Abstract

We aim at investigating the geometric properties of the solutions of the initial-value problem which involves the following third-order linear differential equations: 𝜔(𝑧)+𝑄(𝑧)𝜔(𝑧)=0, 𝜔(0)=0, 𝜔(0)=1, 𝜔(0)=0, where 𝑄(𝑧) is analytic in the open unit disk 𝑈.

1. Introduction

Let 𝐴 denote the class of functions 𝑓 normalized by 𝑓(𝑧)=𝑧+𝑛=2𝑎𝑛𝑧𝑛,(1.1) which are analytic in the open unit disk 𝑈={𝑧: 𝑧 and |𝑧|<1}.

Also let 𝑆, 𝑆, 𝑆(𝛼), 𝐶, and 𝐶(𝛼) denote the subclasses of 𝐴 consisting of functions which are, respectively, univalent, starlike with respect to the origin, starlike of order 𝛼 in 𝑈 (0𝛼<1), convex with respect to the origin, and convex of order 𝛼 in 𝑈 (0𝛼<1) (see, for details, [14]). Furthermore, 𝑆𝑆(𝛽) and 𝐶𝐶(𝛽) denote the subclasses of 𝐴 consisting of functions which are strongly starlike of order 𝛽 and strongly convex of order 𝛽 in 𝑈, 0<𝛽1 (see, [5, 6]).

For functions 𝑓𝐴 with 𝑓(𝑧)0 (𝑧𝑈), we define the Schwarzian derivative of 𝑓 by 𝑓𝑆(𝑓,𝑧)=(𝑧)𝑓(𝑧)12𝑓(𝑧)𝑓(𝑧)2,𝑓𝐴;𝑓.(𝑧)0,𝑧𝑈(1.2) Note that Nehari [7] had proven the quotient of the linearly independent solution of (1.2) is univalent, while Robertson [8] and Miller [9] proved that the unique solution of the equation:𝑊(𝑧)+𝑎(𝑧)𝑊(𝑧)=0,𝑊(0)=0,𝑊(0)=1(1.3) is starlike.

Now, let 𝐵𝐽 denote the class of bounded functions 𝜔(𝑧)=𝜔1𝑧+𝜔2𝑧2+ analytic in the unit disk 𝑈 for which |𝜔(𝑧)|<𝐽. If 𝑔(𝑧)𝐵𝐽, then by using the Schwarz lemma, the function 𝜔(𝑧) defined by 𝜔(𝑧)=𝑧1/2𝑧0𝑔(𝑡)𝑡1/2𝑑𝑡(1.4) is also in 𝐵𝐽. Thus, in terms of derivatives, we have |||12𝜔(𝑧)+𝑧𝜔|||||||(𝑧)<𝐽,(𝑧𝑈),𝜔(𝑧)<𝐽,(𝑧𝑈).(1.5) In 1999, Saitoh [10] proved that the differential equation𝜔(𝑧)+𝑎(𝑧)𝜔(𝑧)+𝑏(𝑧)𝜔(𝑧)=0,(1.6) where 𝑎(𝑧) and 𝑏(𝑧) are analytic in the unit disc 𝑈, has a solution 𝜔(𝑧) univalent and starlike in 𝑈 under some conditions. Then in 2004, Owa et al. [11] studied geometric properties of the solutions of initial-value problem (1.6) and later, Saitoh [12] studied geometric properties of the solutions of the following second-order linear differential equation: 𝜔(𝑧)+𝑃𝑛(𝑧)𝜔(𝑧)=0,(1.7) where 𝑃𝑛(𝑧) is nonconstant polynomial of degree 𝑛1.

In this work, we aim at studying certain geometric properties of the solutions of the following initial-value problem: 𝜔(𝑧)+𝑄(𝑧)𝜔(𝑧)=0,𝜔(0)=0,𝜔(0)=1,𝜔(0)=0,(1.8) where 𝑄(𝑧)=𝑛=0𝑏𝑛𝑧𝑛 is analytic in 𝑈.

In order to prove our main results, we need the following definitions and theorems.

Definition 1.1 (see [13]). Let 𝐻𝐽 be the set of complex functions (𝑢,𝑣) satisfying the following:(i)(𝑢,𝑣) is continuous in a domain 𝐷×;(ii)(0,0)𝐷 and |(0,0)|<𝐽;(iii)|(𝐽𝑒𝑖𝜃,𝐾𝑒𝑖𝜃)|𝐽 when (𝐽𝑒𝑖𝜃,𝐾𝑒𝑖𝜃)𝐷, 𝜃 is real and 𝐾𝐽.

Definition 1.2 (see [13]). Let 𝐻𝐽 with corresponding domain 𝐷. We denote by 𝐵𝐽() those functions 𝜔(𝑧)=𝜔1𝑧+𝜔2𝑧2+ which are analytic in 𝑈 satisfying(i)(𝜔(𝑧),𝑧𝜔(𝑧))𝐷,(ii)|(𝜔(𝑧),𝑧𝜔(𝑧))|<𝐽(𝑧𝑈).

Theorem 1.3 (see [10]). For any 𝐻𝐽, 𝐵𝐽()𝐵𝐽,𝐻𝐽;𝐽>0.(1.9)

Theorem 1.4 (see [10]). Let 𝐻𝐽 and 𝑏(𝑧) be an analytic function in 𝑈 with |𝑏(𝑧)|<𝐽. If the differential equation 𝜔(𝑧),𝑧𝜔(𝑧)=𝑏(𝑧),𝜔(0)=0,𝜔(0)=1(1.10) has a solution 𝜔(𝑧) analytic in 𝑈, then |𝜔(𝑧)|<𝐽.

2. Main Results

Theorem 2.1. Let 𝑄(𝑧)=𝑛=0𝑏𝑛𝑧𝑛 be analytic in 𝑈 with 𝑛=0||𝑏𝑛||<𝐽(𝑧𝑈,𝐽>0),(2.1) and let 𝜔(𝑧) denote the solution of the initial value problem (1.8) in 𝑈. Then 1𝐽<1+𝑧𝜔(𝑧)𝜔(𝑧)>1+𝐽(𝑧𝑈,𝐽>0).(2.2)

Proof. If we let 𝑢(𝑧)=𝑧𝜔(𝑧)𝜔,(𝑧)(2.3) then 𝑢(𝑧) is analytic in 𝑈, such that 𝑢(0)=0 and (1.8) becomes []𝑢(𝑧)2𝑢(𝑧)+𝑧𝑢(𝑧)=𝑧2𝑛=0𝑏𝑛𝑧𝑛(2.4) or, equivalently, 𝑢(𝑧),𝑧𝑢(𝑧)=𝑧2𝑛=0𝑏𝑛𝑧𝑛,(2.5) where, for convenience, (𝜉,𝜂)=𝜉2𝜉+𝜂.(2.6) From assumption, we have 𝑛=0||𝑏𝑛||<𝐽(𝑧𝑈,𝐽>0).(2.7) By using Theorem 1.4, we have ||||𝑢(𝑧)<𝐽(2.8) which, in view of the relationship (2.3), yields ||||𝑧𝜔(𝑧)𝜔||||(𝑧)<𝐽,(2.9) that is, 1𝐽<1+𝑧𝜔(𝑧)𝜔(𝑧)>1+𝐽(𝑧𝑈,𝐽>0).(2.10)

Letting 𝐽=1 in Theorem 2.1, we have the following corollary.

Corollary 2.2. Let 𝑄(𝑧)=𝑛=0𝑏𝑛𝑧𝑛 be analytic in 𝑈 with 𝑛=0||𝑏𝑛||<1.(2.11) Let 𝜔(𝑧) be the solution of the initial-value problem in (1.8) in 𝑈. Then 𝜔(𝑧) is convex in 𝑈.

Example 2.3. Let 𝑄(𝑧)=1 in Corollary 2.2; the solution of the following initial-value problem: 𝜔(𝑧)+𝜔(𝑧)=0,𝜔(0)=0,𝜔(0)=1,𝜔(0)=0(2.12) is given by 𝜔(𝑧)=sin𝑧𝐶.(2.13)

Theorem 2.4. Let 𝑄(𝑧)=𝑛=0𝑏𝑛𝑧𝑛 be analytic in 𝑈 with 𝑛=0||𝑏𝑛||<𝐽(𝑧𝑈,0<𝐽1).(2.14) Let 𝜔(𝑧) be the solution of the initial-value problem in (1.8) in 𝑈. Then 𝜔(𝑧) is strongly convex of order 𝛼, that is, ||||arg1+𝑧𝜔(𝑧)𝜔||||<𝜋(𝑧)2𝛼(2.15) for some 𝛼 (0<𝛼1) and 2𝛼=𝜋sin1𝐽(0<𝐽1).(2.16)

Proof. By using the same technique as in the proof of Theorem 2.1, the required result is obtained.

Remark 2.5. Putting 𝛼=1 in Theorem 2.4, we have Corollary 2.2.

Acknowledgment

The work presented here was partially supported by UKM-ST-FRGS-0244-2010.