Iterative algorithms for function computations

Iterative algorithms such as Newton´s method for square-root extraction offer attractive low-cost techniques for function computation but invariably operate at low-speed. Such algorithms are even more feasible when one considers their use in conjunction with currently available microprocessors. In this letter, two particular iterative algorithms are described and evaluated. These are for the computation of arc (cos (p))/(π) and cos (πp). Error results are given which show that the accuracy for many instrumentation-type applications is tolerable when a small number of iterations, say, 16, is involved.