Hausdorff moments, Hardy spaces, and power series (Q1295899)

Let \(\mathbb{C}_{-1/2}: =\{\lambda: \text{Re}(\lambda)> -1/2\}\). Let \(\{f_n\}\) be a sequence of numbers for which there exists a positive number \(M\) such that \[ (n+1)\sum_{0\leq i\leq n} (C_{n,k})^2 |\Delta^if_{(n-i)}|^2<M, \;n=0,1,\dots. \tag{*} \] Then it is known that \(f_n= \int^1_0 x^nu(x) dx\), where \(u(x)\) belongs to \(L^2(0,1)\). The authors begin by showing that the Laplace transform of \(e^{-t} u(t^{-t})\) is a function of \(\lambda\) holomorphic in \(\mathbb{C}_{-1/2}\) which interpolates \(\{f_n\}\) and has other properties which are sufficient, in view of Carlson's theorem, to make this interpolation unique. In Theorem 1 they consider the series \[ (1/2\pi) \sum_{0\leq n<\infty} a_ne^{-in \theta},\;(\theta= u+iv,u\text{ and }v\text{ real }) \tag{**} \] where the set of numbers \(\{f_n\}\), \(f_n=n^pa_n\), \((p\geq 1,\;n=0,1,\dots)\) satisfies (*). They show that (**) converges to an analytic function \(f(\theta)\) in a bottom half-plane, and that \(f\) has a holomorphic extension to a cut domain. They obtain information about the associated jump function, the discontinuity of \(if (\theta)\) across the cuts. In particular, the Laplace transform of the jump function is the carlsonian interpolation of the coefficients \(a_n\). The substitution \(z=e^{-in\theta}\) and some convenient changes of notation lead to a familiar set of ideas, dating back to Le Roy in 1900. Suppose that \(\{a_n\}\) satisfies the assumptions in Theorem 1, and let \(f(z):=(1/2\pi)\sum_{0\leq n< \infty}a_nz^n\). Then \(f\) is analytic in \(|z|<1\) and admits a holomorphic extension to the cut plane \(\mathbb{C}\setminus[1,\infty)\). The jump function \(F(x):= -i(f_+(x)- f_-(x))\) is of class \(C^{(p-1)} (p\geq 1)\), and its Mellin transform is the carlsonian interpolation of the coefficients \(a_n\). The main purpose of the paper is the reconstruction of the jump function from the coefficients. This is achieved. In Theorem 2', for instance, it is shown that \(F(x)=\sum_{0\leq m<\infty} c_m\varphi_m(x)\), where the \(\varphi_m(x)\) are obtained from Laguerre polynomials, and \(c_m\) is the sum of a series involving \(a_n\) and \(P_m(-i(n+1/2))\), \(P_m\) being Pollaczek polynomials.