Relations between univalent and orthogonal polynomials (Q882025)

The author studies connections between \textit{univalent polynomials} \[ p(z)=z+\sum_{k=2}^n\,b_kz^k,\eqno{(1)} \] sequences of polynomials (orthogonal under certain conditions) satisfying \[ Q_{m+1}(z)=(z+\beta_{m+1})Q_m(z)-\alpha_{m+1}Q_{m-1}(z),\;m\geq 1,\eqno{(2)} \] with \(Q_0=1,\, Q_1(z)=z+\beta_1\) and \(\alpha_m,\,\beta_m\) complex numbers satisfying \(\alpha_{m+1}\not= 0\not=\beta_m,\,m\geq 1\), and so-called \textit{para-orthogonal polynomials} constructed from Szegő polynomials \(\{\rho_n\}_{n=0}^{\infty}\) by \[ B_n(z,\omega_n)=\rho_n(z)+\omega_n\rho_n^{\ast}(z),\;| \omega_n| =1,\eqno{(3)} \] where \(\rho_n^{\ast}(z)=z^n\overline{\rho}_n(1/z)\) are reciprocal polynomials. \vskip0.3cm Using known results on the connection between the location of the zeros, close-to-convex functions (not introduced in the paper) and univalence (close-to-convex\(\,\Rightarrow\,\)univalent in \textbf{D}), the main results proved are \textbf{Theorem 1.} Let \(P_{m+1}\) be a polynomial and assume that \(P'_{m+1}\) satisfies (2) with \(\beta_m=1\) and \(0<\alpha_{k+1}\leq 1,\,1\leq k\leq m-1\). In addition assume \[ (\pi-\theta_m)\leq {\pi (m-1)\over m+2}, \] where \(\theta_m=2\arccos{(\kappa_n\cos{(\pi/(m+1))})}\) and \(\kappa_m=\max_{1\leq k\leq m-1}\,\sqrt{\alpha_{k+1}}\). Then \(P_{m+1}\) is not close-to-convex in \textbf{D}. \textbf{Theorem 2.} Let \(U_{k+1}\) be a polynomial, whose derivative is a para-orthogonal polynomial. Setting \(s=(1-r)/(1+r)\) with \(0<r<1\), then for \[ s^k-\left({k+2\over k+1}\right) s^{k-1}+{1\over k+1}\leq 0, \] \(U_{k+1}\) is close-to-convex. \textbf{Theorem 3.} Let \(U_{n+1}\) be polynomials, whose derivatives are para-orthogonal polynomials. Suppose that the inequalities \[ 0<| a_n| \leq {2\over n^2+n-2},\;n>1, \] are valid. Then \(U_{n+1}\) is close-to-convex. Finally the author considers two special cases of para-orthogonal polynomials (\(\omega_n=\pm 1\)) and gives an example with \(a_n=q^n,\,n\geq \) with \(0\leq q<1\) with explicit figures for the location of the zeroes on the unit circle for several values of \(n\).