Mathematical software and ACM Publications

Abstract

Mathematical software started as a scientific activity almost as soon as serious scientific computing. The field was brought into focus at the symposium Mathematical Software held at Purdue University on April 1-3, 1970. The symposium's organizing committee was John Rice (chairman), Robert Ashenhurst, Charles Lawson, Stuart Lynn and Joseph Traub. It was sponsored by ACM and SIGNUM and financially supported by the Office of Naval Research. Mathematical software was defined then as the set of algorithms in the area of mathematics and it was noted that this definition is much broader than traditional numerical analysis. Even today there are large areas of mathematical software which have yet to be studied systematically or seriously (e.g., geometric algorithms). The first chapter of the symposium proceedings, Mathematical Software [Rice, 1971] presents a brief history of the field up to that point. It is noted there that the first mathematical software published was an EDVAC machine language program to convert base 10 integers to binary; it was in Mathematical Tables and Aids to Computations (now called Mathematics of Computation) on pages 427-431 of Volume 3, 1949. Further noted is that the book [Wilkes, Wheeler and Gill, 1951] contains a thorough discussion of the mathematical software (subroutine library) for the EDSAC. The second chapter of Mathematical Software is The Distribution and Sources of Mathematical Software which summarizes the state of the field as of 1970. The recent book, Sources and Development of Mathematical Software [Cowell, 1984] contains as first chapter the essay Observations on the Mathematical Software Effort by W. J. Cody. Many of the other 13 chapters of Cowell's book contain historical remarks about specific mathematical software areas. Chapter 3 of Mathematical Software is The Challenge for Mathematical Software which raises many points still completely unresolved. It concludes with recommendations for the establishment of:A Journal of Mathematical Software A Center or Focal Point for Mathematical Software The implementation of the first recommendation is the focal point of this article, the other recommendation has yet to be carried out. Perhaps mathematical software is now too big for a “Center” to cover the whole field, but a focal point would still serve a very important scientific function.