Regularity of some `incomplete' Pál-type interpolation problems (Q1612415): Difference between revisions

In this paper the polynomials \[ w_n^{(\alpha)}(z)= (z+\alpha)^n + (1+\alpha z)^n,\quad v_n^{(\alpha)}(z) = (z+\alpha)^n - (1+\alpha z)^n,\tag{*} \] with \(0<\alpha <1\) and \(n\geq 2\), whose zeros are denoted by \(\{w_j\}\) and \(\{v_j\}\), respectively, are considered. From the text: ``It has been proved in [\textit{M. G. de Bruin, A. Sharma} and \textit{J. Szabados, N. K. Govil} (ed.) et al., Approximation theory. In memory of A. K. Varma. New York: Marcel Dekker. Pure Appl. Math., Marcel Dekker. 212, 167-179 (1998; Zbl 0907.41001)] that the Pál-type interpolation problem: (**) given \(2n\) numbers \(b_j\), \(c_j\) \((1<j\leq n)\), find a polynomial \(P(z)\) of degree \(\leq 2n-1\), satisfying \(P(w_j)= b_j\), \(P'(v_j)=c_j\) \((1\leq j\leq n)\) has a unique solution for \(0<\alpha <1\). This is equivalent to: the problem stated above with \(b_j = c_j = 0\) \((1\leq j\leq n)\) has the trivial solution \(P(z)\equiv 0\) only. This uniqueness property is referred to as \textit{regularity}''. The author proves regularity for the Pál-type interpolation problem (**) in nine distinct cases of corresponding pairs of functions of the type (*). In each case the zeros of the first polynomial of the pair are used to interpolate the values \(P(.)\) and the zeros of the second one for the derivative \(P'(.)\).