A continuous extension of the de la Vallée Poussin means (Q1411295): Difference between revisions

Among the many interesting results of their 1958 paper [Pac. J. Math. 8, 295--334 (1958; Zbl 0084.27901)], \textit{G. Pólya} and \textit{I. J. Schoenberg} studied the de la Vallée Poussin means of analytic functions. These are polynomial approximations of a given analytic function on the unit disk \(\mathbb{D}\) obtained by taking Hadamard products ``\(*\)'' of the function \(f\) with certain polynomials \(V_n(z)\), where \(n\) is the degree of the polynomial. The polynomial approximation \(V_n* f\) converge locally uniformly to \(f\) as \(n\to \infty\). In this paper, the authors define a subordination chain \(V_\lambda(z)\), \(\lambda> 0\), \(|\lambda|< 1\), of convex mappings of the disk that for integer values in the same as the previously defined \(V_n(z)\). Precisely Pólya and Schoenberg conjectured that \(f\in\mathbb{C}\) implies \[ V_1* f(\mathbb{D}) \subset\cdots\subset V_n* f(\mathbb{D}) \subset V_{n+1}* f(\mathbb{D}) \subset\cdots\subset f(\mathbb{D}), \quad n\geq 2.\tag{1} \] Note that \[ V_n(z):= \frac{1} {\binom{2n}{n}} \sum_{k=1}^n \binom{2n}{n+k} z^k, \quad n\in \mathbb{N}, \quad f_0(z):= \frac{z}{1-z}\in\mathbb{C} \] and that \(V_n* f_0= V_n\). Therefore, \[ V_1(\mathbb{D}) \subset\cdots\subset V_n(\mathbb{D}) \subset V_{n+1}(\mathbb{D}) \subset\cdots\subset f_0(\mathbb{D}), \quad n\geq 2,\tag{2} \] is a special case of (1). This later relation was also suggested by Pólya and Schoenberg (loc. cit.), but no proof was given. It is the aim of the present paper to provide a proof for (2), in a more general setting. The key to this work is the fact that \(V_n\) satisfies a first-order differential equation: \[ zV_\lambda'(z)+ \lambda \frac{1-z}{1+z} V_\lambda(z)= \frac{\lambda z}{1+z}, \quad V_\lambda(0)= 0, \tag{3} \] where \(\lambda= n\). For arbitrary real \(\lambda> 0\) the solutions of (3) is given explicitly in terms of the Gaussian hypergeometric function (see Theorem 1.1) as: \[ V_\lambda(z)= \frac{\lambda z}{\lambda+1} {}_2 F_1 (1,1-\lambda, 2+\lambda,-z). \] The main result is given in: Theorem 1.2. The solutions \(V_\lambda(z)\) of (3) are in \(\mathbb{C}\) for \(\lambda>0\), and we have \[ V_{\lambda_1}(\mathbb{D}) \subset V_{\lambda_2}(\mathbb{D}) \subset f_0(\mathbb{D}), \quad 0< \lambda_1< \lambda_2. \] In other words, the functions \(V_\lambda\) form a convex subordination chain for \(\lambda>0\). The authors show that this special case, combined with the general theory of convolutions of convex univalent functions, implies that if \(f\) is a conformal mapping of the disk \(\mathbb{D}\) onto a convex domain, then \(V_\lambda* f\to f\) locally uniformly as \(\lambda\to \infty\), and in fact \(V_{\lambda_1}* f(\mathbb{D}) \subset V_{\lambda_2}* f(\mathbb{D}) \subset f(\mathbb{D})\) when \(\lambda_2> \lambda_1\). The authors obtain also other very interesting results considering Hadamard products of the \(V_\lambda\) with complex-valued harmonic mappings of the disk.