First-order Schwartian derivative and some classes of univalent functions Milutin Obradović, Nikola Tuneski Abstract . In this paper we give estimate of | S f ′ ​ ( 0 ) | |S^{\prime}_{f}(0)| , where S f ​ ( z ) = ( f ′′ ​ ( z ) f ′ ​ ( z ) ) ′ − 1 2 ​ ( f ′′ ​ ( z ) f ′ ​ ( z ) ) 2 S_{f}(z)=\left(\frac{f^{\prime\prime}(z)}{f^{\prime}(z)}\right)^{\prime}-\frac{1}{2}\left(\frac{f^{\prime\prime}(z)}{f^{\prime}(z)}\right)^{2} is the Schwartian derivative, and f f belongs to different classes of functions univalent in the open unit disc 𝔻 {\mathbb{D}} . 2000 Mathematics Subject Classification 30C45, 30C50, 30C55 1. Introduction Let {\mathcal{A}} denote the family of all functions analytic in the open unit disk {\mathbb{D}}:=\{z\in{\mathbb{C}}:\,|z|<1\} , satisfying the normalization f(0)=0=f^{\prime}(0)-1 , i.e., (1) f(z)=z+a_{2}z^{2}+a_{3}z^{3}+\cdots. Let \mathcal{S} be the subclass of \mathcal{A} consisting of functions that are univalent in {\mathbb{D}} . Next we will define several classes of univalent functions that will be studied later in the paper. For 0\leq\alpha<1 , let \mathcal{S}^{\star}(\alpha) and \mathcal{C}(\alpha) denote the subclasses of {\mathcal{A}} which are starlike of order \alpha and convex of order \alpha in {\mathbb{D}} , respectively, i.e., with analytical characterisation: \mathcal{S}^{\star}(\alpha)=\left\{f\in\mathcal{A}:\operatorname{Re}\left[\frac{zf^{\prime}(z)}{f(z)}\right]>\alpha,\,z\in{\mathbb{D}}\right\} and \mathcal{C}(\alpha)=\left\{f\in\mathcal{A}:\operatorname{Re}\left[1+\frac{zf^{\prime\prime}(z)}{f^{\prime}(z)}\right]>\alpha,\,z\in{\mathbb{D}}\right\}. Further, let \mathcal{S}\mathcal{S}^{\star}_{\beta} and \mathcal{S}\mathcal{C}_{\beta} , 0<\beta\leq 1 , be the subclasses of \mathcal{A} consisting of functions f satisfying \left|\arg\left(\frac{zf^{\prime}(z)}{f(z)}\right)\right|<\frac{\pi\beta}{2}\quad(z\in{\mathbb{D}}) and \left|\arg\left(1+\frac{zf^{\prime\prime}(z)}{f^{\prime}(z)}\right)\right|<\frac{\pi\beta}{2}\quad(z\in{\mathbb{D}}), respectively. The functions from \mathcal{S}\mathcal{S}^{\star}_{\beta} are called strongly starlike of order \beta , and from \mathcal{S}\mathcal{C}_{\beta} strongly convex of order \beta . Similarly, \mathcal{G}(\alpha) , 0<\alpha\leq 1 , is the class of functions f\in\mathcal{A} for which {\rm Re}\left[1+\frac{zf^{\prime\prime}(z)}{f^{\prime}(z)}\right]<1+\frac{1}{2}\alpha\quad(z\in{\mathbb{D}}). Ozaki in [ 9 ] introduced the class \mathcal{G}(1) and proved that functions in \mathcal{G}(1) are univalent in the unit disk. Later, Umezawa in [ 15 ] , Sakaguchi in [ 11 ] and R. Singh and S. Singh in [ 12 ] showed, respectively, that functions in \mathcal{G}(1) are convex in one direction, close-to-convex and starlike. This, general class is extensively studied in [ 4 ] and [ 6 ] . Further, {\mathcal{U}}(\lambda)=\left\{f\in{\mathcal{A}}:\left|\left(\frac{z}{f(z)}\right)^{2}f^{\prime}(z)-1\right|<\lambda,z\in{\mathbb{D}}\right\}, where 0<\lambda\leq 1 , is subclass of the class of univalent functions, and its special case when \lambda=1 is first studied in [ 3 ] . More details on them can be found in [ 7 , 8 , 13 ] ). Finally, for 0\leq\alpha<1 , \mathcal{R}(\alpha)=\left\{f\in\mathcal{A}:{\rm Re}f^{\prime}(z)>\alpha,\,z\in{\mathbb{D}}\right\} is subclass of the class of function of bounded turning \mathcal{R}(0) . For a locally univalent function f in {\mathbb{D}} , the Schwartian derivative S_{f} is defined as S_{f}(z)=\left(\frac{f^{\prime\prime}(z)}{f^{\prime}(z)}\right)^{\prime}-\frac{1}{2}\left(\frac{f^{\prime\prime}(z)}{f^{\prime}(z)}\right)^{2}. In [ 14 ] it is shown that for the function f given by ( 1 ), (2) S^{\prime}_{f}(0)=24(a_{4}-2a_{2}a_{3}+a_{2}^{3}). Using some properties of Caratheodory class \mathcal{P} consisting of functions p(z)=1+p_{1}z+p_{2}z^{2}+\cdots such that {\rm Re}\{p(z)\}>0 z\in{\mathbb{D}} , in [ 14 ] the authors gave the estimate of |S^{\prime}_{f}(0)| for the classes \mathcal{S}^{\star}(\alpha) , \mathcal{C}(\alpha) , \mathcal{S}\mathcal{S}^{\star}_{\beta} , and \mathcal{S}\mathcal{C}_{\beta} . In this paper, using different approach, we will give the estimates of |S^{\prime}_{f}(0)| for the classes of univalent functions defined above, as well as, for the general class \mathcal{S} . 2. The estimate of |S^{\prime}_{f}(0)| for some classes of univalent functions In the study of |S^{\prime}_{f}(0)| we will need the following Lemma from [ 10 ] , given using the notation from that paper. Lemma 1 . If \omega(z)=c_{1}z+c_{2}z^{2}+c_{3}z^{3}+\cdots is analytic in {\mathbb{D}} satisfying the condition |\omega(z)|<1 , z\in{\mathbb{D}} , and if \Psi(\omega)=|c_{3}+\mu c_{1}c_{2}+\nu c_{1}^{3}|, then the following sharp estimate \Psi(\omega)\leq\Phi(\mu,\nu) holds, where \Phi(\mu,\nu)=\begin{cases}1,&(\mu,\nu)\in D_{1}\cup D_{2}\cup\{(2,1)\},\\ |\nu|,&(\mu,\nu)\in D_{6},\\ \end{cases} where D_{1}=\left\{(\mu,\nu):|\mu|\leq\frac{1}{2},\,-1\leq\nu\leq 1\right\}, D_{2}=\left\{(\mu,\nu):\frac{1}{2}\leq|\mu|\leq 2,\,\frac{4}{27}(|\mu|+1)^{3}-(|\mu|+1)\}\leq\nu\leq 1\right\}, and D_{6}=\left\{(\mu,\nu):2\leq|\mu|\leq 4,\,\nu\geq\frac{1}{12}(\mu^{2}+8)\right\}. Theorem 1 . Let f\in\mathcal{R}(\alpha) , 0\leq\alpha<1 , then |S^{\prime}_{f}(0)|\leq 12(1-\alpha). This result is sharp Proof. From the definition of the class \mathcal{R}(\alpha) , we have f^{\prime}(z)=\alpha+(1-\alpha)\frac{1+\omega(z)}{1-\omega(z)}\quad\left(=2\alpha-1+2(1-\alpha)\frac{1}{1-\omega(z)}\right), where \omega is analytic in {\mathbb{D}} with \omega(0)=0 and |\omega(z)|<1 , z\in{\mathbb{D}} . From here f^{\prime}(z)=1+2(1-\alpha)\left[\omega(z)+\omega^{2}(z)+\cdots\right]. If we put \omega(z)=c_{1}z+c_{2}z^{2}+\cdots , and compare the coefficients on z , z^{2} , and z^{3} , after some calculations, we obtain (3) \begin{split}a_{2}&=(1-\alpha)c_{1},\\ a_{3}&=\frac{2}{3}(1-\alpha)\left(c_{2}+c_{1}^{2}\right),\\ a_{4}&=\frac{1}{2}(1-\alpha)\left(c_{3}+2c_{1}c_{2}+c_{1}^{3}\right).\end{split} Now, using ( 2 ) and ( 3 ) we get \begin{split}|S^{\prime}_{f}(0)|&=24|a_{4}-2a_{2}a_{3}+a_{2}^{3}|\\ &=12(1-\alpha)\left|c_{3}+\frac{2}{3}(4\alpha-1)c_{1}c_{2}+\frac{6\alpha^{2}-4\alpha+1}{3}c_{1}^{3}\right|\\ &=12(1-\alpha)\cdot\Psi(\omega),\end{split} where we used notation from Lemma 1 , with \mu=\frac{2}{3}(4\alpha-1) and \nu=\frac{6\alpha^{2}-4\alpha+1}{3} . Using the lemma, to prove the estimate, it is enough to show that for 0\leq\alpha<1 , (\mu,\nu) is in D_{1} or in D_{2} , which will mean that \Psi(\omega)\leq\Phi(\mu,\nu)=1 . Indeed, if 1/16\leq\alpha<7/16 , or equivalently |\mu|\leq 1/2 , then 0\leq\nu\leq 1 , i.e., (\mu,\nu) is in D_{1} . Otherwise, if 0\leq\alpha\leq 1/16 or 7/16\leq\alpha\leq 1 , we have 1/2\leq|\mu|\leq 2 and easily \nu\leq 1 . In the first case, if 0\leq\alpha\leq 1/16 , then \mu<0 , while in the second case, if 7/16\leq\alpha\leq 1 , then \mu>0 . In both cases it can be easily verified that \frac{4}{27}(|\mu|+1)^{3}-(|\mu|+1)\}\leq\nu, leading to the conclusion (\mu,\nu)\in D_{2} . The result is sharp for the function for which f^{\prime}(z)=\alpha+(1-\alpha)\frac{1+z^{3}}{1-z^{3}}. ∎ Theorem 2 . Let f\in\mathcal{C}(\alpha) -\frac{1}{2}\leq\alpha<1 . Then |S^{\prime}_{f}(0)|\leq 4(1-\alpha) , and the bound is sharp. Proof. From the definition of the class \mathcal{C}(\alpha) , similarly as in the proof of Theorem 1 , we have (4) \left(zf^{\prime}(z)\right)^{\prime}=\left[1+2(1-\alpha)(\omega(z)+\omega^{2}(z)+\cdots)\right]\cdot f^{\prime}(z), where \omega is analytic in {\mathbb{D}} with \omega(0)=0 and |\omega(z)|<1 , z\in{\mathbb{D}} . If we put \omega(z)=c_{1}z+c_{2}z^{2}+\cdots , and compare the coefficients of z , z^{2} , and z^{3} , from the relation ( 4 ) after some simple calculations we obtain, (5) \begin{split}a_{2}&=(1-\alpha)c_{1},\\ a_{3}&=\frac{1}{3}(1-\alpha)\left(c_{2}+(3-2\alpha)c_{1}^{2}\right),\\ a_{4}&=\frac{1}{6}(1-\alpha)\left(c_{3}+(5-3\alpha)c_{1}c_{2}+(2\alpha^{2}-7\alpha+6)c_{1}^{3}\right).\end{split} Next, using ( 2 ) and ( 5 ) we can have \begin{split}|S^{\prime}_{f}(0)|&=24|a_{4}-2a_{2}a_{3}+a_{2}^{3}|\\ &=4(1-\alpha)\left|c_{3}+(\alpha+1)c_{1}c_{2}+\alpha c_{1}^{3}\right|\\ &\leq 4(1-\alpha),\end{split} where we used Lemma 1 , case D_{2} . Namely, for -\frac{1}{2}\leq\alpha<1 we have \frac{1}{2}\leq\mu=\alpha+1\leq 2 , and we have \frac{4}{27}(|\mu|+1)^{3}-(|\mu|+1)\leq\nu\leq 1 being equivalent to \frac{4}{27}(\alpha+2)^{3}-(\alpha+2)\leq\alpha\leq 1 , i.e., to the inequality (\alpha-1)(2\alpha^{2}+14\alpha+11)\leq 0 which can be easily verified to be true. Given result is sharp for the function satisfying 1+\frac{zf^{\prime\prime}(z)}{f^{\prime}(z)}=\alpha+(1-\alpha)\frac{1+z^{3}}{1-z^{3}} , or equivalently, f(z)=\int_{0}^{z}(1-t^{3})^{-\frac{2}{3}(1-\alpha)}dt. We note that in [ 14 ] the authors gave the same result, but only for the case 0\leq\alpha<1 ∎ Theorem 3 . Let f\in\mathcal{G}(\alpha),\,0<\alpha\leq 1 , then |S^{\prime}_{f}(0)|\leq\alpha^{2}+2\alpha , and this result is sharp. Proof. From the definition of the class \mathcal{G}(\alpha) we have 1+\frac{zf^{\prime\prime}(z)}{f^{\prime}(z)}=1+\frac{1}{2}\alpha-\frac{\alpha}{2}\frac{1+\omega(z)}{1-\omega(z)}\quad\left(=1+\alpha-\alpha\frac{1}{1-\omega(z)}\right), where \omega is analytic in {\mathbb{D}} with \omega(0)=0 and |\omega(z)|<1 , z\in{\mathbb{D}} . The last relation can be rewritten as (6) [zf^{\prime}(z)]^{\prime}=\left\{1-\alpha[\omega(z)+\omega^{2}(z)+\cdots]\right\}\cdot f^{\prime}(z). Similarly as in the proofs of previous theorems, putting \omega(z)=c_{1}z+c_{2}z^{2}+\cdots in ( 6 ) and comparing the coefficients on z , z^{2} , and z^{3} , leads to (7) \begin{split}\displaystyle\vskip 3.0pt plus 1.0pt minus 1.0pta_{2}&=-\frac{\alpha}{2}c_{1},\\ a_{3}&=-\frac{\alpha}{6}\left[c_{2}+(1-\alpha)c_{1}^{2}\right],\\ a_{4}&=-\frac{\alpha}{24}\left[2c_{3}+(4-3\alpha)c_{1}c_{2}+(\alpha^{2}-3\alpha+2)c_{1}^{3}\right].\end{split} From ( 2 ) and ( 7 ), after some calculations, we get \begin{split}|S^{\prime}_{f}(0)|&=24|a_{4}-2a_{2}a_{3}+a_{2}^{3}|\\ &=2\alpha\left|c_{3}+\left(2+\frac{1}{2}\alpha\right)c_{1}c_{2}+\left(1+\frac{1}{2}\alpha\right)c_{1}^{3}\right|\\ &\leq 2\alpha\left(1+\frac{1}{2}\alpha\right)\\ &=\alpha^{2}+2\alpha.\end{split} In last step we applied Lemma 1 , case D_{6} , since 2<\mu=2+\frac{1}{2}\alpha\leq\frac{5}{2} and 1<\nu=1+\frac{1}{2}\alpha\leq\frac{3}{2} . The result of this theorem is sharp as the function defined by 1+\frac{zf^{\prime\prime}(z)}{f^{\prime}(z)}=1+\frac{1}{2}\alpha-\frac{\alpha}{2}\cdot\frac{1+z}{1-z}, shows, i.e., for f(z)=\frac{1-(1-z)^{\alpha+1}}{\alpha+1}. ∎ For the proof of the next theorem we will need the following result proven in [ 5 ] as a part of the proof of its Theorem 1. Lemma 2 . For each function f in \mathcal{U}(\lambda), 0<\lambda\leq 1, there exists function \omega, analytic in \mathbb{D}, such that |\omega(z)|\leq|z|<1, and |\omega^{\prime}(z)|\leq 1, for all z\in\mathbb{D}, with (8) \frac{z}{f(z)}=1-a_{2}z-\lambda z\omega(z). Additionally, for \omega(z)=c_{1}z+c_{2}z^{2}+\cdots, (9) |c_{1}|\leq 1\quad\mbox{and}\quad|c_{2}|\leq\frac{1}{2}(1-|c_{1}|^{2}). Theorem 4 . If f belongs to \mathcal{U}(\lambda), 0<\lambda\leq 1, then |S^{\prime}_{f}(0)|\leq 12\lambda and the result is sharp. Proof. Using ( 8 ) we have z=[1-a_{2}z-\lambda z\omega(z)]\cdot f(z), and after equating the coefficients, \begin{split}a_{3}=&\lambda c_{1}+a^{2}_{2},\\ a_{4}=&\lambda c_{2}+2\lambda a_{2}c_{1}+a_{2}^{3}.\end{split} Finally, using previous relations and ( 2 ), we have \begin{split}|S^{\prime}_{f}(0)|&=24|a_{4}-2a_{2}a_{3}+a_{2}^{3}|=24|\lambda c_{2}|\\ &\leq 24\lambda\cdot\frac{1}{2}(1-|c_{1}|^{2})\leq 12\lambda,\end{split} where we used ( 9 ). The result is sharp with extremal function f(z)=\frac{z}{1-\frac{\lambda}{2}z^{3}}. ∎ 3. Estimate of |S^{\prime}_{f}(0)| for the general class \mathcal{S} of univalent functions For our investigation in this section we will use the Grunsky coefficients and its properties. Here are basic definitions and results on those coefficients based on the book of N.A. Lebedev ( [ 2 ] ). Let f\in\mathcal{S} and let \log\frac{f(t)-f(z)}{t-z}=\sum_{p,q=0}^{\infty}\omega_{p,q}t^{p}z^{q}, where \omega_{p,q} are called Grunsky’s coefficients with property \omega_{p,q}=\omega_{q,p} . For those coefficients we have the next Grunsky’s inequalitiy ( [ 1 , 2 ] ): (10) \sum_{q=1}^{\infty}q\left|\sum_{p=1}^{\infty}\omega_{p,q}x_{p}\right|^{2}\leq\sum_{p=1}^{\infty}\frac{|x_{p}|^{2}}{p}, where x_{p} are arbitrary complex numbers such that 0<\sum_{p=1}^{\infty}\frac{|x_{p}|^{2}}{p}<+\infty . If \overline{\lim}_{p\to\infty}\sqrt[p]{|x_{p}|}<1 , then in ( 10 ) we have equality if, and only if, the area of \widehat{\mathbb{C}}\setminus f^{-1}({\mathbb{D}}) is zero, where f^{-1}(z)=\frac{1}{f(z)} . Further, it is well-known that if f given by ( 1 ) belongs to \mathcal{S} , then also (11) f_{2}(z)=\sqrt{f(z^{2})}=z+c_{3}z^{3}+c_{5}z^{5}+\cdots belongs to the class \mathcal{S} . Then, for the function f_{2} we have the appropriate Grunsky’s coefficients of the form \omega_{2p-1,2q-1} , and the inequality ( 10 ) has the form: (12) \sum_{q=1}^{\infty}(2q-1)\left|\sum_{p=1}^{\infty}\omega_{2p-1,2q-1}x_{2p-1}\right|^{2}\leq\sum_{p=1}^{\infty}\frac{|x_{2p-1}|^{2}}{2p-1}. Here, and further in the paper we omit the upper index (2) in \omega_{2p-1,2q-1}^{(2)} if compared with Lebedev’s notation. From the inequality ( 12 ), when x_{2p-1}=0 and p=3,4,\ldots , we have |\omega_{11}x_{1}+\omega_{31}x_{3}|^{2}+3|\omega_{13}x_{1}+\omega_{33}x_{3}|^{2}+5|\omega_{15}x_{1}+\omega_{35}x_{3}|^{2}\leq|x_{1}|^{2}+\frac{|x_{3}|^{2}}{3}. From here, for x_{1}=1 and x_{3}=0 , we obtain (13) |\omega_{11}|^{2}+3|\omega_{13}|^{2}+5|\omega_{15}|^{2}\leq 1. As it has been shown in [ 2 , p. 57] , if f is given by ( 1 ), then the coefficients a_{2} , a_{3} , and a_{4} are expressed by Grunsky’s coefficients \omega_{2p-1,2q-1} of the function f_{2} given by ( 11 ) in the following way: (14) \begin{split}a_{2}&=2\omega_{11},\\ a_{3}&=2\omega_{13}+3\omega_{11}^{2},\\ a_{4}&=2\omega_{33}+8\omega_{11}\omega_{13}+\frac{10}{3}\omega_{11}^{3},\\ 0&=3\omega_{15}-3\omega_{11}\omega_{13}+\omega_{11}^{3}-3\omega_{33}.\end{split} Now, we can formulate and prove the main result of this section. Theorem 5 . If f belongs to \mathcal{S} , then |S^{\prime}_{f}(0)|\leq\frac{64}{5}\sqrt{3}=22.17025\ldots. Proof. Using ( 2 ) and ( 14 ), we have \begin{split}|S^{\prime}_{f}(0)|&=24|a_{4}-2a_{2}a_{3}+a_{2}^{3}|\\ &=48\left|\omega_{33}-\frac{1}{3}\omega_{11}^{3}\right|,\end{split} or, if we use the fourth relation from ( 14 ) and the relation ( 13 ), \begin{split}|S^{\prime}_{f}(0)|&=48|\omega_{15}-\omega_{11}\omega_{13}|\\ &\leq 48(|\omega_{15}|+|\omega_{11}||\omega_{13}|)\\ &\leq 48\left(\frac{1}{\sqrt{5}}\sqrt{1-|\omega_{11}|^{2}-3|\omega_{13}|^{2}}+|\omega_{11}||\omega_{13}|\right)\\ &=:48\cdot g(|\omega_{11}|,|\omega_{13}|),\end{split} where g(x,y)=\frac{1}{\sqrt{5}}\sqrt{1-x^{2}-3y^{2}}+xy and (x,y)\in\Omega=\left\{(x,y):0\leq x\leq 1,\,0\leq y\leq\frac{1}{\sqrt{3}}\sqrt{1-x^{2}}\right\}. Now, let find maximal value of the function g on \Omega . To derive the critical points of g , we need to solve in the interior of \Omega the following system \left\{\begin{split}\frac{\partial g(x,y)}{\partial x}&=y-\frac{1}{\sqrt{5}}\frac{x}{\sqrt{1-x^{2}-3y^{2}}}=0,\\ \frac{\partial g(x,y)}{\partial y}&=x-\frac{1}{\sqrt{5}}\frac{3y}{\sqrt{1-x^{2}-3y^{2}}}=0.\end{split}\right. Now, from 3y\frac{\partial f_{2}(x,y)}{\partial x}-x\frac{\partial f_{2}(x,y)}{\partial y}=3y^{2}-x^{2}=0, we receive y^{2}=\frac{x^{2}}{3}. Substituting it in the second equation of the system gives \frac{x}{\sqrt{3}}-\frac{1}{\sqrt{5}}\frac{x}{\sqrt{1-2x^{2}}}=0, which brings the unique solution of the system in the interior of \Omega , x_{1}=\frac{1}{\sqrt{5}} and y_{1}=\frac{1}{\sqrt{15}} such that g(x_{1},y_{1})=\frac{64}{5}\sqrt{3}=22.17025\ldots . On the edges of \Omega we have: - g(0,y)=\frac{1}{\sqrt{5}}\sqrt{1-3y^{2}}\leq\frac{1}{\sqrt{5}}=0.4472\ldots; - g(1,y)=g(1,0)=0 ; - g(x,0)=\frac{1}{\sqrt{5}}\sqrt{1-x^{2}}\leq\frac{1}{\sqrt{5}}=0.4472\ldots; - h(x):=g\left(x,\sqrt{1-x^{2}}/\sqrt{3}\right)=\frac{1}{\sqrt{3}}x\sqrt{1-x^{2}}\leq h(1/\sqrt{2})=\sqrt{3}/6=0.288675\ldots . Combining the above analysis, we receive that the function g , on the domain \Omega achieves the greatest value \frac{64}{5}\sqrt{3} obtained for x_{1}=\frac{1}{\sqrt{5}} and y_{1}=\frac{1}{\sqrt{15}} , i.e., |S^{\prime}_{f}(0)|\leq\frac{64}{5}\sqrt{3}=22.17025\ldots. ∎