1 Introduction

Let A be the class of analytic functions defined on the unit disk D:={zC:|z|<1} normalized by the condition f(0)=0= f (0)1. For two analytic functions f and g, we say that f is subordinate to g or g is superordinate to f, denoted by fg, if there is a Schwarz function w with |w(z)||z| such that f(z)=g(w(z)). If g is univalent, then fg if and only if f(0)=g(0) and f(D)g(D). For an analytic function φ whose range is starlike with respect to φ(0)=1 and is symmetric with respect to the real axis, let S (φ) denote the class of Ma-Minda starlike functions consisting of all fA satisfying z f (z)/f(z)φ(z). For special choices of φ, S (φ) reduces to well-known subclasses of starlike functions. For example, when 1B<A1, S [A,B]:= S ((1+Az)/(1+Bz)) is the class of Janowski starlike functions [1] (see [2]) and S [12α,1] is the class S (α) of starlike functions of order α and S := S (0) is the class of starlike functions. For φ(z):= 1 + z , the class S (φ) reduces to the class SL introduced by Sokół and Stankiewicz [3] and studied recently by Ali et al. [4, 5]. A function fA is in the class SL if z f (z)/f(z) lies in the region bounded by the right half-plane of the lemniscate of Bernoulli given by | w 2 1|<1. Analytically, SL:={fA:| ( z f ( z ) / f ( z ) ) 2 1|<1}. For b1/2 and a1, a more general class S [a,b] of the functions f satisfying | ( z f ( z ) / f ( z ) ) a b|<b was considered by Paprocki and Sokół [6]. Clearly, S [2,1]=:SL. For some radius problems related with the lemniscate of Bernoulli, see [3, 5, 7, 8]. Estimates for the initial coefficients of functions in the class SL are available in [8].

Let p be an analytic function defined on D with p(0)=1. Recently Ali et al. [4] determined conditions for p(z) 1 + z when 1+βz p (z)/ p k (z) with k=0,1,2 or (1β)p(z)+β p 2 (z)+βz p (z) is subordinated to 1 + z . Motivated by the works in [48], in Section 2 the condition on β is determined so that p(z) 1 + z when 1+βz p (z)/ p k (z)(1+Az)/(1+Bz) (1<k3). Similarly, the condition on β is determined so that p(z)(1+Az)/(1+Bz) when 1+βz p (z)/ p n (z) 1 + z , n=0,1,2. Further, the condition on β is obtained in each case so that p(z) 1 + z when p(z)+βz p (z)/ p n (z) 1 + z , n=0,1,2. At the end of this section, the problem p(z)+βz p (z)/p(z) 1 + z implies p(z)(1+Az)/(1+Bz) is also considered.

Silverman [9] introduced the class G b by

G b := { f A : | z f ( z ) / f ( z ) z f ( z ) / f ( z ) 1 | < b }

and proved G b S (2/(1+ 1 + 8 b )), 0<b1. Further, this result was improved by Obradovič and Tuneski [10] by showing G b S [0,b] S (2/(1+ 1 + 8 b )), 0<b1. Tuneski [11] further obtained the condition for G b S [A,B]. Inspired by the work of Silverman [9], Nunokawa et al. [12] obtained the sufficient conditions for a function in the class G b to be strongly starlike, strongly convex, or starlike in D. By setting p(z)=z f (z)/f(z), the inclusion G b S [A,B] can be written as

1+ z p ( z ) p 2 ( z ) 1+bzp(z) 1 + A z 1 + B z .

Recently Ali et al. [13], obtained the condition on the constants A,B,D,E[1,1] and β so that p(z)(1+Az)/(1+Bz) when 1+βz p (z)/ p n (z)(1+Dz)/(1+Ez), n=0,1. In Section 3, alternate and easy proofs of results [[13], Lemmas 2.1, 2.10] are discussed. Further, this section is concluded with the condition on A,B,D,E[1,1] and β such that 1+βz p (z)/ p 2 (z)(1+Dz)/(1+Ez) implies p(z)(1+Az)/(1+Bz).

The following results are required in order to prove our main results.

Lemma 1.1 [[14], Corollary 3.4h, p.135]

Let q be univalent in D, and let φ be analytic in a domain D containing q(D). Let z q (z)φ(q(z)) be starlike. If p is analytic in D, p(0)=q(0) and satisfies

z p (z)φ ( p ( z ) ) z q (z)φ ( q ( z ) ) ,

then pq and q is the best dominant.

The following is a more general form of the above lemma.

Lemma 1.2 [[14], Corollary 3.4i, p.134]

Let q be univalent in D, and let φ and ν be analytic in a domain D containing q(D) with φ(w)0 when wq(D). Set

Q(z):=z q (z)φ ( q ( z ) ) ,h(z):=ν ( q ( z ) ) +Q(z).

Suppose that

  1. (1)

    h is convex or Q(z) is starlike univalent in D and

  2. (2)

    Re( z h ( z ) Q ( z ) )>0 for zD.

If

ν ( p ( z ) ) +z p (z)φ ( p ( z ) ) ν ( q ( z ) ) +z q (z)φ ( q ( z ) ) ,
(1.1)

then pq and q is the best dominant.

Lemma 1.3 [[14], Corollary 3.4a, p.120]

Let q be analytic in D, let ϕ be analytic in a domain D containing q(D) and suppose

  1. (1)

    Reϕ[q(z)]>0 and either

  2. (2)

    q is convex, or

  3. (3)

    Q(z)=z q (z)ϕ[q(z)] is starlike.

If p is analytic in D, with p(0)=q(0), p(D)D and

p(z)+z p (z)ϕ [ p ( z ) ] q(z),

then p(z)q(z).

2 Results associated with the lemniscate of Bernoulli

In the first result, condition on β is obtained so that the subordination

1+β z p ( z ) p k ( z ) 1 + A z 1 + B z (1<B<A1)

implies p(z) 1 + z .

Lemma 2.1 Let |β| 2 ( k + 3 ) / 2 (AB)+|Bβ|, 1<k3. Let p be an analytic function defined on D with p(0)=1 satisfying

1+β z p ( z ) p k ( z ) 1 + A z 1 + B z (1<B<A1),

then p(z) 1 + z .

Proof Let q(z)= 1 + z . A computation shows that the function

Q(z):=β z q ( z ) q k ( z ) = β z 2 ( 1 + z ) ( k + 1 ) / 2 (1<k3)

is starlike in the unit disk D. Consider the subordination

1+β z p ( z ) p k ( z ) 1+β z q ( z ) q k ( z ) .

Thus in view of Lemma 1.1, it follows that p(z)q(z). In order to prove our result, we need to prove

1 + A z 1 + B z 1+ β z q ( z ) q k ( z ) =1+ β z 2 ( 1 + z ) ( k + 1 ) / 2 :=h(z).

Let w=Φ(z)= 1 + A z 1 + B z . Then Φ 1 (w)= w 1 A B w . The subordination Φ(z)h(z) is equivalent to z Φ 1 (h(z)). Thus in order to prove the result, we need only to show | Φ 1 (h( e i t ))|1. For z= e i t , πtπ, we have

| Φ 1 ( h ( e i t ) ) | | β | 2 ( A B ) ( 2 cos ( t / 2 ) ) ( k + 1 ) / 2 + | B β | =:g(t).

A calculation shows that g(t) attains its minimum at t=0. Further, the value of g(t) at π or −π comes out to be 1/|B| which is naturally greater than the value at the extreme point t=0 because if g(0)g(π), then (AB)|β|0 which is absurd. Thus

g(0)= | β | 2 ( k + 3 ) / 2 ( A B ) + | B β | 1

for |β| 2 ( k + 3 ) / 2 (AB)+|Bβ|. Hence Φ(z)h(z), and the proof is complete now. □

Next result depicts the condition on β such that 1+βz p (z) 1 + z implies p(z)(1+Az)/(1+Bz) (1B<A1). On subsequent lemmas, similar results are obtained by considering the expressions 1+βz p (z)/p(z) and 1+βz p (z)/ p 2 (z).

Lemma 2.2 Let (AB)β 2 ( 1 + | B | ) 2 + ( 1 B ) 2 and 1B<A1. Let p be an analytic function defined on D with p(0)=1 satisfying

1+βz p (z) 1 + z ,

then p(z) 1 + A z 1 + B z .

Proof Define the function q:DC by

q(z)= 1 + A z 1 + B z (1B<A1)

with q(0)=1. A computation shows that

Q(z)=βz q (z)= β ( A B ) z ( 1 + B z ) 2

and

z Q ( z ) Q ( z ) = 1 B z 1 + B z .

Let z=r e i t , r(0,1), πtπ. Then

Re ( 1 B z 1 + B z ) = Re ( 1 B r e i t 1 + B r e i t ) = 1 B 2 r 2 | 1 + B r e i t | 2 .

Since 1 B 2 r 2 >0 (|B|1, 0<r<1) and so Re(z Q (z)/Q(z))>0, this shows that Q is starlike in D. It follows from Lemma 1.1 that the subordination

1+βz p (z)1+βz q (z)

implies p(z)q(z). Now we need to prove the following in order to prove the lemma:

1 + z 1+βz q (z)=1+β ( A B ) z ( 1 + B z ) 2 =:h(z).

Let w=Φ(z)= 1 + z . Then Φ 1 (w)= w 2 1. The subordination Φ(z)h(z) is equivalent to the subordination z Φ 1 (h(z)). Now in order to prove the result, it is enough to show | Φ 1 (h( e i t ))|1, z= e i t , πtπ. Now

| Φ 1 ( h ( e i t ) ) | =| ( 1 + β ( A B ) e i t ( 1 + B e i t ) 2 ) 2 1|1implies that|1+β ( A B ) e i t ( 1 + B e i t ) 2 | 2 .

Further,

| 1 + β ( A B ) e i t ( 1 + B e i t ) 2 | = | 1 + ( 2 B + β ( A B ) ) e i t + B 2 e 2 i t | | 1 + 2 B e i t + B 2 e 2 i t | Re ( 2 B + β ( A B ) + B 2 e i t + e i t ) 1 + 2 | B | + B 2 = 2 B + β ( A B ) + ( 1 + B 2 ) cos t ( 1 + | B | ) 2 2 B + β ( A B ) ( 1 + B 2 ) ( 1 + | B | ) 2 2

for (AB)β 2 ( 1 + | B | ) 2 + ( 1 B ) 2 . Therefore Φ(z)h(z) and this completes the proof. □

Lemma 2.3 Let (AB)β( 2 1)(1+|A|)(1+|B|) and 1B<A1. Let p be an analytic function defined on D with p(0)=1 satisfying

1+β z p ( z ) p ( z ) 1 + z ,

then p(z) 1 + A z 1 + B z .

Proof Let the function q:DC be defined by

q(z)= 1 + A z 1 + B z (1B<A1).

A computation shows that

Q(z):= β z q ( z ) q ( z ) = β ( A B ) z ( 1 + A z ) ( 1 + B z )

and

z Q ( z ) Q ( z ) = 1 A B z 2 ( 1 + A z ) ( 1 + B z ) .

Let z=r e i t , r(0,1), πtπ. Then

Re ( 1 A B z 2 ( 1 + A z ) ( 1 + B z ) ) = Re ( 1 A B r 2 e 2 i t ( 1 + A r e i t ) ( 1 + B r e i t ) ) = ( 1 A B r 2 ) ( 1 + ( A + B ) r cos t + A B r 2 ) | 1 + A r e i t | 2 | 1 + B r e i t | 2 .

Since 1+AB r 2 +(A+B)rcost(1Ar)(1Br)>0 for A+B0 and, similarly, 1+AB r 2 +(A+B)rcost(1+Ar)(1+Br)>0 for A+B0, it follows that Q is starlike in D. Lemma 1.1 suggests that the subordination

1+β z p ( z ) p ( z ) 1+β z q ( z ) q ( z )

implies p(z)q(z). Now we have to prove

1 + z 1+β z q ( z ) q ( z ) =1+ β ( A B ) z ( 1 + A z ) ( 1 + B z ) =:h(z).

Let w=Φ(z)= 1 + z . Then Φ 1 (w)= w 2 1. The subordination Φ(z)h(z) is equivalent to the subordination z Φ 1 (h(z)). Now in order to prove the result, it is enough to show | Φ 1 (h( e i t ))|1, πtπ. Now

| Φ 1 ( h ( e i t ) ) | = | ( 1 + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) ) 2 1 | 1 implies that | 1 + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) | 2 .

Further,

| 1 + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) | Re ( 1 + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) ) 1 + ( A B ) β ( 1 + | A | ) ( 1 + | B | ) 2

for (AB)β( 2 1)(1+|A|)(1+|B|). Therefore Φ(z)h(z) and this completes the proof. □

Lemma 2.4 Let (AB)β( 2 1) ( 1 + | A | ) 2 + ( 1 A ) 2 and 1B<A1. Let p be an analytic function defined on D with p(0)=1 satisfying

1+β z p ( z ) p 2 ( z ) 1 + z ,

then p(z) 1 + A z 1 + B z .

Proof Let the function q:DC be defined by

q(z)= 1 + A z 1 + B z (1B<A1)

with q(0)=1. Then

Q(z)= β z q ( z ) q 2 ( z ) = β ( A B ) z ( 1 + A z ) 2

and

z Q ( z ) Q ( z ) = 1 A z 1 + A z .

Let z=r e i t , πtπ, 0<r<1. Then

Re ( 1 A z 1 + A z ) = 1 A 2 r 2 | 1 + A r e i t | 2 .

Since 1 A 2 r 2 >0 (|A|1, 0<r<1). Hence Re(z Q (z))/Q(z)>0, this shows that Q is starlike in D. An application of Lemma 1.1 reveals that the subordination

1+β z p ( z ) p 2 ( z ) 1+β z q ( z ) q 2 ( z )

implies p(z)q(z). Now our result is established if we prove

1 + z 1+β z q ( z ) q 2 ( z ) =1+β ( A B ) z ( 1 + A z ) 2 =:h(z).

The rest of the proof is similar to that of Lemma 2.2, and therefore it is skipped here. □

In the next result, the condition on β is obtained so that p(z)+βz p (z) 1 + z implies p(z) 1 + z . On subsequent lemmas, similar results are discussed by considering the expressions p(z)+βz p (z)/p(z) and p(z)+βz p (z)/ p 2 (z).

Lemma 2.5 Let p be an analytic function defined on D with p(0)=1 satisfying p(z)+βz p (z) 1 + z , β>0. Then p(z) 1 + z .

Proof Define the function q:DC by q(z)= 1 + z with q(0)=1. Since q(D)={w:| w 2 1|<1} is the right half of the lemniscate of Bernoulli, q(D) is a convex set, and hence q is a convex function. Let us define ϕ(w)=β, then

Reϕ [ q ( z ) ] =β>0.

Consider the function Q defined by

Q(z):=z q (z)ϕ ( q ( z ) ) =β z 2 1 + z .

Further,

Re ( z Q ( z ) Q ( z ) ) = 1 Re ( z 2 ( 1 + z ) ) 3 4 > 0 .

Thus the function Q is starlike, and the result now follows by an application of Lemma 1.3. □

Lemma 2.6 Let p be an analytic function defined on D with p(0)=1 satisfying

p(z)+β z p ( z ) p ( z ) 1 + z ,β>0.

Then p(z) 1 + z .

Proof As before, let q be given by q(z)= 1 + z with q(0)=1. Then q is a convex function. Let us define ϕ(w)=β/w. Since q(D)={w:| w 2 1|<1} is the right half of the lemniscate of Bernoulli, so

Reϕ [ q ( z ) ] = β | 1 + z | 2 Re( 1 + z )>0.

Consider the function Q defined by

Q(z):=β z q ( z ) q ( z ) =β z 2 ( 1 + z ) .

Further,

Re ( z Q ( z ) Q ( z ) ) =1Re ( z 1 + z ) 1 2 >0.

Thus the function Q is starlike, and the result now follows by an application of Lemma 1.3. □

Lemma 2.7 Let p be an analytic function defined on D with p(0)=1 satisfying

p(z)+β z p ( z ) p 2 ( z ) 1 + z ,β>0.

Then p(z) 1 + z .

Proof Let q be given by q(z)= 1 + z with q(0)=1. Then q is a convex function. Let us define ϕ(w)=β/ w 2 and

Reϕ [ q ( z ) ] =Re ( β 1 + z ) > β 2 >0.

Consider the function Q defined by

Q(z):=β z q ( z ) q 2 ( z ) =β z 2 ( 1 + z ) 3 2 .

Further,

Re ( z Q ( z ) Q ( z ) ) =1 3 2 Re ( z 1 + z ) 1 4 >0.

Thus the function Q is starlike, and the result now follows by an application of Lemma 1.3. □

In the next result, the condition on β is obtained such that p(z)+βz p (z)/p(z) 1 + z implies that p(z)(1+Az)/(1+Bz).

Lemma 2.8 Let 1B<A1, (AB)β 2 (1+|A|)(1+|B|)+ | A | 2 1 and

1 β max { 0 , A B ( 1 + | A | ) ( 1 + | B | ) 1 | B | 1 + | B | } .

Let p be an analytic function defined on D with p(0)=1 satisfying

p(z)+β z p ( z ) p ( z ) 1 + z .

Then p(z) 1 + A z 1 + B z .

Proof Define the function q:DC by q(z)=(1+Az)/(1+Bz), 1B<A1. Consider the subordination

p(z)+β z p ( z ) p ( z ) q(z)+β z q ( z ) q ( z ) .

Thus, in view of Lemma 1.2, the above subordination can be written as (1.1) by defining the functions ν and φ as ν(w):=w and φ(w):=β/w (β0). Clearly, the functions ν and φ are analytic in ℂ and φ(w)0. Let the functions Q(z) and h(z) be defined by

Q(z):=z q (z)φ ( q ( z ) ) =β z q ( z ) q ( z )

and

h(z):=ν ( q ( z ) ) +Q(z)=q(z)+β z q ( z ) q ( z ) .

A computation shows that Q(z) is starlike univalent in D. Further,

z h ( z ) Q ( z ) = 1 β +1+ z q ( z ) q ( z ) z q ( z ) q ( z ) .

Let z= e i t , πtπ. Then

Re ( e i t h ( e i t ) Q ( e i t ) ) = 1 β + Re ( 1 B e i t 1 + B e i t ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) ) 1 β + 1 | B | 1 + | B | A B ( 1 + | A | ) ( 1 + | B | ) > 0 .

Thus by Lemma 1.2, it follows that p(z)q(z). In order to prove our result, we need to prove that

Φ(z):= 1 + z q(z)+β z q ( z ) q ( z ) = 1 + A z 1 + B z + β ( A B ) z ( 1 + A z ) ( 1 + B z ) :=h(z).

The subordination Φ(z)h(z) is equivalent to the subordination z Φ 1 (h(z)). Now in order to prove the result, it is enough to show | Φ 1 (h( e i t ))|1, πtπ. Now

| Φ 1 ( h ( e i t ) ) |=| ( 1 + A e i t 1 + B e i t + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) ) 2 1|1

implies

| 1 + A e i t 1 + B e i t + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) | 2 .

Further,

| 1 + A e i t 1 + B e i t + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) | Re ( 1 + A e i t 1 + B e i t + β ( A B ) e i t ( 1 + A e i t ) ( 1 + B e i t ) ) 1 | A | 1 + | B | + ( A B ) β ( 1 + | A | ) ( 1 + | B | ) 2

for (AB)β 2 (1+|A|)(1+|B|)+ | A | 2 1. This completes the proof. □

3 Sufficient condition for Janowski starlikeness

The following first two results (Lemmas 3.1, 3.2) are essentially due to Ali et al. [[13], Lemmas 2.1, 2.10]. However, an alternate proof of the same result, which is much easier than that given by Ali et al. [13], is presented below.

Lemma 3.1 Assume that 1B<A1, 1E<D1 and β(AB)(DE)(1+ B 2 )+|2B(DE)Eβ(AB)|. Let p be an analytic function defined on D with p(0)=1 satisfying

1+βz p (z) 1 + D z 1 + E z ,β0.

Then p(z) 1 + A z 1 + B z .

Proof Define the function q:DC by

q(z)= 1 + A z 1 + B z ,1B<A1.

Then q is convex in D with q(0)=1. Further computation shows that

Q(z)=βz q (z)= β ( A B ) z ( 1 + B z ) 2

and Q is starlike in D. It follows from Lemma 1.1 that the subordination

1+βz p (z)1+βz q (z)

implies p(z)q(z). In view of the above result, it is sufficient to prove

1 + D z 1 + E z 1+βz q (z)=1+β ( A B ) z ( 1 + B z ) 2 =h(z).

Let w=Φ(z)= 1 + D z 1 + E z . Then Φ 1 (w)= w 1 D E w and

Φ 1 ( h ( z ) ) = β ( A B ) z D ( 1 + B z ) 2 E ( 1 + B z ) 2 β E ( A B ) z = β ( A B ) z ( D E ) ( 1 + B 2 z 2 ) + ( 2 B ( D E ) β E ( A B ) ) z .

Let z= e i t , πtπ. Thus

| Φ 1 ( h ( e i t ) ) | | β | ( A B ) ( D E ) ( 1 + B 2 ) + | ( 2 B ( D E ) β E ( A B ) ) | 1

for |β|(AB)(DE)(1+ B 2 )+|(2B(DE)Eβ(AB))|. Hence q(D)h(D), that is, q(z)h(z), this completes the proof. □

It should be noted that Ali et al. [13] made the assumption AB>0 in order to prove the result [[13], Lemma 2.10], whereas in the following lemma this condition has been dropped.

Lemma 3.2 Assume that 1B<A1, 1E<D1 and β(AB)(DE)(1+|AB|)+|(A+B)(DE)Eβ(AB)|. Let p be an analytic function defined on D with p(0)=1 satisfying

1+β z p ( z ) p ( z ) 1 + D z 1 + E z ,β0.

Then p(z) 1 + A z 1 + B z .

Proof As above, define the function q:DC by

q(z)= 1 + A z 1 + B z ,1B<A1.

Then q is convex in D with q(0)=1. A computation shows that

Q(z)= β z q ( z ) q ( z ) = β ( A B ) z ( 1 + A z ) ( 1 + B z )

and Q is starlike in D. It follows from Lemma 1.1 that the subordination

1+β z p ( z ) p ( z ) 1+β z q ( z ) q ( z )

implies p(z)q(z). Now we need to prove

1 + D z 1 + E z 1+β z q ( z ) q ( z ) =1+β ( A B ) z ( 1 + B z ) 2 =h(z).

Let w=Φ(z)= 1 + D z 1 + E z . Then Φ 1 (w)= w 1 D E w and

Φ 1 ( h ( z ) ) = β ( A B ) z ( D E ) ( 1 + A z ) ( 1 + B z ) β E ( A B ) z = β ( A B ) z ( D E ) ( 1 + A B z 2 ) + ( ( A + B ) ( D E ) β E ( A B ) ) z .

Let z= e i t , πtπ. Thus

| Φ 1 ( h ( e i t ) ) | | β | ( A B ) ( D E ) ( 1 + | A B | ) + | ( A + B ) ( D E ) β E ( A B ) | 1

for |β|(AB)(DE)(1+|AB|)+|(A+B)(DE)Eβ(AB)|. Hence q(D)h(D), that is, q(z)h(z), this completes the proof. □

Lemma 3.3 Assume that 1B<A1, 1E<D1 and |β|(AB)(DE)(1+ A 2 )+|2A(DE)Eβ(AB)|. Let p be an analytic function defined on D with p(0)=1 satisfying

1+β z p ( z ) p 2 ( z ) 1 + D z 1 + E z .

Then p(z) 1 + A z 1 + B z .

Proof Define the function q:DC by

q(z)= 1 + A z 1 + B z ,1B<A1.

Then q is convex in D with q(0)=1. A computation shows that

Q(z)= β z q ( z ) q 2 ( z ) = β ( A B ) z ( 1 + A z ) 2

and

z Q ( z ) Q ( z ) = 1 A z 1 + A z .

As before, a computation shows Q is starlike in D. It follows from Lemma 1.1 that the subordination

1+β z p ( z ) p 2 ( z ) 1+β z q ( z ) q 2 ( z )

implies p(z)q(z). To prove result, it is enough to show that

1 + D z 1 + E z 1+β z q ( z ) q 2 ( z ) =1+β ( A B ) z ( 1 + A z ) 2 =h(z).

The remaining part of the proof is similar to that of Lemma 3.1, and therefore it is skipped here. □

Remark 3.4 When β=1, Lemma 3.3 reduces to [[13], Lemma 2.6] due to Ali et al.