betaFeedback

WorksMathematics

Monotonicity and bounded variation for harmonic mappings with applications to special functions

Abstract

We investigate sufficient conditions for the close-to-convexity and univalence of planar harmonic mappings. First, we establish a unified monotonicity criterion for harmonic mappings that incorporates the ordinary and odd harmonic cases. We then develop a criterion based on the total variation of successive coefficient differences. Unlike conventional monotonicity conditions, this approach allows the coefficient sequences to exhibit non-monotone behavior. Several previously established criteria for harmonic close-to-convexity are recovered as particular consequences. Applications are developed for harmonic mappings associated with Gaussian hypergeometric functions and Legendre polynomials, with explicit restrictions on their parameters.

Introduction

Coefficient criteria for univalence form a standard topic in geometric function theory. Among the different approaches developed for this purpose, the results expressed directly in terms of its series coefficients are useful, since they often permit geometric conclusions without requiring an explicit description of the image domain.

Let D={z∈C:∣z∣<1}\mathbb{D}=\{z\in\mathbb{C}: |z|<1\} denote the open unit disk, and let A\mathcal{A} be the class of analytic functions normalized by h(z)=z+∑n=2∞anznh(z)=z+\sum_{n=2}^{\infty}a_nz^n, z∈Dz\in\mathbb{D}. We denote by S\mathcal{S}, the subclass of univalent functions in A\mathcal{A}. The study of coefficient criteria ensuring membership of f∈Af\in\mathcal{A} in S\mathcal{S} has a long history. Alexander [1] investigated the connection between geometric properties of analytic functions and their derivatives. In subsequent works, Fejér, Szegő and Robertson [6, 20, 15] extended these results by examining sequences of real numbers with monotone properties.

A complex-valued harmonic function ff defined in a simply connected domain admits a canonical decomposition f=h+g‾f=h+\overline{g}, where hh and gg are analytic. Unlike analytic mappings, harmonic mappings need not preserve orientation, and their local univalence depends on derivatives of both hh and gg. Therefore, coefficient conditions that ensure univalence in the analytic case do not automatically extend to harmonic mappings.

For a normalized harmonic mapping

f(z)=z+∑n=2∞anzn+∑n=2∞bnzn‾f(z)=z+\sum_{n=2}^{\infty}a_nz^n+\overline{\sum_{n=2}^{\infty}b_nz^n}

the associated analytic family is Fε(z)=h(z)+εg(z)F_{\varepsilon}(z)=h(z)+\varepsilon g(z), ∣ε∣=1|\varepsilon|=1. If members of this family are close-to-convex, the corresponding harmonic mappings are close-to-convex and univalent. Recently, Sahoo and Wankhede [16] established several sufficient conditions involving monotonicity of coefficients and their successive differences. They also obtained applications to Gaussian hypergeometric functions. However, many power series may fail to be monotone even when the corresponding functions possess geometric properties. This observation motivates the search for criteria that control the total variation rather than requiring successive differences to have the same sign. A related motivation comes from lacunary power series. Consider the harmonic mapping

fp(z)=z+∑n=1∞apn+1zpn+1+∑n=1∞bpn+1zpn+1‾f_p(z)=z+\sum_{n=1}^{\infty}a_{pn+1}z^{pn+1}+\overline{\sum_{n=1}^{\infty}b_{pn+1}z^{pn+1}}

where p=1p=1 or p=2p=2. The choice p=1p=1 corresponds to an unrestricted normalized harmonic mapping, whereas p=2p=2 gives an odd harmonic mapping. Although these cases have been studied separately, their coefficients suggest that a common argument may be developed by introducing the weighted sequences cn=(pn+1)apn+1c_n=(pn+1)a_{pn+1} and dn=(pn+1)bpn+1d_n=(pn+1)b_{pn+1}. This formulation allows the two cases to be studied through a single coefficient inequality.

For complex numbers aa, bb and cc with c∉{0,−1,−2,…}c\notin\{0,-1,-2,\ldots\}, the Gaussian hypergeometric function 2F1(a,b;c;z){}_2F_1(a,b;c;z) is defined by the series

2F1(a,b;c;z)=∑n=0∞(a)n(b)n(c)nznn!(1){}_2F_1(a,b;c;z)=\sum_{n=0}^{\infty}\frac{(a)_n(b)_n}{(c)_n}\frac{z^n}{n!} \tag*{(1)}

where (α)n(\alpha)_n is the Pochhammer symbol defined as (α)0=1(\alpha)_0=1 and (α)n=α(α+1)⋯(α+n−1)(\alpha)_n=\alpha(\alpha+1)\cdots(\alpha+n-1) with n≥1n\geq1. Also note that

2F1′(a,b;c;z)=abc2F1(a+1,b+1;c+1;z).(2){}_2F_1'(a,b;c;z)=\frac{ab}{c}{}_2F_1(a+1,b+1;c+1;z). \tag*{(2)}

The Gaussian hypergeometric function contains many classical special functions as particular cases. In particular, the hypergeometric representation of the Legendre function is given by

Pν(x)=2F1(−ν,ν+1;1;1−x2).(3)P_{\nu}(x)={}_2F_1\left(-\nu,\nu+1;1;\frac{1-x}{2}\right). \tag*{(3)}

Putting ν=−a\nu=-a, x=1−2zx=1-2z, we obtain 2F1(a,1−a;1;z)=P−a(1−2z){}_2F_1(a,1-a;1;z)=P_{-a}(1-2z). The basic information on hypergeometric functions and their special cases may be found in [2].

The coefficients of hypergeometric functions possess explicit recurrence relations, while their asymptotic behavior can be expressed through gamma-function identities. The geometric theory of hypergeometric functions has been investigated recently by several researchers (see, for example, [3, 8, 9, 12, 13, 14, 17, 21]). More recently, investigations of shifted hypergeometric functions and their geometric properties have further emphasized the importance of understanding the influence of parameters on univalence and convexity (see [10, 18, 19]).

In this paper, we first use the monotonicity of weighted coefficients to provide a unified close-to-convexity result for ordinary and odd harmonic mappings. In another result, we show that the total variation of weighted coefficient differences yields a criterion that does not require monotonicity. Recall that the total variation of a sequence {xn}n≥1\{x_n\}_{n\ge1} is defined by V(x)=∑n=1∞∣xn+1−xn∣V(x)=\sum_{n=1}^{\infty}|x_{n+1}-x_n|. The sequence is said to have bounded variation if V(x)<∞V(x)<\infty. Every bounded-variation sequence converges, but such a sequence need not be monotone. In particular, the condition V(x)<∞V(x)<\infty permits successive differences to change sign infinitely often, provided their absolute values are summable. This observation is essential for extending monotonicity-based criteria to a wider family of harmonic mappings.

Let H\mathcal{H} denote the normalized class of harmonic mappings f(z)=h+g‾f(z)=h+\overline{g} in D\mathbb{D} satisfying h(0)=g(0)=0h(0)=g(0)=0, h′(0)=1h'(0)=1, g′(0)=0g'(0)=0. Thus

h(z)=z+∑n=2∞anzn,andg(z)=∑n=2∞bnzn,z∈D.h(z)=z+\sum_{n=2}^{\infty}a_nz^n,\qquad\text{and}\qquad g(z)=\sum_{n=2}^{\infty}b_nz^n,\qquad z\in\mathbb{D}.

The Jacobian of ff is Jf(z)=∣h′(z)∣2−∣g′(z)∣2J_f(z)=|h'(z)|^2-|g'(z)|^2. The mapping ff is locally sense-preserving when Jf(z)>0J_f(z)>0 in D\mathbb{D}. Whenever h′≠0h'\ne0 in D\mathbb{D}, the analytic dilatation is defined by ωf(z)=g′(z)/h′(z)\omega_f(z)=g'(z)/h'(z). Hence the sense-preserving condition becomes ∣ωf(z)∣<1|\omega_f(z)|<1.

An analytic univalent function hh is said to be convex in D\mathbb{D} if h(D)h(\mathbb{D}) is a convex domain. The standard analytic characterization for convex functions in D\mathbb{D} is f′(0)≠0f'(0)\ne0 and Re⁡(1+zf′′(z)/f′(z))>0\operatorname{Re}(1+zf''(z)/f'(z))>0, z∈Dz\in\mathbb{D}. A function f∈Af\in\mathcal{A} is said to be close-to-convex, if there exist a convex function g∈Ag\in\mathcal{A} and a real number θ\theta such that Re⁡(eiθf′(z)/g′(z))>0\operatorname{Re}(e^{i\theta}f'(z)/g'(z))>0. Every close-to-convex function is univalent, although the converse does not hold. The basic information on analytic convexity and close-to-convexity may be found in [5, 7].

The convex functions ϕ1(z)=−log⁡(1−z)\phi_1(z)=-\log(1-z) and ϕ2(z)=(1/2)log⁡[(1+z)/(1−z)]\phi_2(z)=(1/2)\log[(1+z)/(1-z)] play an important role in our arguments. We unify these functions by

ϕp(z)=∫0z11−tp dt(z∈D, p=1,2).\phi_p(z)=\int_0^z\frac{1}{1-t^p}\,dt\qquad(z\in\mathbb{D},\ p=1,2).

Clearly, ϕp\phi_p is convex for p={1,2}p=\{1,2\}.

We shall need the following result to prove the main results:

Lemma 1.1 [Clunie–Sheil-Small [4]]. Let hh and gg be analytic in D\mathbb{D}, with ∣h′(0)∣>∣g′(0)∣|h'(0)|>|g'(0)|. If the analytic function Fε=h+εgF_\varepsilon=h+\varepsilon g is close-to-convex in D\mathbb{D} for every ∣ε∣=1|\varepsilon|=1, then f=h+g‾f=h+\overline{g} is a close-to-convex and univalent in D\mathbb{D}.

Main Results

The following theorem combines coefficient criteria for ordinary harmonic functions and odd harmonic functions.

Theorem 2.1 Let p∈{1,2}p \in\{1,2\} and let {apn+1}n≥0\{a_{pn+1}\}_{n\geq0} and {bpn+1}n≥0\{b_{pn+1}\}_{n\geq0} be sequences of real numbers with a1=1a_1=1 and b1=0b_1=0. Suppose that {(pn+1)apn+1}n≥0\{(pn+1)a_{pn+1}\}_{n\geq0} and {(pn+1)bpn+1}n≥0\{(pn+1)b_{pn+1}\}_{n\geq0} are non-decreasing sequences and lim⁡n→∞(pn+1)(apn+1+bpn+1)≤2\lim_{n\to\infty}(pn+1)(a_{pn+1}+b_{pn+1})\leq2. Then the harmonic function

fp(z)=z+∑n=1∞apn+1zpn+1+∑n=1∞bpn+1zpn+1‾=hp(z)+gp(z)‾f_p(z)=z+\sum_{n=1}^{\infty}a_{pn+1}z^{pn+1}+\overline{\sum_{n=1}^{\infty}b_{pn+1}z^{pn+1}}=h_p(z)+\overline{g_p(z)}

is close-to-convex and univalent in D\mathbb{D}.

Proof. Let cn=(pn+1)apn+1c_n=(pn+1)a_{pn+1}, dn=(pn+1)bpn+1d_n=(pn+1)b_{pn+1}, n≥0n\geq0, where c0=1c_0=1 and d0=0d_0=0. By hypothesis, cn≥1c_n\geq1, dn≥0d_n\geq0, and 1≤cn+dn≤21\leq c_n+d_n\leq2. Thus the sequences apn+1a_{pn+1} and bpn+1b_{pn+1} are bounded, hence the series for hh and gg converge locally uniformly in D\mathbb{D}, and fp=hp+gp‾f_p=h_p+\overline{g_p} is harmonic in D\mathbb{D}.

For every ε∈C\varepsilon\in\mathbb{C} with ∣ε∣=1|\varepsilon|=1, consider Fε(z)=hp(z)+εgp(z)F_\varepsilon(z)=h_p(z)+\varepsilon g_p(z). Hence

(1−zp)Fε′(z)=1+∑n=1∞[(cn−cn−1)+ε(dn−dn−1)]zpn.(1-z^p)F_\varepsilon'(z)=1+\sum_{n=1}^{\infty}[(c_n-c_{n-1})+\varepsilon(d_n-d_{n-1})]z^{pn}.

For N≥1N\geq1, define

SN(z)=1+∑n=1N[(cn−cn−1)+ε(dn−dn−1)]zpn.S_N(z)=1+\sum_{n=1}^{N}[(c_n-c_{n-1})+\varepsilon(d_n-d_{n-1})]z^{pn}.

Since the coefficient sequences {cn}n≥1\{c_n\}_{n\geq1} and {dn}n≥1\{d_n\}_{n\geq1} are non-decreasing, for z∈Dz\in\mathbb{D},

Re⁡SN(z)≥1−∑n=1N∣cn−cn−1∣∣z∣pn−∑n=1N∣dn−dn−1∣∣z∣pn\operatorname{Re}S_N(z)\geq1-\sum_{n=1}^{N}|c_n-c_{n-1}||z|^{pn}-\sum_{n=1}^{N}|d_n-d_{n-1}||z|^{pn}
≥1−∑n=1N(cn−cn−1)−∑n=1N(dn−dn−1)=2−(cN+dN)≥0.\begin{aligned} \geq1-\sum_{n=1}^{N}(c_n-c_{n-1})-\sum_{n=1}^{N}(d_n-d_{n-1}) \\ &=2-(c_N+d_N)\geq0. \end{aligned}

Choosing N→∞N\to\infty, we obtain Re⁡{(1−zp)Fε′(z)}>0\operatorname{Re}\{(1-z^p)F_\varepsilon'(z)\}>0, z∈Dz\in\mathbb{D}. Now define ϕp(z)=∫0z(1−tp)−1 dt\phi_p(z)=\int_0^z(1-t^p)^{-1}\,dt. Then ϕp\phi_p is convex for p=1,2p=1,2. Moreover

Re⁡{Fε′(z)ϕp′(z)}=Re⁡{(1−zp)Fε′(z)}>0.\operatorname{Re}\left\{\frac{F_\varepsilon'(z)}{\phi_p'(z)}\right\}=\operatorname{Re}\{(1-z^p)F_\varepsilon'(z)\}>0.

Hence, FεF_\varepsilon is close-to-convex for every ∣ε∣=1|\varepsilon|=1. Now the use of Lemma 1.1 shows that the harmonic mapping fpf_p is close-to-convex and univalent in D\mathbb{D}.

To prove the sharpness, we assume for every n≥1n\geq1, apn+1=1/(pn+1)a_{pn+1}=1/(pn+1), bpn+1=(1+δ)/(pn+1)b_{pn+1}=(1+\delta)/(pn+1). Thus both weighted sequences {cn}\{c_n\} and {dn}\{d_n\} are non-decreasing, and lim⁡n→∞(pn+1)(apn+1+bpn+1)=2+δ\lim_{n\to\infty}(pn+1)(a_{pn+1}+b_{pn+1})=2+\delta. The corresponding harmonic mapping is fp,δ(z)=hp(z)+gp,δ(z)‾f_{p,\delta}(z)=h_p(z)+\overline{g_{p,\delta}(z)}, where

hp(z)=∑n=0∞zpn+1pn+1=∫0zdt1−tpandgp,δ(z)=(1+δ)∑n=1∞zpn+1pn+1=(1+δ)(hp(z)−z).h_p(z)=\sum_{n=0}^{\infty}\frac{z^{pn+1}}{pn+1}=\int_0^z\frac{dt}{1-t^p} \qquad\text{and}\qquad g_{p,\delta}(z)=(1+\delta)\sum_{n=1}^{\infty}\frac{z^{pn+1}}{pn+1}=(1+\delta)(h_p(z)-z).

Therefore, the Jacobian is

Jfp,δ(z)=∣hp′(z)∣2−∣gp,δ′(z)∣2=1−(1+δ)2∣z∣2p∣1−zp∣2.J_{f_{p,\delta}}(z)=\lvert h'_{p}(z)\rvert^{2}-\lvert g'_{p,\delta}(z)\rvert^{2}=\frac{1-(1+\delta)^{2}\lvert z\rvert^{2p}}{\lvert1-z^{p}\rvert^{2}}.

If we choose rδ=(1+δ)−1/pr_{\delta}=(1+\delta)^{-1/p}, then 0<rδ<10<r_{\delta}<1. Moreover at z=rδz=r_{\delta}, we have Jfp,δ(rδ)=0J_{f_{p,\delta}}(r_{\delta})=0. Hence Jfp,δJ_{f_{p,\delta}} vanishes inside D\mathbb{D}. Therefore ff is not univalent in D\mathbb{D}. Since δ>0\delta>0 was arbitrary, the limiting constant 2 cannot be increased. Therefore, the constant 2 is sharp for p=1,2p=1,2.

Remark 2.2

(i) Choosing p=1p=1 in Theorem 2.1 gives the criterion for ordinary harmonic mappings [16 Theorem 3.2].

(ii) Choosing p=2p=2 in Theorem 2.1 gives the criterion for odd harmonic mappings [16 Theorem 3.2].

(iii) Choosing p=1p=1 and g≡0g\equiv0 in Theorem 2.1 gives [11 Theorem 1].

(iv) Choosing p=2p=2 and g≡0g\equiv0 in Theorems 2.1 gives [11 Theorem 2].

The next result replaces monotonicity by a restriction on the total variation of weighted coefficient differences.

Theorem 2.3 Let {an}n≥1\{a_n\}_{n\geq1} and {bn}n≥1\{b_n\}_{n\geq1} be two sequences of real numbers such that a1=1a_1=1, and b1=0b_1=0. For n≥1n\geq1, suppose that cn=(n+1)an+1−nanc_n=(n+1)a_{n+1}-na_n, dn=(n+1)bn+1−nbnd_n=(n+1)b_{n+1}-nb_n. If

∣c1−1∣+∣d1∣+∑n=1∞(∣cn+1−cn∣+∣dn+1−dn∣)≤1,(4)\lvert c_1-1\rvert+\lvert d_1\rvert+\sum_{n=1}^{\infty}\left(\lvert c_{n+1}-c_n\rvert+\lvert d_{n+1}-d_n\rvert\right)\leq1, \tag*{(4)}

then the harmonic mapping f(z)=h(z)+g(z)‾=z+∑n=2∞anzn+∑n=2∞bnzn‾f(z)=h(z)+\overline{g(z)}=z+\sum_{n=2}^{\infty}a_nz^n+\sum_{n=2}^{\infty}\overline{b_nz^n}, z∈Dz\in\mathbb{D}, is sense-preserving, close-to-convex, and univalent in D\mathbb{D}.

Proof. By hypothesis both sequences {cn}\{c_n\} and {dn}\{d_n\} are bounded. Since nan=1+∑j=1n−1cjna_n=1+\sum_{j=1}^{n-1}c_j, nbn=∑j=1n−1djnb_n=\sum_{j=1}^{n-1}d_j, the sequences {an}\{a_n\} and {bn}\{b_n\} are bounded. Thus hh and gg are analytic in D\mathbb{D}. For ∣ε∣=1\lvert\varepsilon\rvert=1, consider Fε(z)=h(z)+εg(z)F_{\varepsilon}(z)=h(z)+\varepsilon g(z). Now for z∈Dz\in\mathbb{D}, we obtain

Re⁡{(1−z)2Fε′(z)}≥1−∣c1−1∣∣z∣−∣d1∣∣z∣−∑n=1∞(∣cn+1−cn∣+∣dn+1−dn∣)∣z∣n+1≥1−∣z∣>0\begin{aligned} \operatorname{Re}\{(1-z)^2F_{\varepsilon}'(z)\} &\geq1-\lvert c_1-1\rvert\lvert z\rvert-\lvert d_1\rvert\lvert z\rvert-\sum_{n=1}^{\infty}\left(\lvert c_{n+1}-c_n\rvert+\lvert d_{n+1}-d_n\rvert\right)\lvert z\rvert^{n+1}\\ &\geq1-\lvert z\rvert>0 \end{aligned}

The function ψ(z)=z/(1−z)\psi(z)=z/(1-z) is convex in D\mathbb{D}, and Re⁡{Fε′(z)/ψ′(z)}=Re⁡{(1−z)2Fε′(z)}>0\operatorname{Re}\{F_{\varepsilon}'(z)/\psi'(z)\}=\operatorname{Re}\{(1-z)^2F_{\varepsilon}'(z)\}>0. Hence every FεF_{\varepsilon} is close-to-convex in D\mathbb{D}. Now using Lemma 1.1, f=h+g‾f=h+\overline{g} is close-to-convex and univalent in D\mathbb{D}.

Moreover, the last inequality implies h′(z)+εg′(z)≠0h'(z)+\varepsilon g'(z)\ne0 for every ε\varepsilon such that ∣ε∣=1\lvert\varepsilon\rvert=1. Therefore, ∣h′(z)∣≠∣g′(z)∣\lvert h'(z)\rvert\ne\lvert g'(z)\rvert in D\mathbb{D}. Since h′(0)=1h'(0)=1 and g′(0)=0g'(0)=0, the continuity gives ∣g′(z)∣<∣h′(z)∣\lvert g'(z)\rvert<\lvert h'(z)\rvert, z∈Dz\in\mathbb{D}. Hence, ff is sense-preserving. This completes the proof.

Remark 2.4. Theorem 2.3 remains applicable when the sequences {cn}\{c_n\} and {dn}\{d_n\} are non-monotone. This distinguishes the bounded-variation method from coefficient criteria requiring monotonicity.

The next Corollary relates Theorem 2.3 to three results of Sahoo and Wankhede [16].

Corollary 2.5. Let {an}n≥1\{a_n\}_{n\geq1} and {bn}n≥1\{b_n\}_{n\geq1} be two sequences of real numbers such that a1=1a_1=1, and b1=0b_1=0. Suppose that cn=(n+1)an+1−nanc_n=(n+1)a_{n+1}-na_n, dn=(n+1)bn+1−nbnd_n=(n+1)b_{n+1}-nb_n, and L=lim⁡n→∞(cn+dn)L=\lim_{n\to\infty}(c_n+d_n). Then each of the following assumptions implies (4):

  1. The sequences {cn}\{c_n\} and {dn}\{d_n\} are non-decreasing, c1≥1c_1\geq1, d1≥0d_1\geq0, and L≤2L\leq2.

  2. The sequences {cn}\{c_n\} and {dn}\{d_n\} are non-decreasing, c1≤1c_1\leq1, d1≥0d_1\geq0, and L≤2c1L\leq2c_1.

  3. The sequences {cn}\{c_n\} and {dn}\{d_n\} are non-increasing, 0≤c1≤10\leq c_1\leq1, d1≥0d_1\geq0, and L≥2d1L\geq2d_1.

Proof. Let SS denote the left-hand side of (4), and let SN=∣c1−1∣+∣d1∣+∑n=1N(∣cn+1−cn∣+∣dn+1−dn∣)S_N=|c_1-1|+|d_1|+\sum_{n=1}^{N}(|c_{n+1}-c_n|+|d_{n+1}-d_n|)

Under (i), we get

S=lim⁡N→∞SN=lim⁡N→∞[cN+dN−1]=L−1≤1.S=\lim_{N\to\infty}S_N=\lim_{N\to\infty}[c_N+d_N-1]=L-1\leq1.

Under (ii), we get

S=lim⁡N→∞SN=lim⁡N→∞[1−2c1+cN+dN]=1−2c1+L≤1.S=\lim_{N\to\infty}S_N=\lim_{N\to\infty}[1-2c_1+c_N+d_N]=1-2c_1+L\leq1.

Finally, under (iii), we get

S=lim⁡N→∞SN=lim⁡N→∞[1+2d1−cN−dN]=1+2d1−L≤1.S=\lim_{N\to\infty}S_N=\lim_{N\to\infty}[1+2d_1-c_N-d_N]=1+2d_1-L\leq1.

Hence the result follows from Theorem 2.3.

Remark 2.6. Theorem 2.3 contains Theorems 3.4, 3.5, and 3.6 of Sahoo and Wankhede [16].

Applications

Theorem 3.1 Let p∈{1,2}p\in\{1,2\}, a,b>0a,b>0, and c=a+bc=a+b. Suppose that pab≥1pab\geq1 and a+b≤(p+1)aba+b\leq(p+1)ab. Let λ>0\lambda>0 satisfy p(λ+p(a+b))Γ(a+b)≤2λΓ(a)Γ(b)p(\lambda+p(a+b))\Gamma(a+b)\leq2\lambda\Gamma(a)\Gamma(b). Then the harmonic mapping

Hp,λa,b(z)=z 2F1(a,b;a+b;zp)+z2λ2F1′(a,b;a+b+1;zp)(5)\mathcal{H}_{p,\lambda}^{a,b}(z)=z\,{}_2F_1(a,b;a+b;z^p)+\frac{z^2}{\lambda}{}_2F_1'(a,b;a+b+1;z^p) \tag*{(5)}

is close-to-convex and univalent in D\mathbb{D}.

Proof. We shall use Theorem 2.1. For c=a+bc=a+b, and for p∈{1,2}p\in\{1,2\}, we write Hp,λa,b(z)=hp(z)+gp(z)\mathcal{H}_{p,\lambda}^{a,b}(z)=h_p(z)+g_p(z). Then

hp(z)=z+∑n=1∞Apn+1zpn+1andgp(z)=∑n=1∞Bpn+1zpn+1.h_p(z)=z+\sum_{n=1}^{\infty}A_{pn+1}z^{pn+1} \qquad\text{and}\qquad g_p(z)=\sum_{n=1}^{\infty}B_{pn+1}z^{pn+1}.

where

Apn+1=(a)n(b)n(c)nn!,Bpn+1=pnλ(a)n(b)n(c+1)nn!.(6)A_{pn+1}=\frac{(a)_n(b)_n}{(c)_n n!}, \qquad B_{pn+1}=\frac{pn}{\lambda}\frac{(a)_n(b)_n}{(c+1)_n n!}. \tag*{(6)}

Also, A1=1A_1=1 and B1=0B_1=0.

Note that Apn+1≥0A_{pn+1}\geq0 for all n≥0n\geq0. Define Cn=(pn+1)Apn+1C_n=(pn+1)A_{pn+1}, n≥0n\geq0. Then using c=a+bc=a+b, we get for n≥0n\geq0,

Cn+1−Cn=(p(n+1)+1)(a+n)(b+n)−(pn+1)(c+n)(n+1)(c+n)(n+1)Apn+1=(pab−1)n+(p+1)ab−a−b(c+n)(n+1)Apn+1≥0.(7)\begin{aligned} C_{n+1}-C_n &=\frac{(p(n+1)+1)(a+n)(b+n)-(pn+1)(c+n)(n+1)}{(c+n)(n+1)}A_{pn+1}\\ &=\frac{(pab-1)n+(p+1)ab-a-b}{(c+n)(n+1)}A_{pn+1}\geq0. \tag*{(7)} \end{aligned}

Hence {(pn+1)Apn+1}n≥0\{(pn+1)A_{pn+1}\}_{n\geq0} is non-decreasing.

Moreover, define Dn=(pn+1)Bpn+1D_n=(pn+1)B_{pn+1}, n≥0n\geq0, where D0=0D_0=0. From (6), we have

Bpn+1=pcnλ(c+n)Apn+1,(n≥1)andD1=(p+1)Bp+1=p(p+1)abλ(c+1)>0.B_{pn+1}=\frac{pcn}{\lambda(c+n)}A_{pn+1},\quad(n\geq1) \qquad\text{and}\qquad D_1=(p+1)B_{p+1}=\frac{p(p+1)ab}{\lambda(c+1)}>0.

Now using c=a+bc=a+b, we obtain

Dn+1−Dn=(p(n+1)+1)(a+n)(b+n)−(pn+1)n(c+n+1)n(c+n+1)Bpn+1=(pab+pc−1)n+(p+1)abn(c+n+1)Bpn+1>0\begin{aligned} D_{n+1}-D_n =\frac{(p(n+1)+1)(a+n)(b+n)-(pn+1)n(c+n+1)}{n(c+n+1)}B_{pn+1} \\ &=\frac{(pab+pc-1)n+(p+1)ab}{n(c+n+1)}B_{pn+1}>0 \end{aligned}

for every n≥1n\geq1. Hence, the sequence {(pn+1)Bpn+1}n≥0\{(pn+1)B_{pn+1}\}_{n\geq0} is non-decreasing.

To complete the proof, it remains to show that lim⁡n→∞(pn+1)(Apn+1+Bpn+1)≤2\lim_{n\to\infty}(pn+1)(A_{pn+1}+B_{pn+1})\leq2. By using the standard asymptotic formula for the Gamma function

lim⁡n→∞Γ(n+α)Γ(n+β)nβ−α=1\lim_{n\to\infty}\frac{\Gamma(n+\alpha)}{\Gamma(n+\beta)}n^{\beta-\alpha}=1

and using c=a+bc=a+b, we obtain

lim⁡n→∞(pn+1)Apn+1=lim⁡n→∞Γ(c)Γ(a)Γ(b)pn+1n(Γ(n+a)Γ(n+1)n1−a)(Γ(n+b)Γ(n+c)nc−b)na+b−c=pΓ(c)Γ(a)Γ(b),\begin{aligned} \lim_{n\to\infty}(pn+1)A_{pn+1} =\lim_{n\to\infty}\frac{\Gamma(c)}{\Gamma(a)\Gamma(b)}\frac{pn+1}{n} \left(\frac{\Gamma(n+a)}{\Gamma(n+1)}n^{1-a}\right) \left(\frac{\Gamma(n+b)}{\Gamma(n+c)}n^{c-b}\right)n^{a+b-c} \\ &=\frac{p\Gamma(c)}{\Gamma(a)\Gamma(b)}, \end{aligned}

and

lim⁡n→∞(pn+1)Bpn+1=pcλlim⁡n→∞nc+n(pn+1)Apn+1=p2cΓ(c)λΓ(a)Γ(b).\lim_{n\to\infty}(pn+1)B_{pn+1} =\frac{pc}{\lambda}\lim_{n\to\infty}\frac{n}{c+n}(pn+1)A_{pn+1} =\frac{p^2c\Gamma(c)}{\lambda\Gamma(a)\Gamma(b)}.

Thus

lim⁡n→∞(pn+1)(Apn+1+Bpn+1)=pΓ(c)Γ(a)Γ(b)(1+pcλ)≤2\lim_{n\to\infty}(pn+1)(A_{pn+1}+B_{pn+1}) =\frac{p\Gamma(c)}{\Gamma(a)\Gamma(b)} \left(1+\frac{pc}{\lambda}\right)\leq2

for c=a+bc=a+b. This concludes the proof.

Remark 3.2.

  • (i) Taking p=1p=1 in Theorem 3.1, we obtain [16 Theorem 4.1].

  • (ii) Taking p=2p=2 in Theorem 3.1, we obtain [16 Theorem 4.3].

Theorem 3.3 Let a,b,c>0a,b,c>0 satisfy c>a+b+1c>a+b+1. Suppose that α,β>0\alpha,\beta>0 and

α+β≤Γ(c−a)Γ(c−b)4Γ(c)Γ(c−a−b−1)[ab+2(c−a−b−1)].\alpha+\beta\leq\frac{\Gamma(c-a)\Gamma(c-b)}{4\Gamma(c)\Gamma(c-a-b-1)}[ab+2(c-a-b-1)].

Then the harmonic mapping

f(z)=z1−z+αz22F1(a,b;c;z)+βz22F1(a,b;c;z)f(z)=\frac{z}{1-z}+\alpha z^2{}_2F_1(a,b;c;z)+\beta z^2{}_2F_1(a,b;c;z)

is sense-preserving, close-to-convex, and univalent in D\mathbb{D}.

Proof. Let f=h+g‾f=h+\overline{g}, then h(z)=z+∑n=2∞anznh(z)=z+\sum_{n=2}^{\infty}a_nz^n and g(z)=∑n=2∞bnzng(z)=\sum_{n=2}^{\infty}b_nz^n, where for n≥2n\geq2 and

Cn=(a)n(b)n(c)n(n)!,C_n=\frac{(a)_n(b)_n}{(c)_n(n)!},

we have an=1+αCn−2a_n=1+\alpha C_{n-2} and bn=βCn−2b_n=\beta C_{n-2}. Define u1=0u_1=0 and un=u_n= ncn−2n c_{n-2}, n≥2n \ge2. Then nan=n+αunn a_n = n + \alpha u_n and nbn=βunn b_n = \beta u_n (n≥2n \ge2). Thus the associated sequences are

cn=(n+1)an+1−nan=1+α(un+1−un)c_n = (n+1)a_{n+1} - n a_n = 1 + \alpha(u_{n+1} - u_n)

and

dn=(n+1)bn+1−nbn=β(un+1−un).d_n = (n+1)b_{n+1} - n b_n = \beta(u_{n+1} - u_n).

Since u2=2u_2 = 2, we have c1=1+2αc_1 = 1 + 2\alpha, d1=2βd_1 = 2\beta. Hence ∣c1−1∣+∣d1∣=2(α+β)|c_1 - 1| + |d_1| = 2(\alpha+ \beta). Moreover,

cn+1−cn=α(un+2−2un+1+un)anddn+1−dn=β(un+2−2un+1+un).c_{n+1} - c_n = \alpha(u_{n+2} - 2u_{n+1} + u_n) \quad\text{and}\quad d_{n+1} - d_n = \beta(u_{n+2} - 2u_{n+1} + u_n).

Let SS denote the left-hand side of (4), then

S=2(α+β)+(α+β)∑n=1∞∣un+2−2un+1+un∣.S = 2(\alpha+ \beta) + (\alpha+ \beta)\sum_{n=1}^{\infty} |u_{n+2} - 2u_{n+1} + u_n|.

Since un≥0u_n \ge0, ∣un+2−2un+1+un∣≤un+2+2un+1+un|u_{n+2} - 2u_{n+1} + u_n| \le u_{n+2} + 2u_{n+1} + u_n. Now summing over n≥1n \ge1, and using u1=0u_1 = 0, u2=2u_2 = 2, we obtain

∑n=1∞∣un+2−2un+1+un∣≤∑n=1∞un+2+2∑n=1∞un+1+∑n=1∞un=4∑n=2∞un−2.\begin{aligned} \sum_{n=1}^{\infty} |u_{n+2} - 2u_{n+1} + u_n| \le \sum_{n=1}^{\infty} u_{n+2} + 2\sum_{n=1}^{\infty} u_{n+1} + \sum_{n=1}^{\infty} u_n \\ &= 4\sum_{n=2}^{\infty} u_n - 2. \end{aligned}

Therefore S≤4(α+β)∑n=2∞unS \le4(\alpha+ \beta)\sum_{n=2}^{\infty}u_n. Now,

∑n=2∞un=∑k=0∞(k+2)(a)k(b)k(c)kk!=2F(1)+F′(1).\sum_{n=2}^{\infty}u_n = \sum_{k=0}^{\infty}(k+2)\frac{(a)_k(b)_k}{(c)_k k!} = 2F(1)+F'(1).

Since c>a+b+1c>a+b+1, using Gauss’s summation formula, eq (2), and the hypothesis, we obtain

S≤4(α+β)Γ(c)Γ(c−a−b−1)Γ(c−a)Γ(c−b)[2(c−a−b−1)+ab]≤1,S \le4(\alpha+\beta) \frac{\Gamma(c)\Gamma(c-a-b-1)} {\Gamma(c-a)\Gamma(c-b)} \left[2(c-a-b-1)+ab\right]\le1,

Therefore, Theorem 2.3 implies that ff is sense-preserving, close-to-convex, and univalent in D\mathbb{D}.

If {cn}\{c_n\} and {dn}\{d_n\} are monotone, their total variations can be expressed through their values and limits. However, if monotonicity is absent, the same criterion remains meaningful, provided the total variations are finite and sufficiently small. As a simple illustration, we give the following result.

Theorem 3.4 Let mm be a nonnegative integer and let

An(m)={(−m)n(m+1)n(n!)2,0≤n≤m,0,n>m.A_n^{(m)} = \begin{cases} \dfrac{(-m)_n(m+1)_n}{(n!)^2}, & 0\le n\le m,\\ 0, & n>m. \end{cases}

Suppose that u1=0u_1=0, un=nAn−2(m)u_n=nA_{n-2}^{(m)} (n≥2n\ge2), and vm=2+∑n=1m+2∣un+2−2un+1+un∣v_m=2+\sum_{n=1}^{m+2}|u_{n+2}-2u_{n+1}+u_n|. If α,β>0\alpha,\beta>0 satisfy α+β≤1/vm\alpha+\beta\le1/v_m, then the harmonic mapping

fm(z)=z1−z+αz2Pm(1−2z)+βz2Pm(1−2z)f_m(z)=\frac{z}{1-z}+\alpha z^2P_m(1-2z)+\beta z^2P_m(1-2z)

is sense-preserving, close-to-convex, and univalent in D\mathbb{D}.

Proof. Using the terminating hypergeometric series,

Pm(1−2z)=∑k=0m(−m)k(m+1)k(k!)2zkP_m(1-2z)=\sum_{k=0}^{m}\frac{(-m)_k(m+1)_k}{(k!)^2}z^k

we write fm=hm+gm‾f_m=h_m+\overline{g_m}, where

hm(z)=z1−z+αz2Pm(1−2z),gm(z)=βz2Pm(1−2z).h_m(z)=\frac{z}{1-z}+\alpha z^2P_m(1-2z),\qquad g_m(z)=\beta z^2P_m(1-2z).

For n≥2n\ge2, the Taylor coefficients are an=1+αAn−2(m)a_n=1+\alpha A_{n-2}^{(m)}, and bn=βAn−2(m)b_n=\beta A_{n-2}^{(m)}. Hence, the sequences appearing in Theorem 2.3 are

cn=(n+1)an+1−nan=1+α(un+1−un),c_n=(n+1)a_{n+1}-na_n=1+\alpha(u_{n+1}-u_n),

and

dn=(n+1)bn+1−nbn=β(un+1−un).d_n=(n+1)b_{n+1}-nb_n=\beta(u_{n+1}-u_n).

Since u1=0u_1=0 and u2=2u_2=2, we have c1=1+2αc_1=1+2\alpha, d1=2βd_1=2\beta. Furthermore,

cn+1−cn=α(un+2−2un+1+un)c_{n+1}-c_n=\alpha(u_{n+2}-2u_{n+1}+u_n)

and

dn+1−dn=β(un+2−2un+1+un).d_{n+1}-d_n=\beta(u_{n+2}-2u_{n+1}+u_n).

Therefore, the left-hand side of (2) becomes

S=∣c1−1∣+∣d1∣+∑n=1∞(∣cn+1−cn∣+∣dn+1−dn∣)=(α+β)(2+∑n=1m+2∣un+2−2un+1+un∣)=(α+β)vm.\begin{aligned} S=|c_1-1|+|d_1|+\sum_{n=1}^{\infty}\left(|c_{n+1}-c_n|+|d_{n+1}-d_n|\right) \\ &=(\alpha+\beta)\left(2+\sum_{n=1}^{m+2}|u_{n+2}-2u_{n+1}+u_n|\right)=(\alpha+\beta)v_m. \end{aligned}

By hypothesis, S≤1S\le1. This proves the assertion.

Remark 3.5.

  1. Taking m=0m=0 in Theorem 3.4, we obtain u2=2u_2=2, un=0u_n=0 (n≥3n\ge3), hence v0=8v_0=8. Therefore, α+β≤1/8\alpha+\beta\le1/8 is sufficient for the harmonic mapping

f0(z)=z1−z+αz2+βz2f_0(z)=\frac{z}{1-z}+\alpha z^2+\beta z^2

to be sense-preserving, close-to-convex, and univalent.

  1. Taking m=1m=1 in Theorem 3.4, we obtain: If α,β>0\alpha,\beta>0 satisfy α+β≤1/32\alpha+\beta\le1/32, then the harmonic mapping

f(z)=z1−z+αz2(1−2z)+βz2(1−2z)f(z)=\frac{z}{1-z}+\alpha z^2(1-2z)+\beta z^2(1-2z)

is sense-preserving, close-to-convex, and univalent in D\mathbb{D}.

References

References

  1. [1]Alexander, J.W.: Functions which map the interior of the unit circle upon simple regions. Ann. of Math. 17(1915) 12–22.DOI
  2. [2]Andrews, G.E., Askey, R., Roy, R.: Special Functions. Encyclopedia of Mathematics and Its Applications, vol. 71. Cambridge University Press, Cambridge (1999).
  3. [3]Carlson, B.C., Shaffer, D.B.: Starlike and prestarlike hypergeometric functions. SIAM J. Math. Anal. 15(1984) 737–745.DOI
  4. [4]Clunie, J., Sheil-Small, T.: Harmonic univalent functions. Ann. Acad. Sci. Fenn. Ser. A I Math. 9(1984) 3–25.DOI
  5. [5]Duren, P.L.: Harmonic Mappings in the Plane. Cambridge Tracts in Mathematics, vol. 156. Cambridge University Press, Cambridge (2004).DOI
  6. [6]Fejér, L.: Trigonometrische Reihen und Potenzreihen mit mehrfach monotoner Koeffizientenfolge. Trans. Amer. Math. Soc. 39(1936) 18–59.DOI
  7. [7]Goodman, A.W.: Univalent Functions, vols. I and II. Mariner Publishing Company, Tampa (1983).
  8. [8]Hästö, P., Ponnusamy, S., Vuorinen, M.: Starlikeness of the Gaussian hypergeometric functions. Complex Var. Elliptic Equ. 55(2010), 173–184.DOI
  9. [9]Kumari, N., Prajapat, J.K., Dash, P.P.: Convexity of the hypergeometric functions _0F_2 and its applications. Proc. Natl. Acad. Sci., India, Sect. A Phys. Sci. (2026). https://doi.org/10.1007/s40010-026-01219-7.
  10. [10]Küstner, R.: On the order of starlikeness of the shifted Gauss hypergeometric function. J. Math. Anal. Appl. 334(2007), 1363–1385.DOI
  11. [11]MacGregor, T.H.: Univalent power series whose coefficients have monotonic properties. Math. Z. 112(1969) 222–228.DOI
  12. [12]Miller, S.S., Mocanu, P.T.: Univalence of Gaussian and confluent hypergeometric functions. Proc. Amer. Math. Soc. 110(1990) 333–342.DOI
  13. [13]Ponnusamy, S., Rønning, F.: Geometric properties for convolutions of hypergeometric functions and functions with the derivative in a half plane. Integral Transforms Spec. Funct. 8(1999) 121–138.DOI
  14. [14]Ponnusamy, S., Vuorinen, M.: Univalence and convexity properties for Gaussian hypergeometric functions. Rocky Mountain J. Math. 31(2001) 327–353.DOI
  15. [15]Robertson, M.S.: Univalent power series with multiply monotonic sequences of coefficients. Ann. of Math. 46(1945) 533–555.DOI
  16. [16]Sahoo, S.K., Wankhede, S.: Monotone coefficients of univalent harmonic functions with applications to special functions. J. Math. Anal. Appl. 554(2026) 129898.DOI
  17. [17]Sugawa, T., Wang, L.-M.: Geometric properties of the shifted hypergeometric functions. Complex Anal. Oper. Theory 11(2017) 1879–1893.arxiv.org/abs/1704.07948
  18. [18]Sugawa, T., Wang, L.-M.: On geometric properties of the ratio of two hypergeometric functions. Comput. Methods Funct. Theory 24(2024) 241–256.DOI
  19. [19]Sugawa, T., Wang, L.-M., Wu, C.: Universal convexity and range problems of shifted hypergeometric functions. Proc. Amer. Math. Soc. 152(2024) 3521–3535.arxiv.org/abs/2309.02934
  20. [20]Szegő, G.: Power series with multiply monotonic sequences of coefficients. Duke Math. J. 8(1941) 559–564.DOI
  21. [21]Yıldız, İ., Akyar, A.: An analytical investigation on starlikeness and convexity properties for hypergeometric functions. Turkish J. Math. 44(2020), 1453–1465.DOI

Paper details

vlnc:W_286aj7uv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents