Let Tn(A,B,γ,α) (−1≤B<1,B<A,0<γ≤1 and α>0) denote the class of functions of the form f(z)=z+∑k=n+1∞akzk (n∈N={1,2,3,…}), which are analytic in the open unit disk U and satisfy the following subordination condition f′(z)+αzf′′(z)≺((1+Az)/(1+Bz))γ, for(z∈U;A≤1;0<γ<1),(1+Az)/(1+Bz), for(z∈U;γ=1). We obtain sharp bounds on Ref'(z),Ref(z)/z,|f(z)|, and coefficient estimates for functions f(z) belonging to the class Tn(A,B,γ,α). Conditions for univalency and starlikeness, convolution properties, and the radius of convexity are also considered.
1. Introduction
Let An denote the class of functions of the form f(z)=z+∑k=n+1∞akzk(n∈N={1,2,3,…}),
which are analytic in the open unit disk U={z:z∈Cand|z|<1}. Let Sn and Sn* denote the subclasses of An whose members are univalent and starlike, respectively.
For functions f(z) and g(z) analytic in U, we say that f(z) is subordinate to g(z) in U and we write f(z)≺g(z)(z∈U), if there exists an analytic function w(z) in U such that |w(z)|≤|z|,f(z)=g(w(z))(z∈U).
Furthermore, if the function g(z) is univalent in U, then f(z)≺g(z)(z∈U)⟺f(0)=g(0),f(U)⊂g(U).
Throughout our present discussion, we assume that n∈N,-1≤B<1,B<A,α>0,β<1,0<γ≤1.
We introduce the following subclass of An.
Definition 1.1.
A function f(z)∈An is said to be in the class Tn(A,B,γ,α) if it satisfies
f′(z)+αzf′′(z)≺h(z)(z∈U),
where
h(z)={(1+Az1+Bz)γ,(A≤1;0<γ<1),1+Az1+Bz,(γ=1).
The classes
T1(1-2β,-1,1,1)=R(β)(β=0orβ<1),T1(A,0,1,α)=R̃(α,A)(A>0)
have been studied by several authors (see [1–5]). Recently, Gao and Zhou [6] showed some mapping properties of the following subclass of A1:
R(β,α)={f(z)∈A1:Re{f′(z)+αzf′′(z)}>β(z∈U)}.
Note that
R(β,1)=R(β),T1(1-2β,-1,1,α)=R(β,α).
For further information of the above classes (with γ=1) and related analytic function classes, see Srivastava et al. [7], Yang and Liu [8], Kim [9], and Kim and Srivastava [10].
In this paper, we obtain sharp bounds on Ref′(z),Re(f(z)/z),|f(z)|, and coefficient estimates for functions f(z) belonging to the class Tn(A,B,γ,α). Conditions for univalency and starlikeness, convolution properties, and the radius of convexity are also presented. One can see that the methods used in [6] do not work for the more general class Tn(A,B,γ,α) than R(β,α).
2. The bounds on Ref′(z), Re(f(z)/z), and |f(z)| in Tn(A,B,γ,α)
In this section, we let λm(A,B,γ)={∑j=0m(γj)(-γm-j)AjBm-j,(A≤1;0<γ<1),(A-B)(-B)m-1,(γ=1),
where m∈N and (γj)={γ(γ-1)⋯(γ-j+1)j!,(j=1,2,…,m),1,(j=0).
With (2.1), it is easily seen that the function h(z) given by (1.6) can be expressed as h(z)=1+∑m=1∞λm(A,B,γ)zm(z∈U).
Theorem 2.1.
Let f(z)∈Tn(A,B,γ,α). Then, for |z|=r<1,
Ref′(z)≥1+∑m=1∞(-1)mλm(A,B,γ)αnm+1rnm,Ref′(z)≤1+∑m=1∞λm(A,B,γ)αnm+1rnm.
The bounds in (2.4) are sharp for the function fn(z) defined by
fn(z)=z+∑m=1∞λm(A,B,γ)(nm+1)(αnm+1)znm+1(z∈U).
Proof.
The analytic function h(z) given by (1.6) is convex (univalent) in U (cf. [11]) and satisfies h(z¯)=h(z)¯(z∈U). Thus, for |ζ|≤σ(ζ∈Candσ<1),
h(-σ)≤Reh(ζ)≤h(σ).
Let f(z)∈Tn(A,B,γ,α). Then, we can write
f′(z)+αzf′′(z)=h(w(z))(z∈U),
where w(z)=wnzn+wn+1zn+1+⋯ is analytic and |w(z)|<1 for z∈U. By the Schwarz lemma, we know that |w(z)|≤|z|n(z∈U). It follows from (2.7) that
(z1/αf′(z))′=1αz(1/α)-1h(w(z)),
which leads to
f′(z)=1αz-1/α∫0zζ(1/α)-1h(w(ζ))dζ
or to
f′(z)=1α∫01t(1/α)-1h(w(tz))dt(z∈U).
Since
|w(tz)|≤(tr)n(|z|=r<1;0≤t≤1),
we deduce from (2.6) and (2.10) that
1α∫01t(1/α)-1h(-(tr)n)dt≤Ref′(z)≤1α∫01t(1/α)-1h((tr)n)dt.
Now, by using (2.3) and (2.12), we can obtain (2.4).
Furthermore, for the function fn(z) defined by (2.5), we find that
fn′(z)=1+∑m=1∞λm(A,B,γ)αnm+1znm,fn′(z)+αzfn′′(z)=1+∑m=1∞λm(A,B,γ)znm=h(zn)≺h(z)(z∈U).
Hence, fn(z)∈Tn(A,B,γ,α) and from (2.13), we see that the bounds in (2.4) are the best possible.
Hereafter, we write
Tn(A,B,1,α)=Tn(A,B,α).
Corollary 2.2.
Let f(z)∈Tn(A,B,α). Then, for z∈U,
Ref′(z)>1-(A-B)∑m=1∞Bm-1αnm+1,Ref′(z)<1+(A-B)∑m=1∞(-B)m-1αnm+1(B≠-1).
The results are sharp.
Proof.
For γ=1, it follows from (2.12) (used in the proof of Theorem 2.1) that
Ref′(z)>1α∫01t(1/α)-1(1-Atn1-Btn)dt,Ref′(z)<1α∫01t(1/α)-1(1+Atn1+Btn)dt(B≠-1),
for z∈U. From these, we have the desired results.
The bounds in (2.16) and (2.17) are sharp for the function fn(z)=z+(A-B)∑m=1∞(-B)m-1(nm+1)(αnm+1)znm+1∈Tn(A,B,α).
Theorem 2.3.
Let f(z)∈Tn(A,B,γ,α). Then, for |z|=r<1,
Ref(z)z≥1+∑m=1∞(-1)mλm(A,B,γ)(nm+1)(αnm+1)rnm,Ref(z)z≤1+∑m=1∞λm(A,B,γ)(nm+1)(αnm+1)rnm.
The results are sharp.
Proof.
Noting that
f(z)=z∫01f′(uz)du,Ref(z)z=∫01Ref′(uz)du(z∈U),
an application of Theorem 2.1 yields (2.20). Furthermore, the results are sharp for the function fn(z) defined by (2.5).
Corollary 2.4.
Let f(z)∈Tn(A,B,α). Then, for z∈U,
Ref(z)z>1-(A-B)∑m=1∞Bm-1(nm+1)(αnm+1),Ref(z)z<1+(A-B)∑m=1∞(-B)m-1(nm+1)(αnm+1).
The results are sharp for the function fn(z) defined by (2.19).
Proof.
For f(z)∈Tn(A,B,α), it follows from (2.6) and (2.10) (with γ=1) that
1α∫01t(1/α)-1(1-A(ut)n1-B(ut)n)dt<Ref′(uz)<1α∫01t(1/α)-1(1+A(ut)n1+B(ut)n)dt,
for z∈U and 0<u≤1. Making use of (2.21) and (2.23), we can obtain (2.22).
Theorem 2.5.
Let f(z)∈T1(A,B,α) and g(z)∈T1(A0,B0,α0)(-1≤B0<1,B0<A0 and α0>0). If
(A0-B0)∑m=1∞B0m-1(m+1)(α0m+1)≤12,
then (f*g)(z)∈T1(A,B,α), where the symbol * stands for the familiar Hadamard product (or convolution) of two analytic functions in U.
Proof.
Since g(z)∈T1(A0,B0,α0)(-1≤B0<1,B0<A0 and α0>0), it follows from Corollary 2.4 (with n=1) and (2.24) that
Reg(z)z>1-(A0-B0)∑m=1∞B0m-1(m+1)(α0m+1)≥12(z∈U).
Thus, g(z)/z has the Herglotz representation
g(z)z=∫|x|=1dμ(x)1-xz(z∈U),
where μ(x) is a probability measure on the unit circle |x|=1 and ∫|x|=1dμ(x)=1.
For f(z)∈T1(A,B,α), we have
(f*g)′(z)+αz(f*g)′′(z)=F(z)*g(z)z(z∈U),
where
F(z)=f′(z)+αzf′′(z)≺1+Az1+Bz(z∈U).
In view of the function (1+Az)/(1+Bz) is convex (univalent) in U, we deduce from (2.26) to (2.28) that
(f*g)′(z)+αz(f*g)′′(z)=∫|x|=1F(xz)dμ(x)≺1+Az1+Bz(z∈U).
This shows that (f*g)(z)∈T1(A,B,α).
Corollary 2.6.
Let f(z)∈T1(A,B,α), g(z)∈R(β,1) and
β≥-π2-912-π2.
Then, (f*g)(z)∈T1(A,B,α).
Proof.
By taking A0=1-2β, B0=-1 and α0=1, (2.24) in Theorem 2.5 becomes
2(1-β)∑m=1∞(-1)m-1(m+1)2=2(1-β)(1-π212)≤12,
that is,
β≥-π2-912-π2.
Hence, the desired result follows as a special case from Theorem 2.5.
Remark 2.7.
R. Singh and S. Singh [4, Theorem 3] proved that, if f(z) and g(z) belong to R(0,1), then (f*g)(z)∈R(0,1). Obviously, for
-π2-912-π2≤β<0,
Corollary 2.6 generalizes and improves Theorem 3 in [4].
Theorem 2.8.
Let f(z)∈Tn(A,B,γ,α) and AB≤1. Then, for |z|=r<1,
|f(z)|≤r+∑m=1∞λm(A,B,γ)(nm+1)(αnm+1)rnm+1.
The result is sharp, with the extremal function fn(z) defined by (2.5).
Proof.
It is well known that for ζ∈C and |ζ|≤σ<1,
|1+Aζ1+Bζ-1-ABσ21-B2σ2|≤(A-B)σ1-B2σ2.
Since AB≤1, we have 1-ABσ2>0 and so (2.35) leads to
|1+Aζ1+Bζ|γ≤(|1-ABσ21-B2σ2|+(A-B)σ1-B2σ2)γ=(1+Aσ1+Bσ)γ(|ζ|≤σ<1).
By virtue of (1.6), (2.10), and (2.36), we have
|f′(uz)|≤1α∫01t(1/α)-1|h(w(utz))|dt≤1α∫01t(1/α)-1h((ut|z|)n)dt,
for z∈U and 0≤u≤1. Now, by using (2.3), (2.21) and (2.37), we can obtain (2.34).
Theorem 2.9.
Let
f(z)=z+∑k=n+1∞akzk∈Tn(A,B,γ,α).
Then,
|ak|≤γ(A-B)k(α(k-1)+1)(k≥n+1).
The result is sharp for each k≥n+1.
Proof.
It is known (cf. [12]) that, if
φ(z)=∑k=1∞bkzk≺ψ(z)(z∈U),
where φ(z) is analytic in U and ψ(z)=z+⋯ is analytic and convex univalent in U, then |bk|≤1(k∈N).
By (2.38), we have
f′(z)+αzf′′(z)-1γ(A-B)=∑k=n+1∞k(α(k-1)+1)γ(A-B)akzk-1≺ψ(z)(z∈U),
where
ψ(z)=h(z)-1γ(A-B)=z+⋯
and h(z) is given by (1.6). Since the function ψ(z) is analytic and convex univalent in U, it follows from (2.41) that
k(α(k-1)+1)γ(A-B)|ak|≤1(k≥n+1),
which gives (2.39).
Next, we consider the function
fk-1(z)=z+∑m=1∞λm(A,B,γ)(m(k-1)+1)(αm(k-1)+1)zm(k-1)+1(z∈U;k≥n+1).
It is easy to verify that
fk-1′(z)+αzfk-1′′(z)=h(zk-1)≺h(z)(z∈U),fk-1(z)=z+γ(A-B)k(α(k-1)+1)zk+⋯.
The proof of Theorem 2.9 is completed.
3. The Univalency and Starlikeness of Tn(A,B,α)Theorem 3.1.
Tn(A,B,α)⊂Sn if and only if
(A-B)∑m=1∞Bm-1αnm+1≤1.
Proof.
Let f(z)∈Tn(A,B,α) and (3.1) be satisfied. Then, by (2.16) in Corollary 2.2, we see that Ref′(z)>0(z∈U). Thus, f(z) is close-to-convex and univalent in U.
On the other hand, if
(A-B)∑m=1∞Bm-1αnm+1>1,
then the function fn(z) defined by (2.19) satisfies fn′(0)=1>0 and
fn′(reπi/n)=1-(A-B)∑m=1∞Bm-1αnm+1rnm⟶1-(A-B)∑m=1∞Bm-1αnm+1<0
as r→1. Hence, there exists a point zn=rneπi/n(0<rn<1) such that fn′(zn)=0. This implies that fn(z) is not univalent in U and so the theorem is proved.
Theorem 3.2.
Let (3.1) in Theorem 3.1 be satisfied. If α≥1 and
(α-1)(1-(A-B)∑m=1∞Bm-1αnm+1)+nα2(1-(A-B)∑m=1∞Bm-1(nm+1)(αnm+1))≥A-11-B,
then Tn(A,B,α)⊂Sn*.
Proof.
We first show that
∑m=1∞Bm-1αnm+1≥∑m=1∞Bm-1(nm+1)(αnm+1)(α≥1).
Equation (3.5) is obvious when B≥0. For 0>B≥-1, we have
∑m=1∞Bm-1αnm+1-∑m=1∞Bm-1(nm+1)(αnm+1)=μ1-μ2+μ3-μ4+⋯+(-1)m-1μm+⋯,
where
μm=nm|B|m-1(nm+1)(αnm+1)>0.
Since |B|≤1 and
ddx(x(x+1)(αx+1))=1-αx2(x+1)2(αx+1)2≤0(x≥1;α≥1),{μm} is a monotonically decreasing sequence. Therefore, the inequality (3.5) follows from (3.6).
Let f(z)∈Tn(A,B,α). Then,
Re{f′(z)+αzf′′(z)}>1-A1-B(z∈U).
Define p(z) in U by
p(z)=zf′(z)f(z).
In view of (3.1) in Theorem 3.1 is satisfied, the function f(z) is univalent in U, and so p(z)=1+pnzn+pn+1zn+1+⋯ is analytic in U. Also it follows from (3.10) that
f′(z)+αzf′′(z)=(1-α)f′(z)+αf(z)z[zp′(z)+(p(z))2].
We want to prove now that Rep(z)>0 for z∈U. Suppose that there exists a point z0∈U such that Rep(z)>0(|z|<|z0|),Rep(z0)=0.
Then, applying a result of Miller and Mocanu [13, Theorem 4], we havez0p′(z0)+(p(z0))2≤-n2Re(1-p(z0))-(Imp(z0))2≤-n2.
For α≥1, we deduce from Corollaries 2.2 and 2.4, (3.1), (3.5), (3.11), (3.13), and (3.4) that Re{f′(z0)+αz0f′′(z0)}≤(1-α)(1-(A-B)∑m=1∞Bm-1αnm+1)-nα2(1-(A-B)∑m=1∞Bm-1(nm+1)(αnm+1))≤1-A1-B.
But this contradicts (3.9) at z=z0. Therefore, we must have Rep(z)>0(z∈U) and the proof of Theorem 3.2 is completed.
Remark 3.3.
In [6, Theorem 4(ii)], the authors gave the following: if 0<α<1 and β1 is the solution of the equation
1-3α2=β+(1-β)∑m=2∞(-1)m-1α+2(α-1)mm(α(m-1)+1),
then R(β,α)⊂S1* for β≥β1. However, this result is not true because the series in (3.15) diverges.
4. The Radius of ConvexityTheorem 4.1.
Let f(z) belong to the class Tn(γ) defined by
Tn(γ)=Tn(1,-1,γ,0)={f(z)∈An:f′(z)≺(1+z1-z)γ,(z∈U)},0<δ≤1 and 0≤ρ<1. Then,
Re{(1-δ)(f′(z))1/γ+δ(1+zf′′(z)f′(z))}>ρ(|z|<rn(γ,δ,ρ)),
where rn(γ,δ,ρ) is the root in (0,1) of the equation
(1-2δ+ρ)r2n-2(1-δ+nδγ)rn+1-ρ=0.
The result is sharp.
Proof.
For f(z)∈Tn(γ), we can write
(f′(z))1/γ=1+znφ(z)1-znφ(z),
where φ(z) is analytic and |φ(z)|≤1 in U. Differentiating both sides of (4.4) logarithmically, we arrive at
1+zf′′(z)f′(z)=1+2nγznφ(z)1-(znφ(z))2+2γzn+1φ′(z)1-(znφ(z))2(z∈U).
Put |z|=r<1 and (f′(z))1/γ=u+iv(u,v∈R). Then, (4.4) implies that
znφ(z)=u-1+ivu+1+iv,1-rn1+rn≤u≤1+rn1-rn.
With the help of the Carathéodory inequality
|φ′(z)|≤1-|φ(z)|21-r2,
it follows from (4.5) and (4.6) that
Re{(1-δ)(f′(z))1/γ+δ(1+zf′′(z)f′(z))}≥(1-δ)u+δ+2nδγRe{znφ(z)1-(znφ(z))2}-2δγ|zn+1φ′(z)1-(znφ(z))2|≥(1-δ)u+δ+nδγ2(u-uu2+v2)+δγ2(u-1)2+v2-r2n((u+1)2+v2)rn-1(1-r2)(u2+v2)1/2=Fn(u,v)(say),∂∂vFn(u,v)=δγvGn(u,v),
where 0<r<1,0<δ≤1 and
Gn(u,v)=nu(u2+v2)2+1-r2nrn-1(1-r2)(u2+v2)1/2+r2n((u+1)2+v2)-((u-1)2+v2)2rn-1(1-r2)(u2+v2)3/2>0
because of (4.6) and (4.7). In view of (4.10) and (4.11), we see that
Fn(u,v)≥Fn(u,0)=(1-δ)u+δ+nδγ2(u-1u)+δγ2rn-1(1-r2)×{(1-r2n)(u+1u)-2(1+r2n)}.
Let us now calculate the minimum value of Fn(u,0) on the closed interval [(1-rn)/(1+rn),(1+rn)/(1-rn)]. Noting that 1-r2nrn-1(1-r2)≥n(see[8])
and (4.7), we deduce from (4.12) that dduFn(u,0)=1-δ+δγ2[(1-r2nrn-1(1-r2)+n)-1u2(1-r2nrn-1(1-r2)-n)]≥1-δ+δγ2[(1-r2nrn-1(1-r2)+n)-(1+rn1-rn)2(1-r2nrn-1(1-r2)-n)]=1-δ+2δγIn(r)(1-rn)2,
where In(r)=n2(1+r2n)-r(1+r2+⋯+r2n-2).
Also In′(r)=n2r2n-1-(1+3r2+⋯+(2n-1)r2n-2)
and I1′(r)=r-1<0. Suppose that In′(r)<0. Then,In+1′(r)=(n+1)2r2n+1-(2n+1)r2n-(1+3r2+⋯+(2n-1)r2n-2)<n2r2n-(1+3r2+⋯+(2n-1)r2n-2)<In′(r)<0.
Hence, by virtue of the mathematical induction, we have In′(r)<0 for all n∈N and 0≤r<1. This implies that In(r)>In(1)=0(n∈N;0≤r<1).
In view of (4.14) and (4.18), we see that dduFn(u,0)>0(1-rn1+rn≤u≤1+rn1-rn).
Further it follows from (4.9), (4.12), and (4.19) thatRe{(1-δ)(f′(z))1/γ+δ(1+zf′′(z)f′(z))}-ρ≥Fn(1-rn1+rn,0)-ρ=(1-δ)1-rn1+rn+δ1-2nγrn-r2n1-r2n-ρ=Jn(r)1-r2n,
where 0≤ρ<1 and Jn(r)=(1-2δ+ρ)r2n-2(1-δ+nδγ)rn+1-ρ.
Note that Jn(0)=1-ρ>0 and Jn(1)=-2nδγ<0. If we let rn(γ,δ,ρ) denote the root in (0,1) of the equation Jn(r)=0, then (4.20) yields the desired result (4.2).
To see that the bound rn(γ,δ,ρ) is the best possible, we consider the function f(z)=∫0z(1-tn1+tn)γdt∈Tn(γ).
It is clear that for z=r∈(rn(γ,δ,ρ),1), (1-δ)(f′(r))1/γ+δ(1+rf′′(r)f′(r))-ρ=Jn(r)1-r2n<0,
which shows that the bound rn(γ,δ,ρ) cannot be increased.
Setting δ=1, Theorem 4.1 reduces to the following result.
Corollary 4.2.
Let f(z)∈Tn(γ) and 0≤ρ<1. Then, f(z) is convex of order ρ in
|z|<[((nγ)2+(1-ρ)2)1/2-nγ1-ρ]1/n.
The result is sharp.
AliR. M.On a subclass of starlike functions1994242447451127733710.1216/rmjm/1181072410ChichraP. N.New subclasses of the class of close-to-convex functions197762137430425097SilvermanH.A class of bounded starlike functions1994172249252126107010.1155/S0161171294000360ZBL0796.30006SinghR.SinghS.Convolution properties of a class of starlike functions1989106114515299438810.1090/S0002-9939-1989-0994388-6ZBL0672.30007PonnusamyS.SinghV.Criteria for strongly starlike functions19973432672911487463ZBL0892.30005GaoC. Y.ZhouS. Q.Certain subclass of starlike functions20071871176182232356710.1016/j.amc.2006.08.113ZBL1130.30010SrivastavaH. M.XuN. E.YangD. G.Inclusion relations and convolution properties of a certain class of analytic functions associated with the Ruscheweyh derivatives20073311686700230603210.1016/j.jmaa.2006.09.019ZBL1111.30011YangD. G.LiuJ. L.On a class of analytic functions with missing coefficients2010215934733481257683810.1016/j.amc.2009.10.043ZBL1182.30025KimY. C.Mapping properties of differential inequalities related to univalent functions20071871272279232357910.1016/j.amc.2006.08.124ZBL1115.30004KimY. C.SrivastavaH. M.Some applications of a differential subordination1999223649654171734310.1155/S016117129922649XZBL0960.30009SrivastavaH. M.YangD. G.XuN. E.Some subclasses of meromorphically multivalent functions associated with a linear operator200819511123237919110.1016/j.amc.2007.04.065ZBL1175.30028DurenP. L.1983259New York, NY, USASpringerxiv+382Fundamental Principles of Mathematical Sciences708494MillerS. S.MocanuP. T.Second order differential inequalities in the complex plane197865228930550630710.1016/0022-247X(78)90181-6ZBL0367.34005