One-dimensional P-algorithm with convergence rate \(O(n^{-3+\delta})\) for smooth functions (Q1594873): Difference between revisions

The authors describe an adaptive algorithm (one-dimensional P-algorithm with a decreasing sequence of threshold values \(\varepsilon_{n}\)) for approximating the global minimum of a smooth one-dimensional objective function \(f.\) The case where \(f\) is not unimodal is investigated. The main result of the paper under review is an algorithm with a decreasing sequence of threshold values \(\varepsilon_{n} = O(n^{-1 + \delta})\) which has a convergence rate \(O(n^{-3+\delta})\) for a large class of smooth objective functions. It is a continuation of the authors' article [J. Optim. Theory Appl. 104, 479-495 (1999; Zbl 0985.90075)], and the Chapter 6 the book by \textit{A. Torn} and \textit{A. Žilinskas}, [Global Optimization. Springer, Berlin (1989; Zbl 0752.90075)]. The organization of the paper under review is as follows: 1. Introduction. 2. P-algorithm. 3. Asymptotic normalized error bounds. 4. Numerical experiments.