Browse Prior Art Database

Institutionalization of FORTRAN: Early FORTRAN User Experience

IP.com Disclosure Number: IPCOM000129433D
Original Publication Date: 1984-Jan-01
Included in the Prior Art Database: 2005-Oct-06
Document File: 3 page(s) / 20K

Publishing Venue

Software Patent Institute

Related People

HERBERT S. BRIGHT: AUTHOR [+2]

Abstract

I want to mention a little-known aspect of how Westinghouse-Bettis, a nuclear-power-reactor development laboratory, together with a lot of other groups who ultimately became known as ";the nuclear-codes crowd,"; got interested in large-scale computing in the middle 1950s.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 33% of the total text.

Page 1 of 3

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

Copyright ©; 1984 by the American Federation of Information Processing Societies, Inc. Used with permission.

Institutionalization of FORTRAN: Early FORTRAN User Experience

HERBERT S. BRIGHT

(Image Omitted: Author s Address: H. S. Bright, 7840 Aberdeen Road, Bethesda, MD 20814.)

I want to mention a little-known aspect of how Westinghouse-Bettis, a nuclear-power-reactor development laboratory, together with a lot of other groups who ultimately became known as "the nuclear-codes crowd," got interested in large-scale computing in the middle 1950s.

Systems of elliptic partial differential equations are used to describe fixed-geometry nuclear- power reactors for criticality calculations. One group, reputed to be the world's "outstanding authorities," investigated the use of digital computers to perform such calculations by relaxation or successive-approximation techniques. They concluded on theoretical grounds that the rate of approach to a correct solution, which must decrease with problem size, went to zero for numerical models of order 600. Using what was then the world's most powerful computer, the NORC (Naval Ordnance Research Calculator), they performed experiments that seemed to support that conclusion.

Problems of the order 2500 were already in use for two-dimensional reactor design work, represented by passive electric-network models. Using a special-purpose analog simulator, one such solution took about six weeks of day-and-night chain-gang-style labor for several skilled technicians. The 600-limit "proof" had pretty well convinced the reactor designers that digital computers weren't going to help them.

A team of mathematicians, headed by Elizabeth Cuthill at the U.S. Navy's David Taylor Model Basin near Washington, concluded that the proof applied to the mathematical technique instead of to the problem. Using a new technique, they wrote a program to solve problems up to order 2500.

The machine they had available was a UNIVAC computer that had about as much memory as a modern pocket-size key-driven calculator and executed roughly 1000 instructions per second. The program took between 30 and 40 hours of machine time per solution, but it ran! Results were correct and usable. It was used to design several reactors.

Although Betty would never have named anything after herself, her 2500-point program became known as the Cuthill Code -- now a household word in the nuclear-codes community. Without such a successful demonstration that the world's outstanding authorities could be wrong, there would have been no early large-scale nuclear codes. The demand for more and more powerful computers would not have gained a major push.

When I recently discussed this development with a distinguished computer historian, I was startled to learn that few people in other fields of applied mathematics have even heard of the work. As of today, he will no longer be able to make that stateme...