The trapezoidal rule for analytic functions of rapid decrease (Q1824979): Difference between revisions

The aim of the paper is to give estimates for the error of the trapezoidal rule for smooth functions of rapid decrease (e.g. \(| f(z)| \leq M \exp (\alpha | z|^{\beta}),\) \(\alpha >0\), \(\beta\geq 2\), \(z\in {\mathbb{C}})\). Inequalities of the form \(| \int^{\infty}_{-\infty}f(x)dx-\sum^{N}_{k=-N}f(kh)| \leq C(N)\exp (-\gamma_ N)\) and of similar formulas for transformed rules are obtained in case that f is analytic in a strip about the real axis of width d or f is analytic in a transformed set. The choice of the mesh size h is optimized for each of these rules. C(N) is growing at most algebraically with N. The last section considers the implementation of the rules obtained above in the case that f is not entire.