Quo Vadimus: computer science in a decade
J. F. Traub
A panel discussion was held during the third biennial meeting of chairmen of Ph.D.-granting computer science departments in June, 1978 at Snowbird, Utah, a meeting sponsored by the Computer Science Board. Invitees from industry and government were also present. A report was prepared from tapes made of the discussion (Department of Computer Science, Carnegie-Mellon University: Report #CMU-CS-80-127, June 1980). It contained all the prepared statements of the panelists, lightly edited, and the panelists' discussion in its entirety. A selection of the audience discussion was also included, rather heavily edited. The following presentation is derived from that report.
Author Archives
Numerical mathematics and computer science
Numerical mathematics is viewed as the analysis of continuous algorithms. Four of the components of numerical mathematics are discussed. These are: foundations (finite precision number systems, computational compexity), synthesis and analysis of algorithms, analysis of error, programs and program libraries.
Algorithm 419: zeros of a complex polynomial [C2]
The subroutine CPOLY is a Fortran program to find all the zeros of a complex polynomial by the three-stage complex algorithm described in Jenkins and Traub [4]. (An algorithm for real polynomials is given in [5].) The algorithm is similar in spirit to the two-stage algorithms studied by Traub [1, 2]. The program finds the zeros one at a time in roughly increasing order of modulus and deflates the polynomial to one of lower degree. The program is extremely fast and the timing is quite insensitive to the distribution of zeros. Extensive testing of an Algol version of the program, reported in Jenkins [3], has shown the program to be very reliable.
ACM proposes to republish contents of Communications Algorithms section is useable looseleaf format, with bimonthly updating service, provided there is sufficient demand. For details, see News item on page 583.
The Vocabulary Subcommittee of the International Standards Organization's Technical Committee on Computers and Information Processing (ISO/TC97/SC1) held its third meeting in New York City in May, 1964. (More precisely, this was the subcommittee's first meeting. Its earlier meetings in Geneva and Paris were as a Working Group.) The program of work agreed upon at the New York meeting marks a sharp reversal of SC1's earlier plans.
USA participation in an international standard glossary on information processing
A considerable number of glossaries in the area of information processing have been produced in the USA in the last ten years [1, 2]. In some cases the glossaries were reworked versions of earlier glossaries, while in other cases major new contributions were made. All told, the glossary effort has cost thousands of man-hours of work.
Several years ago the ASA X3 sectional committee sponsored by BEMA was established to prepare standards for the USA in the information processing field. (See Appendix 1 for the meaning of abbreviations and acronyms. See also [3].) ASA X3.5 was assigned the double scope of advising the other X3.n subcommittees on the establishment of definitions required for their proposed standards and of establishing a standard glossary, pASGIP, for general use.
At the same time there was important British standardization activity. After reworking a number of earlier drafts, the BSI released the “Glossary of Terms Used in Automatic Data Processing,” British Standard 3527: 1962. The British effort differed in at least one very important respect from the USA glossaries. It was organized along subject rather than alphabetical lines. This was to have important consequences, as we shall see.
On a class of iteration formulas and some historical notes
The class of iteration formulas obtainable by rational approximations of “Euler's formula” is derived with the corresponding error estimates. Some historical notes on iterative procedures are followed by a derivation of Euler's formula with the associated error estimate in a new notation which simplifies the error estimate and suggests generalizations. The final section considers the Padé approximants to the “Euler polynomial” and shows how a number of known formulas may be derived from this unified approach. There is a short discussion of the “best” formula.
Comparison of iterative methods for the calculation of nth roots
Three iterative methods for calculation of nth roots (including one proposed by the author) are compared in two ways: (1) Theoretical convergence estimates are given. (2) A new macro-compiler which estimates machine running time is used to compare the running time of the three methods for a variety of input data.
Shape the Future of Computing
ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.
Get Involved