Rational Arithmetic in Floating-Point

摘要

Professor Kahan pursued research to provide improved subroutines for common arithmetic functions for scientific computations. He produced an algorithm for the accurate implementation of rational arithmetic operations without resorting to multi-precision arithmetic. This was described in a paper entitled Rational arithmetic in floating point. He has also made a careful study of how to make branch cuts in the complex plane so as to allow evaluation of the dementary functions without any anomalies. This was presented in a talk at the conference on State-of -the-Art in Numerical Analysis held in Birmingham, England, April 14-18, 1986. Keywords: Computations.