G. Kohr, Quasiconformal extension of biholomorphic mappings in several complex variables, J. Anal. Math. 96 (2005), 269–282.
[15] W.K. Hayman, Research Problems in Function Theory, The Athlone Press, London, 1967.
[16] G. Kohr, Basic Topics in Holomorphic Funcions of Several Complex Variables, Cluj University Press, Cluj-Napoca, 2003.
[17] T. Matsuno, On starlike theorems and convexlike theorems in the complex vector space, Sci. Rep. Tokyo Kyoiku Daigaku Sect. A 5 (1955), 88–95.
[18] T.H. MacGregor, The univalence of a linear combination of convex mappings, J. London Math. Soc. 44 (1969), 210–212.
[19] E.P. Merkes, On the convex sum of certain univalent functions and the identity function, Rev. Colombiana Math. 21 (1987), 5–12.
[20] J.A. Pfaltzgraff, Subordination Chains and Univalence of Holomorphic Mappings in Cn, Math. Ann. 210 (1974), 55–68.
[21] J.A. Pfaltzgraff, T.J. Suffridge, Close-to-starlike holomorphic functions of several complex variables, Pacif. J. Math. 57 (1975), 271–279.
[22] C. Pommerenke, Univalent Functions, Vandenhoeck and Ruprecht, Gottingen, 1975.
[23] S. Reich, D. Shoikhet, Nonlinear Semigroups, Fixed Points and the Geometry of Domains in Banach Spaces, World Scientific Publisher, Imperial College Press, London, 2005.
[24] K. Roper, T.J. Suffridge, Convexity properties of holomorphic mappings in Cn, Trans. Amer. Math. Soc., 351 (1999), 1803–1833.
[25] T.J. Suffridge, Starlikeness, convexity and other geometric properties of holomorphic maps in higher dimensions, Lecture Notes in Math., 599 (1976), 146–159.
[26] S. Trimble, The convex sum of convex functions, Math. Z. 109 (1969), 112–114.

On Some Convex Combinations of Biholomorphic Mappings in Several Complex Variables

On Some Convex Combinations of Biholomorphic Mappings in Several Complex Variables

Eduard Ştefan Grigoriciuc eduard.grigoriciuc@ubbcluj.ro “Babeş-Bolyai” University, Faculty of Mathematics and Computer Science, 1 M. Kogălniceanu Street, 400084, Cluj-Napoca, Romania and "Tiberiu Popoviciu" Insitute of Numerical Analysis, Romanian Academy, Cluj-Napoca, Romania
Abstract

In this paper, our interest is devoted to study the convex combinations of the form (1λ)f+λg, where λ(0,1), of biholomorphic mappings on the Euclidean unit ball 𝔹n in the case of several complex variables. Starting from a result proved by S. Trimble trimble and then extended by P.N. Chichra and R. Singh (chichra-singh, , Theorem 2) which says that if f is starlike such that Re[f(z)]>0, then (1λ)z+λf(z) is also starlike, we are interested to extend this result to higher dimensions. In the first part of the paper, we construct starlike convex combinations using the identity mapping on 𝔹n and some particular starlike mappings on 𝔹n. In the second part of the paper, we define the class λ*(𝔹n) and prove results involving convex combinations of normalized locally biholomorphic mappings and Loewner chains. Finally, we propose a conjecture that generalize the result proved by Chichra and Singh.

1 Introduction

Let n denote the space of n complex variables z=(z1,,zn) with the Euclidean inner product z,w=j=1nzjw¯j and the Euclidean norm z=z,z, for all z,wn. Also, let 𝔹n denote the Euclidean unit ball in n. In the case of one complex variable, the unit disc 𝔹1 is denoted by U.

Let H(𝔹n) denote the set of all holomorphic mappings from 𝔹n into n. If fH(𝔹n), we say that f is normalized if f(0)=0 and Df(0)=In, where Df(z) is the complex Jacobian matrix of f at z and In is the identity operator in n. Let

S(𝔹n)={fH(𝔹n):f is normalized and univalent }

be the set of all normalized biholomorphic mappings on 𝔹n.

A mapping fS(𝔹n) is called convex (starlike) if its image is a convex (respectively, starlike with respect to the origin) domain in n. We denote by

K(𝔹n)={fS(𝔹n):f(𝔹n) is a convex domain in n}

the class of normalized convex mappings on 𝔹n and by

S*(𝔹n)={fS(𝔹n):f(𝔹n) is a starlike domain with respect to zero in n}.

the class of normalized starlike mappings on 𝔹n. In the case of one complex variable, the sets S(U), K(U) and S*(U) are denoted by S, K and S*.

If fH(𝔹n), we say that f is locally biholomorphic on 𝔹n if Jf(z)0, for all z𝔹n, where Jf(z)=det(Df(z)), for all z𝔹n. We denote by

𝒮n(𝔹n)={f:𝔹nn:f is normalized and locally biholomorphic on 𝔹n}

the set of all normalized locally biholomorphic mappings on 𝔹n. If n=1, then 𝒮1(𝔹1) is denoted by 𝒮.

Another important class of normalized holomorphic functions on the unit disc U is the Carathéodory class (for details, one may consult (duren, , Chapter 2), (graham-kohr1, , p. 27) or (pommerenke, , Chapter 2)), denoted by

𝒫={pH(U):p(0)=1 and Re[p(ζ)]>0,ζU}.

In the case of several complex variables, we use the family

(𝔹n)={hH(𝔹n):h(0)=0,Dh(0)=In and Reh(z),z>0,z𝔹n{0}}

of normalized holomorphic mappings on the Euclidean unit ball. It is important to mention that the class (𝔹n) plays the role of the Carathéodory family in n. This class will be very important in the section that contains remarks about Loewner chains and Herglotz vector fields. For more details, one may consult graham-hamada-kohr1 , graham-hamada-kohr-kohr1 , graham-kohr1 , kohr1 , pfaltzgraff1 and reich-shoikhet .

Next, we recall the notions of Loewner chain and Herglotz vector field on the Euclidean unit ball in n. We will use these notions to prove that (under some particular assumptions) the convex combination of two Loewner chains is also a Loewner chain.

Definition 1.1

(see e.g. (graham-kohr1, , Definition 8.1.2) or (graham-kohr-kohr2, , Definition 1.1)): A mapping L=L(z,t):𝔹n×[0,)n is said to be a Loewner chain (normalized univalent subordination chain) if the following conditions hold:

  1. 1.

    etL(,t)S(𝔹n), for all t[0,);

  2. 2.

    L(𝔹n,s)L(𝔹n,t), for all 0st<.

Definition 1.2

(see e.g. (graham-kohr1, , Chapter 8)): Let h:𝔹n×[0,)n be a mapping. We say that h is a Herglotz vector field if the following conditions hold:

  1. 1.

    h(,t)(𝔹n), for all t[0,);

  2. 2.

    h(z,) is measurable on [0,), for all z𝔹n.

The following theorem gives a sufficient condition for a mapping L=L(z,t) to be a Loewner chain (see (graham-hamada-kohr1, , Lemma 1.6), graham-kohr-kohr2 , (hamada-kohr2, , Lemma 2.3) or (pfaltzgraff1, , Theorem 2.2)).

Theorem 1.3

Let L=L(z,t):𝔹n×[0,)n be a mapping which satisfies the following conditions:

  1. 1.

    L(,t)H(𝔹n), L(0,t)=0 and DL(0,t)=etIn, for all t[0,);

  2. 2.

    L(z,) is locally Lipschitz continuous on the interval [0,) locally uniformly with respect to z𝔹n.

Assume that there exists a Herglotz vector field h:𝔹n×[0,)n such that

Lt(z,t)=DL(z,t)h(z,t),a.e.t[0,),z𝔹n.

Moreover, assume that {etL(,t)}t0 is a normal family on 𝔹n. Then L(z,t) is a Loewner chain.

The connection between the class S and the Loewner chains in the case of one complex variable is given by a result due to Pommerenke (see e.g. (graham-kohr1, , Theorem 3.1.8)) which says that any function fS can be embedded as the first element of a Loewner chain (i.e. for each fS, there exists a Loewner chain L(ζ,t) such that L(ζ,0)=f(ζ), for all ζU).

This result is no longer true for the class S(𝔹n), and by this reason I. Graham, H. Hamada and G. Kohr (see graham-hamada-kohr1 ) defined the class

S0(𝔹n)={fS(𝔹n):L(z,t) a Loewner chain s.t. {etL(,t)}t0 is a normal family on 𝔹n and f=L(,0)}

of normalized univalent mappings which have parametric representation on 𝔹n. It is clear that S0(𝔹1)=S (see pommerenke ), but S0(𝔹n)S(𝔹n), for n2 (see graham-hamada-kohr1 and graham-kohr1 ). For details, one may consult also graham-hamada-kohr1 , (graham-kohr1, , Chapter 8) and graham-kohr-kohr2 .

2 Remarks on convex combinations

An interesting fact about the class of normalized univalent functions on the unit disc U in is that the class S is not convex. Namely, starting from two normalized univalent functions on the unit disc even the average of these functions does not necessarily belong to S. To show this, we present two examples in the case of one complex variable (see e.g. duren , hallenbeck-macgregor or macgregor ) and one example on the Euclidean unit ball in 2 (see e.g. graham-kohr1 or kohr1 ).

Example 2.1

(see e.g. (duren, , Exercise 3, Chapter 2)): Let

f(ζ)=ζ(1ζ)2andg(ζ)=ζ(1+ζ)2,ζU.

Then h=(f+g)/2 does not belong to S.

In Example 2.1, the functions f and g are not only normalized and univalent - they are even starlike on the unit disc U. However, the function h is not starlike on U (in fact, h is not even univalent on U). Hence, f,gS*, but hS* because hS. Another important example was given by MacGregor in (macgregor, , Section 3). He proved that the linear combination of two convex functions is not necessarily univalent on the unit disc.

Example 2.2

(see (macgregor, , Section 3)): Let

f(ζ)=ζ1ζandg(ζ)=ζ1+iζ,ζU.

Also let h(ζ)=tf(ζ)+(1t)g(ζ), for all 0<t<1. Then h is not univalent in U for each t(0,1).

In the previous example, f,gK are convex functions, but h is not univalent on U because there exists a point z0=(a+i)/(a+1) with |z0|<1, where a=(1t)/t, 0<t<1 such that h(z0)=0 (for the complete proof, one may consult (macgregor, , Section 3)).

Next, we can extend the statement of Example 2.1 to the case of several complex variables. For n=2, we obtain the following example (considered in (graham-kohr1, , Problem 6.2.3) and (kohr1, , Problem 4.3.4)):

Example 2.3

(see (graham-kohr1, , Problem 6.2.3)): Let

f(z)=(z1(1z1)2,z2(1z2)2)andg(z)=(z1(1+z1)2,z2(1+z2)2),z=(z1,z2)𝔹2.

Then h=(f+g)/2 is not starlike on the Euclidean unit ball 𝔹2. In fact h does not belong to S(𝔹2).

Remark 2.4

We can also analyze the convex combination between the mapping f(z)=z(1z1)2, for z𝔹n and one of the mappings g(z)=z(1+z1)2 or g(z)=z(1±z2)2, for z=(z1,,zn)𝔹n. Similar arguments like in the previous examples will show us if the mapping h=(f+g)/2 is starlike or not on the Euclidean unit ball 𝔹n.

Although linear combinations of univalent functions are not always univalent (for more details about these results, one may consult campbell , hallenbeck-macgregor , hayman or macgregor ), there exist subclasses of the class S that satisfy this condition (see chichra-singh , merkes or trimble in the case n=1). The goal of this paper is to extend in the case of several complex variables a result proved by P.N. Chichra and R. Singh in (chichra-singh, , Theorem 2) for the case of one complex variable. This result shows that the convex combination between a starlike function with positive real part of the derivative and the identity function is also starlike on the unit disc U, as it follows:

Theorem 2.5

Let λ[0,1]. If fS* and Re[f(ζ)]>0, for all ζU, then

h(ζ)=(1λ)ζ+λf(ζ) (1)

is starlike with respect to zero in U and Re[h(ζ)]>0, for all ζU.

In the case of several complex variables we start with some particular forms of the mapping f in order to construct convex combinations which are starlike on the Euclidean unit ball 𝔹n. Like in the case of result proved by Chichra and Singh, we also consider convex combinations between a starlike mapping and the identity map in n. First, we prove a general result for normalized locally bihomomorphic mappings on 𝔹n which satisfies some additional conditions in order to obtain the starlikeness of the convex combination. Then, in the final part, we propose a conjecture which generalize Theorem 2.5 in the case of several complex variables.

3 Preliminary results

Next we present some important results the will be used in the proofs of the main results from this paper. We recall, without proofs, the analytical characterization of starlikeness in n (proved by Matsuno, see matsuno or (graham-kohr1, , Theorem 6.2.2); see also elin-reich-shoikhet and suffridge ), the connection between Loewner chains and starlike mappings (see (graham-kohr1, , Corollary 8.2.3) or (pfaltzgraff-suffridge, , Corollary 2)) and also some important criteria for univalence in n.

Theorem 3.1

Let f:𝔹nn be a locally biholomorphic mapping such that f(0)=0. Then f is starlike if and only if

Re[Df(z)]1f(z),z>0,z𝔹n{0}. (2)

Using Theorem 3.1 we can prove that a locally biholomorphic mapping with f(0)=0 is starlike on the Euclidean unit ball 𝔹n. Another important characterization of starlikeness was given by Pfaltzgraff and Suffridge (see (graham-kohr1, , Corollary 8.2.3) or (pfaltzgraff-suffridge, , Corollary 2)) in terms of Loewner chains.

Theorem 3.2

Let f:𝔹nn be a normalized locally biholomorphic mapping on 𝔹n. Then f is starlike on 𝔹n if and only if L(z,t)=etf(z) is a Loewner chain.

We end this section with two important results that ensure the univalence of a normalized holmorphic mapping on the Euclidean unit ball 𝔹n. The first result, proved by Suffridge in (suffridge, , Theorem 7), is a version of the Noshiro-Warschawski’s univalence criteria (see e.g. (duren, , Theorem 2.16) for the case n=1) in the case of several complex variables.

Theorem 3.3

(see (suffridge, , Theorem 7)): Let f:𝔹nn be a normalized holomorphic mapping such that

ReDf(z)(u),u>0,z𝔹n,un,u=1. (3)

Then f is univalent on 𝔹n.

Another important criteria for univalence in n is presented in the following result proved by I. Graham, H. Hamada and G. Kohr in (graham-hamada-kohr2, , Lemma 2.2) (see also hamada-kohr1 and hamada-kohr2 ).

Theorem 3.4

(see (graham-hamada-kohr2, , Lemma 2.2)): Let f:𝔹nn be a normalized holomorphic mapping such that

Df(z)In<1,z𝔹n. (4)

Then fS0(𝔹n). In particular, f is univalent on 𝔹n.

Taking into account that S0(𝔹n)S(𝔹n), for n2 (see graham-hamada-kohr1 ), it is clear that there is an important difference between Theorem 3.3 (which assures us the univalence of a mapping on 𝔹n) and Theorem 3.4 (which assures us that a mapping admits parametric representation on 𝔹n), i.e. there exist normalized holomorphic mappings on 𝔹n that satisfy condition (3), but do not satisfy condition (4).

Such an example, that shows us that S0(𝔹n) is a proper subclass of S(𝔹n) for n2, is presented in (graham-kohr1, , Example 8.3.21).

4 Univalence of convex combinations in n

In view of the results presented in the previous section we can prove some criteria for univalence of a convex combination of normalized holomorphic mappings on the Euclidean unit ball 𝔹n. In fact, we can obtain a condition for a convex combination to be a mapping which has parametric representation on 𝔹n.

Lemma 4.1

Let f:𝔹nn be a normalized holomorphic mapping such that

ReDf(z)(u),u>0,

for all z𝔹n, un with u=1. Also let

h(z)=(1λ)z+λf(z),z𝔹n,λ[0,1].

Then h is univalent on 𝔹n.

Proof 4.2.

Clearly, h is a normalized holomorphic mapping. Moreover,

Dh(z)=(1λ)In+λDf(z),z𝔹n,λ[0,1]

and then

ReDh(z)(u),u=Re[(1λ)In+λDf(z)](u),u
=Re(1λ)In(u),u+ReλDf(z)(u),u=(1λ)Reu,u+λReDf(z)(u),u
=(1λ)u2+λReDf(z)(u),u=(1λ)+λReDf(z)(u),u>0,

for all z𝔹n, un with u=1. According to Theorem 3.3 we obtain that h is univalent on 𝔹n. Since h is also normalized, it means that hS(𝔹n).

Remark 4.3.

Notice that, in view of the previous proof, we obtain that if ReDf(z)(u),u>0, for all z𝔹n and un with u=1, then ReDh(z)(u),u>0, for all z𝔹n and un with u=1.

Let k*. We say that a mapping Pk:nn is a homogenous polynomial of degree k if there exist Qk:n××nn an k--linear operator such that Pk(z)=Qk(zk). For details, one may consult graham-kohr1 or kohr1 . Recall that we define the norm of the operator Pk by Pk=max{Pk(z):z=1}.

Lemma 4.4.

Let f:𝔹nn be a normalized holomorphic mapping such that

Df(z)In<1,

for all z𝔹n and

h(z)=(1λ)z+λf(z),z𝔹n,λ[0,1].

Then hS0(𝔹n). In particular, h is univalent on 𝔹n.

Proof 4.5.

Frist, it is clear that h is a normalized holomorphic mapping on 𝔹n. Moreover,

Dh(z)=(1λ)In+λDf(z),z𝔹n,λ[0,1]

and then

Dh(z)In=InλIn+λDf(z)In=λ(Df(z)In)=|λ|Df(z)InDf(z)In<1,

for all z𝔹n. In view of Theorem 3.4 (see (graham-hamada-kohr2, , Lemma 2.2)) we obtain that hS0(𝔹n), so h is also univalent on 𝔹n.

Lemma 4.6.

Let f:𝔹nn be a normalized holomorphic mapping such that f(z)=z+k=2Ak(zk), for all z𝔹n. Also let

h(z)=(1λ)z+λf(z),z𝔹n,λ(0,1).

If k=2kAk1, then hS0(𝔹n).

Proof 4.7.

Since f(z)=z+k=2Ak(zk), for all z𝔹n, we deduce that

Df(z)=In+k=2kAk(zk1,),z𝔹n,

where we used the fact that Ak(zk) is a homogenous polynomial of degree k. On the other hand,

Dh(z)=(1λ)In+λDf(z),z𝔹n.

Hence,

Dh(z)In=λ(Df(z)In)=λk=2kAk(zk1,)=|λ|k=2kAk(zk1,)
k=2kAk(zk1,)k=2kAkzk1k=2kAkz.

Then we obtain

Dh(z)Inzk=2kAkz<1,z𝔹n.

Consequently,

Dh(z)In<1,z𝔹n

and in view of Theorem 3.4 we deduce that hS0(𝔹n).

5 Particular starlike mappings on the Euclidean unit ball in n

In this section we present the first important result of this paper together with some examples to illustrate how this result can be applied in several particular cases. We begin this section with a well-known result related to starlike mappings on the Euclidean unit ball 𝔹n (see graham-kohr1 ).

Lemma 5.1.

Let f:𝔹nn be of the form f(z)=(f1(z1),,fn(zn)), for all z=(z1,,zn)𝔹n.

  1. 1.

    If f1,,fnS*, then fS*(𝔹n).

  2. 2.

    If, in addition, Re[fj(zj)]>0, for all j=1,n¯, then ReDf(z)(u),u>0, for all z𝔹n and un with u=1.

Proof 5.2.

Indeed, if f1,,fnS*, then fS*(𝔹n) (see e.g. (graham-kohr1, , Problem 6.2.5) or (kohr1, , Example 4.3.4)) and this completes the first part of the proof.

For the second part of the lemma, we have

Df(z)(u)=(f1(z1)0000f2(z2)00000fn(zn))(u1u2un)=(u1f1(z1),,unfn(zn))

and

ReDf(z)(u),u=|u1|2Re[f1(z1)]++|un|2Re[fn(zn)]>0,

for all z𝔹n and un with u=1 and this completes the proof.

According to Lemma 5.1 we can obtain a first version of Theorem 2.5 in the case of several complex variables. However, in this case we have a particular form of the mapping f (it has on each component a starlike function of one variable). The following result is very simple and its proof is immediate.

Proposition 5.3.

Let 0λ1 and let f1,,fnS* be such that Re[fj(zj)]>0, for all j=1,n¯. Also, let f(z)=(f1(z1),,fn(zn)), for all z𝔹n. Then

h(z)=(1λ)z+λf(z) (5)

is starlike, for all z𝔹n and λ[0,1]. Moreover, ReDh(z)(u),u>0, for all z𝔹n and un with u=1. In particular, h is univalent on 𝔹n.

Proof 5.4.

In view of Lemma 5.1 we have that fS*(𝔹n) and ReDf(z)(u),u>0, for all z𝔹n and un with u=1. On the other hand,

h(z)=(1λ)z+λf(z)=(1λ)(z1,,zn)+λ(f1(z1),,fn(zn))
=((1λ)z1+λf1(z1),,(1λ)zn+λfn(zn)).

If we denote hj(zj)=(1λ)zj+λfj(zj), for all j=1,n¯ and zjU, then according to Theorem 2.5 we obtain that hjS*, for all j=1,n¯ and λ[0,1]. Hence,

h(z)=(h1(z1),,hn(zn))

is a starlike mapping on 𝔹n, for all λ[0,1]. Moreover,

Dh(z)(u)=(h1(z1)0000h2(z2)00000hn(zn))(u1u2un)=(u1h1(z1),,unhn(zn))

and

ReDh(z)(u),u=|u1|2Re[h1(z1)]++|un|2Re[hn(zn)]>0,

for all z𝔹n and un with u=1 and this completes the proof.

It is clear that the mapping f used in the previous result has a very particular form (has on each component a starlike function of one complex variable). However, we can obtain similar results for an arbitrary starlike mapping on the Euclidean unit ball.

In the following examples (considered also in elin-reich-shoikhet , (graham-kohr-kohr1, , Example 3.5), (hamada-kohr1, , Example 3.4), roper-suffridge or (suffridge, , Examples 3 and 7)) we use arbitrary starlike mappings to construct starlike mappings (as convex combinations on 𝔹n). On the other hand, we can obtain starlike mappings h of the form (5) on 𝔹n using starlike mappings f that do not satisfy condition ReDf(z)(u),u>0, for all z𝔹n and un with u=1.

Example 5.5.

Let n=2 and f:𝔹22 be given by

f(z)=(z1+az22,z2),z=(z1,z2)𝔹2 (6)

with |a|33/2. According to (roper-suffridge, , Example 5), we know that fS*(𝔹2). Moreover,

h(z)=(1λ)z+λf(z)=(1λ)(z1,z2)+λ(z1+az22,z2)=((1λ)z1+λz1+λaz22,(1λ)z2+λz2),

so

h(z)=(z1+λaz22,z2),z=(z1,z2)𝔹2.

Then hS*(𝔹2) if and only if |λa|33/2. In particular, this is true for λ[0,1] and |a|33/2. Hence, hS*(𝔹2).

Example 5.6.

Let n=2 and f:𝔹22 be given by

f(z)=(z1+az1z2,z2),z=(z1,z2)𝔹2 (7)

with |a|1. According to (roper-suffridge, , Example 6), we know that fS*(𝔹2). Moreover,

h(z)=(1λ)z+λf(z)=(1λ)(z1,z2)+λ(z1+az1z2,z2)=((1λ)z1+λz1+λaz1z2,(1λ)z2+λz2),

so

h(z)=(z1+λaz1z2,z2),z=(z1,z2)𝔹2.

Then hS*(𝔹2) if and only if |λa|1. In particular, this is true for λ[0,1] and |a|1. Hence, hS*(𝔹2).

It is important to mention here that the results contained in Examples 5.5 and 5.6 can be directly verified. However, we can obtain starlike mappings h as convex combinations of two starlike mappings such that at least one of them does not satisfy the condition ReDf(z)(u),u>0, for all z𝔹n and un with u=1.

Remark 5.7.

Notice that in Examples 5.5 and 5.6 we can consider also a general case: a complex parameter λ with the property that |λ|1.

In the following example (considered also in graham-kohr1 and roper-suffridge ) we use a convex mapping f on 𝔹n to construct a starlike univalent mapping h on the Euclidean unit ball 𝔹n. The condition |a|1/2 also ensures the univalence of the mapping h on 𝔹n.

Example 5.8.

Let n=2 and f:𝔹22 be given by

f(z)=(z1+az22,z2),z=(z1,z2)𝔹2 (8)

with |a|1/2. Then h(z)=(1λ)z+λf(z) is starlike on 𝔹2, for all z𝔹2 and λ[0,1].

Proof 5.9.

According to (roper-suffridge, , Example 7) we know that fK(𝔹2). Moreover,

h(z)=(z1+λaz22,z2),z=(z1,z2)𝔹2.

Then hS*(𝔹2) because |λa|12<332. In addition, Df(z)=(12az201) and

ReDf(z)(u),u=Re(u1+2az2u2,u2),(u1,u2)=|u1|2+Re(2az2u2u¯1)+|u2|2=1+2Re(az2u2u¯1)
12|a||z2||u1||u2|>0,

for all z𝔹2, u2 with u=1 and |a|1/2. On the other hand,

Dh(z)=D((1λ)z+λf(z))=(1λ)I2+λDf(z) (9)

and

Dh(z)(u)=(1λ)u+λDf(z)(u),u2,u=1.

Then

Dh(z)(u),u=(1λ)u+λDf(z)(u),u=(1λ)u,u+λDf(z)(u),u=(1λ)u,u+λDf(z)(u),u
=(1λ)u2+λDf(z)(u),u=(1λ)+λDf(z)(u),u

and

ReDh(z)(u),u=(1λ)+λReDf(z)(u),u>0, (10)

for all z𝔹2, u2 with u=1 and λ[0,1]. Hence, h is a starlike univalent mapping on the Euclidean unit ball 𝔹2.

6 A general result on convex combinations of locally biholomorphic mappings in several complex variables

Next, we present another suggestive result which can be seen as a second version of Theorem 2.5 in the case of several complex variables. We can consider the following result as a generalization of the theorem proved by Chichra and Singh (see (chichra-singh, , Theorem 2)).

Theorem 6.1.

Let 0<λ<1 and μ=λ/(1λ). Also let f:𝔹nn be a normalized locally biholomorphic mapping such that

Df(z)In<1λ (11)

and

Re(In+μDf(z))1(z+μf(z)),z>0, (12)

for all z𝔹n{0}. Consider h:𝔹nn be given by

h(z)=(1λ)z+λf(z),z𝔹n. (13)

Then hS*(𝔹n), for all λ(0,1).

Proof 6.2.

In order to prove that hS*(𝔹n), for all λ(0,1), it is enough to show that h is locally biholomorphic on 𝔹n, h(0)=0 and

Re[Dh(z)]1h(z),z>0,z𝔹n{0}.

Since f is normalized on 𝔹n it follows that

h(0)=λf(0)=0andDh(0)=(1λ)In+λDf(0)=In.

Moreover,

Dh(z)In=(1λ)In+λDf(z)In=λDf(z)λIn=|λ|Df(z)In=λDf(z)In<λλ=1,

for all λ(0,1), in view of relation (11). According to Theorem 3.4 the above inequality assures us that h is univalent on 𝔹n. Obviously, in particular, h is locally biholomorphic on the Euclidean unit ball 𝔹n. Moreover, the relation

Dh(z)In<1

implies that Dh(z) is invertible, i.e. there exists the inverse operator [Dh(z)]1. Finally, the inequality

Re[Dh(z)]1h(z),z>0,z𝔹n{0}

is equivalent to

Re((1λ)I2+λDf(z))1((1λ)z+λf(z)),z=Re11λ(I2+μDf(z))1(1λ)(z+μf(z)),z
=Re(I2+μDf(z))1(z+μf(z)),z>0,z𝔹n{0}.

Hence, in view of relation (12), we obtain that

Re[Dh(z)]1h(z),z>0,

for all z𝔹n{0}. According to Theorem 3.1 we conclude that hS*(𝔹n), for all λ(0,1).

Remark 6.3.

It is clear that

  • if λ=0, then h(z)=z;

  • if λ=1, then h(z)=f(z),

for all z𝔹n and then hS*(𝔹n).

7 The class λ*(𝔹n)

Taking into account the main result from the previous section, we can define a class of normalized locally biholomorphic mappings on the Euclidean unit ball that satisfies conditions from Theorem 6.1. Hence, the convex combination between the identity map and a function from this class will be a starlike mapping on the Euclidean unit ball 𝔹n.

Definition 7.1.

Let us consider λ(0,1) and μ=λ/(1λ). We say that fλ*(𝔹n) if f is normalized locally biholomorphic on 𝔹n and f satisfies

  1. (a1)

    Df(z)In<1λ;

  2. (a2)

    Re(In+μDf(z))1(z+μf(z)),z>0, for all z𝔹n{0}.

In view of the above definition, we denote the class

λ*(𝔹n)={f𝒮n(𝔹n):f satisfies (a1) and (a2)}
Remark 7.2.

Note that λ*(𝔹n) as the identity mapping ϕ:𝔹nn given by ϕ(z)=z, for all z𝔹n belongs to λ*(𝔹n).

Next, we present an example (considered also in elin-reich-shoikhet , (graham-kohr-kohr1, , Example 3.5) or (suffridge, , Examples 3 and 7)) of a mapping f that has been used in Example 5.8 and which satisfies also the conditions (a1) and (a2) from Definition 7.1. This means that fλ*(𝔹2) and then we can construct a starlike mapping h which is a convex combination between the identity mapping and the mapping f.

Example 7.3.

Let n=2 and f:𝔹22 be given by

f(z)=(z1+az22,z2),z=(z1,z2)𝔹2, (14)

where |a|1/2. Then fλ*(𝔹2). In particular, hS*(𝔹2), where h(z)=(1λ)z+λf(z), for all z𝔹2 and λ(0,1).

Proof 7.4.

Let 0<λ<1 and μ=λ/(1λ). Then f is normalized locally biholomorphic on 𝔹2 and

Df(z)=(12az201).

If we denote

A=Df(z)I2=(12az201)(1001)=(02az200),

then

Df(z)I2=A=max{A(w):w=1}.

Let us consider wn such that w=1. It follows that

A(w)=|2az2w2|=2|a||z2||w2|

and A<1λ if and only if |a|12λ. Since λ(0,1) and |a|1/2, we obtain that condition (a1) from Definition 7.1 is satisfied. On the other hand,

B=I2+μDf(z)=(1001)+(μ2aμz20μ)=(1+μ2aμz201+μ)

and

B1=1(1+μ)2(1+μ2aμz201+μ)=(11+μ2aμz2(1+μ)2011+μ). (15)

Let us denote

C(z)=z+μf(z)=(z1,z2)+(μz1+2aμz22,μz2)=((1+μ)z1+2aμz22,(1+μ)z2),z=(z1,z2)𝔹2.

Then

ReB1C(z),z=Re((1+μ)z1+2aμz221+μ2aμz22(1+μ)(1+μ)2,(1+μ)z21+μ),(z1,z2)
=Re(z1+2aμz221+μ2aμz221+μ,z2),(z1,z2)=Re(z1,z2),(z1,z2)=z2>0,

for all z𝔹2{0}. Hence, condition (a2) from Definition 7.1 is also satisfied and then we conclude that fλ*(𝔹2). In particular, in view of Theorem 6.1 we obtain that hS*(𝔹2), where h is given by relation (13).

Remark 7.5.

In the second part of this section, let us to refer to the case n=1. Consider λ(0,1), μ=λ/(1λ) and f:U a normalized locally univalent function on the unit disc U in . Then

  1. 1.

    Condition (a1) can be written in one of the following form

    |f(ζ)1|<1λf(ζ)Uλ(1;1/λ),ζU, (16)

    where Uλ(1;1/λ) is the disc with center wλ=1 and radius rλ=1/λ. The smallest disc Uλ can be constructed for λ1 (in this case, we obtain the disc U1 of center w1=1 and radius r1=1). On the other hand, for λ0, it is clear that Re[f(ζ)]>0 for ζU implies condition (16), but the converse result is not necessarily true.

  2. 2.

    Condition (a2) can be written in one of the following form

    Re(1+μf(ζ))1(ζ+μf(ζ)),ζ>0

    or

    Re[ζ¯(ζ+μf(ζ))1+μf(ζ)]>0|ζ|2Re[ζ+μf(ζ)ζ+μζf(ζ)]>0, (17)

    for all ζU with ζ0. Clearly, if fS* and Re[f(ζ)]>0 for ζU, then the above condition is satisfied (for details, one may consult bernardi or chichra-singh ). But again, the converse result, is not necessarily true.

    Hence, we conclude that if λ(0,1) is sufficiently small and fS* is a function with the property that Re[f(ζ)]>0 for ζU, then (a1) and (a2) take place, but the converse implication is not necessarily true.

  3. 3.

    In view of previous remarks we can define the class

    λ*(𝔹1)=λ*(U)={fH(U):f(0)=0,f(0)=1,|f(ζ)1|<1λ and Re[ζ+μf(ζ)ζ+μζf(ζ)]>0,ζU{0}}

    for the case of one complex variable, where λ(0,1) and μ=λ/(1λ).

Remaining in the case n=1 we obtain the following result:

Proposition 7.6.

Let λ(0,1) and fλ*(U). Consider the function h:U be given by

h(ζ)=(1λ)ζ+λf(ζ),ζU.

Then hS*.

Proof 7.7.

Since fλ*(U) we deduce that hH(U), h(0)=0 and h(0)=1 (in fact, hS). Moreover,

Re[ζh(ζ)h(ζ)]=Re[ζ+μζf(ζ)ζ+μf(ζ)]>0,

for all ζU{0}, where μ=λ/(1λ). Hence, in view of the analytical characterization of starlikeness in (see (graham-kohr1, , Theorem 2.2.2)) we obtain that hS*.

Question 7.8.

What is the connection between the starlikeness of the mapping f and conditions (a1) and (a2) for n2? For sure, there will be no equivalence between the conditions, but the question would be whether the implication from the case n=1 is true.

8 Remarks on Loewner chains

Another interesting approach to the class λ* is that in terms of Loewner chains. We can prove that starting from a function fλ*(𝔹n) we can easily construct an associated Loewner chain according to Theorem 6.1 and the characterization of starlikeness with Loewner chains given by Theorem 3.2. In particular, we can obtain a Loewner chain that is the convex combination of another two Loewner chains.

Proposition 8.1.

Let λ(0,1). If fλ*(𝔹n), then

H(z,t)=(1λ)etz+λetf(z) (18)

is a Loewner chain, for all z𝔹n and t[0,).

Proof 8.2.

Let fλ*(𝔹n) and h:𝔹nn be given by

h(z)=(1λ)z+λf(z),z𝔹n,λ(0,1).

Then h(0)=0, Dh(0)=In and h is locally biholomorphic on 𝔹n. Moreover,

H(z,t)=(1λ)etz+λetf(z)=eth(z),z𝔹n,t[0,).

According to Theorem 6.1 we know that hS*(𝔹n). Since h is normalized locally biholomorphic on 𝔹n, it follows in view of Theorem 3.2 that H(z,t) is a Loewner chain, for all z𝔹n and t[0,).

Remark 8.3.

In view of Proposition 7.6 and Theorem 3.2 we obtain the previous result also in the case of one complex variable.

In the following remark we replace the mapping fλ*(𝔹n) with a starlike mapping on the Euclidean unit ball. However in order to obtain a Loewner chain (which is also a convex combination of the identity mapping and a starlike mapping on 𝔹n) we still need the assumption (a2) from Definition 7.1. According to this remark we deduce that in our context, for n2 this condition is very important.

Remark 8.4.

Let λ(0,1) and fS*(𝔹n) be such that Df(z)In<1λ, for all z𝔹n. Also consider the mapping H=H(z,t):𝔹n×[0,)n be given by

H(z,t)=(1λ)etz+λetf(z),z𝔹n,t[0,). (19)

In this case, H=H(z,t) is the convex combination of two Loewner chains

L1(z,t)=etz

and

L2(z,t)=etf(z),z𝔹n,t[0,).

Moreover, H(,t) is holomorphic on 𝔹n, H(0,t)=0 and DH(0,t)=etIn, for all t[0,). On the other hand, H(z,) is locally Lipschitz continuous on [0,) locally uniformly with respect to z𝔹n.

According to relation (19) we have that

etH(z,t)=et[(1λ)etz+λetf(z)]=(1λ)z+λf(z),

for all z𝔹n and t[0,). Let us denote h:𝔹nn given by

h(z)=(1λ)z+λf(z)=etH(z,t),

for all z𝔹n, t[0,) and λ(0,1). Since fS*(𝔹n), we deduce that {etH(,t)}t0 is a normal family on 𝔹n (see (duren, , Chapter 1) and kohr1 ).

In order to prove that H=H(z,t) is a Loewner chain (according to Theorem 1.3) we have to construct a Herglotz vector field P=P(z,t):𝔹n×[0,)n such that

Ht(z,t)=DH(z,t)P(z,t),a.e.t[0,),z𝔹n. (20)

Using the assumption Df(z)In<1λ, for all z𝔹n, we deduce that

Dh(z)In=(1λ)In+λDf(z)In=λDf(z)λIn=|λ|Df(z)In<λλ=1,

for all z𝔹n. Then the operator In+[Dh(z)In]=Dh(z) is invertible on 𝔹n and we can consider the inverse operator [Dh(z)]1 on the Euclidean unit ball 𝔹n. In view of this remark and relation (20) we obtain

P(z,t)=[DH(z,t)]1Ht(z,t)=et[(1λ)In+λDf(z)]1H(z,t)

or equivalently,

P(z,t)=[(1λ)In+λDf(z)]1[(1λ)z+λf(z)]

and hence

P(z,t)=[Dh(z)]1h(z), (21)

for all z𝔹n and t[0,). Clearly, P(z,) is measurable on [0,), for all z𝔹n because is constant with respect to t and then it remains to prove that P(,t)(𝔹n).

For simplicity let us consider n=2. Since Dh(z)=(1λ)I2+λDf(z) is invertible, it follows that P(,t) is holomorphic on 𝔹2, P(0,t)=0 and DP(0,t)=I2, for all t[0,). Indeed,

P(0,t)=[(1λ)I2+λDf(0)]1[λf(0)]=I2(0)=0,

for all t[0,). On the other hand, if we denote h(z)=(h1(z),h2(z)), then

Dh(z)=(h1z1h1z2h2z1h2z2)and[Dh(z)]1=1Jh(z)(h2z2h1z2h2z1h1z1),

where Jh(z)=det(Dh(z)), for all z𝔹2. Taking into account the previous relations we obtain the mapping

P(z,t)=1Jh(z)(h1(z)h2z2(z)h2(z)h1z2(z),h2(z)h1z1(z)h1(z)h2z1(z)),

for all z𝔹2 and t[0,). Now it is clear that P(0,t)=0, for all t[0,). Moreover, after some computations we deduce that

DP(z,t)=1Jh(z)(p11(z)p12(z)p21(z)p22(z)),

where

p11(z)=h1z1(z)h2z2(z)+h1(z)2h2z2z1(z)h2z1(z)h1z2(z)h2(z)2h1z2z1(z)
p12(z)=h1z2(z)h2z2(z)+h1(z)2h2z22(z)h2z2(z)h1z2(z)h2(z)2h1z22(z)
p21(z)=h2z1(z)h1z1(z)+h2(z)2h1z12(z)h1z1(z)h2z1(z)h1(z)2h2z12(z)
p22(z)=h2z2(z)h1z1(z)+h2(z)2h1z1z2(z)h1z2(z)h2z1(z)h1(z)2h2z1z2(z)

and then

DP(0,t)=1Jh(0)(h1z1(0)h2z2(0)h2z1(0)h1z2(0)00h1z1(0)h2z2(0)h2z1(0)h1z2(0))
=h1z1(0)h2z2(0)h2z1(0)h1z2(0)Jh(0)(1001)=Jh(0)Jh(0)I2=I2,

for all t[0,) since f is normalized and h1(0)=h2(0)=0. In order to complete the proof it remains to show that

ReP(z,t),z>0,z𝔹n{0},t[0,).

Indeed, we have

ReP(z,t),z=Re[Dh(z)]1h(z),z=Re[(1λ)In+λDf(z)]1[(1λ)z+λf(z)],z>0, (22)

for all z𝔹n{0} and t[0,). But relation (22) is the same as condition (a2) from Definition 7.1. Concluding the above arguments, we obtain the following result:

Proposition 8.5.

Let λ(0,1) and fS*(𝔹n) be such that

Df(z)In<1λ (23)

and

Re[(1λ)In+λDf(z)]1[(1λ)z+λf(z)],z>0, (24)

for all z𝔹n{0}. Then H=H(z,t) given by (19) is a Loewner chain. In particular, the Loewner chain H=H(z,t) is the convex combination of two Loewner chains.

Remark 8.6.

Let n=1, λ(0,1) and fS* be such that |f(ζ)1|<1, for all ζU. In view of relation (17) we obtain that condition (24) is

Re[ζ+μf(ζ)ζ+μζf(ζ)]>0,ζU{0},μ=λ/(1λ). (25)

If we denote h(ζ)=(1λ)ζ+λf(ζ), for all ζU, then h is holomorphic and normalized on U. In fact, in view of the assumption |f(ζ)1|<1, for all ζU, we obtain that hS. Next, let us define p:U×[0,) given by

p(ζ,t)=h(ζ)ζh(ζ),ζU,t[0,).

According to previous remarks, we deduce that p(,t)𝒫, for all t[0,) and p(ζ,) is measurable on the interval [0,), for all ζU, where

𝒫={pH(U):p(0)=1 and Re[p(ζ)]>0,ζU}

is the Carathéodory class in the case of one complex variable (for details, one may consult (duren, , Chapter 2), (graham-kohr1, , p. 27) or (pommerenke, , Chapter 2)).

If we consider H=H(ζ,t):U×[0,) given by H(ζ,t)=eth(ζ), then H(,t)H(U), H(0,t)=0 and H(0,t)=et, for all t[0,). Moreover,

Ht(ζ,t)=eth(ζ)=ζeth(ζ)p(ζ,t)=zH(ζ,t)p(ζ,t)a.e.t[0,),ζU.

Hence, taking into account the n=1 version of Theorem 1.3 (see (graham-kohr1, , Theorem 3.1.13)), we conclude that H=H(ζ,t) is a Loewner chain in .

9 Conjecture related to Chichra-Singh’s result in several complex variables

In this last section we propose a conjecture (for the case of several complex variables) which generalize Theorem 2.5 proved by Chichra and Singh in chichra-singh (in the case of one complex variable).

Conjecture 9.1.

Let λ[0,1]. If fS*(𝔹n) and ReDf(z)(u),u>0, for all z𝔹n and un with u=1, then

h(z)=(1λ)z+λf(z)

is a starlike mapping on 𝔹n. Moreover, ReDh(z)(u),u>0, for all z𝔹n and un with u=1. In particular, h is univalent on 𝔹n.

Remark 9.2.

In the case of one complex variable, the statement of Conjecture 9.1 is true, as it reduces to Theorem 2.5 obtained by Chichra and Singh in chichra-singh .

Acknowledgments

The author thanks to his first doctoral supervisor, Professor PhD Gabriela Kohr, for her advice and very useful suggestions during the preparation of this paper. The author is also grateful to his current PhD advisor, Professor PhD Mirela Kohr, for her continuous support and encouragement. The author thanks the referee(s) for carefully reading the manuscript and providing helpful suggestions.

References

  • (1) S.D. Bernardi, Convex and Starlike Univalent Functions, Trans. Amer. Math. Soc. 135 (1969), 429–446.
  • (2) D.M. Campbell, A survey of properties of the convex combination of univalent functions, Rocky Mountain J. Math. 5 (1975), 475–492.
  • (3) P. Chichra, R. Singh, Convex sum of univalent functions, J. Austral. Math. Soc. 14 (1972), 503–507.
  • (4) P.L. Duren, Univalent Functions, Springer-Verlag, New York, 1983.
  • (5) M. Elin, S. Reich, D. Shoikhet, Complex Dynamical Systems and the Geometry of Domains in Banach Spaces, Dissertationes Math. (Rozprawy Mat.) 427 (2004), 62 pp.
  • (6) I. Graham, H. Hamada, G. Kohr, Parametric Representation of Univalent Mappings in Several Complex Variables, Canad. J. Math. 54 (2002), 324–351.
  • (7) I. Graham, H. Hamada, G. Kohr, Radius problems for holomorphic mappings on the unit ball in n, Math. Nachr. 279 (2006), 1474–1490.
  • (8) I. Graham, H. Hamada, G. Kohr, M. Kohr, Extreme points, support points and the Loewner variation in several complex variables, Sci. China Math. 55(7) (2012), 1353–1366.
  • (9) I. Graham, G. Kohr, Geometric Function Theory in One and Higher Dimensions, Marcel Dekker Inc., New York, 2003.
  • (10) I. Graham, G. Kohr, M. Kohr, Loewner Chains and the Roper-Suffridge Extension Operator, J. Math. Anal. Appl. 247 (2000), 448–465.
  • (11) I. Graham, G. Kohr, M. Kohr, Loewner chains and parametric representation in several complex variables, J. Math. Anal. Appl. 281 (2003), 425–438.
  • (12) D.J. Hallenbeck, T.H. MacGregor, Linear Problems and Convexity Techniques In Geometric Function Theory, Pitman, Boston, 1984.
  • (13) H. Hamada, G. Kohr, Loewner chains and quasiconformal extension of holomorphic mappings, Ann. Polon. Math. 81 (2003), 85–100.
  • (14) H. Hamada, G. Kohr, Quasiconformal extension of biholomorphic mappings in several complex variables, J. Anal. Math. 96 (2005), 269–282.
  • (15) W.K. Hayman, Research Problems in Function Theory, The Athlone Press, London, 1967.
  • (16) G. Kohr, Basic Topics in Holomorphic Funcions of Several Complex Variables, Cluj University Press, Cluj-Napoca, 2003.
  • (17) T. Matsuno, On starlike theorems and convexlike theorems in the complex vector space, Sci. Rep. Tokyo Kyoiku Daigaku Sect. A 5 (1955), 88–95.
  • (18) T.H. MacGregor, The univalence of a linear combination of convex mappings, J. London Math. Soc. 44 (1969), 210–212.
  • (19) E.P. Merkes, On the convex sum of certain univalent functions and the identity function, Rev. Colombiana Math. 21 (1987), 5–12.
  • (20) J.A. Pfaltzgraff, Subordination Chains and Univalence of Holomorphic Mappings in n, Math. Ann. 210 (1974), 55–68.
  • (21) J.A. Pfaltzgraff, T.J. Suffridge, Close-to-starlike holomorphic functions of several complex variables, Pacif. J. Math. 57 (1975), 271–279.
  • (22) C. Pommerenke, Univalent Functions, Vandenhoeck and Ruprecht, Gottingen, 1975.
  • (23) S. Reich, D. Shoikhet, Nonlinear Semigroups, Fixed Points and the Geometry of Domains in Banach Spaces, World Scientific Publisher, Imperial College Press, London, 2005.
  • (24) K. Roper, T.J. Suffridge, Convexity properties of holomorphic mappings in n, Trans. Amer. Math. Soc., 351 (1999), 1803–1833.
  • (25) T.J. Suffridge, Starlikeness, convexity and other geometric properties of holomorphic maps in higher dimensions, Lecture Notes in Math., 599 (1976), 146–159.
  • (26) S. Trimble, The convex sum of convex functions, Math. Z. 109 (1969), 112–114.
2022

Related Posts