Stable numerical differentiation for the second order derivatives (Q609545)

A simple scheme for stable differentiation from given noisy data was proposed by \textit{S.~Ahn}, \textit{U.~J.~Choi}, and \textit{A.~G.~Ramm} [J. Comput. Appl. Math. 186, No.~2, 325--334 (2006; Zbl 1086.65013)]. In the paper under review, this method is extended to compute the second derivative \(f''\) of a real-valued function \(f \in H^2(0,\,1)\) from given noisy data. Using Tikhonov regularization, this ill-posed differentiation problem is reformulated as a linear Volterra integral equation of the second kind. The regularizing solution is explicitly given. The convergence analysis and error estimates of the regularizing solution for both the \textit{a priori} and the \textit{a posteriori} choice strategies of the regularization parameter are presented. Several numerical examples are given.