Browse Prior Art Database

Anecdotes Some Memories of EDSAC I: 1950-1952

IP.com Disclosure Number: IPCOM000129324D
Original Publication Date: 1979-Oct-01
Included in the Prior Art Database: 2005-Oct-05
Document File: 3 page(s) / 21K

Publishing Venue

Software Patent Institute

Related People

Sandy Douglas: AUTHOR [+2]

Abstract

The EDSAC, like most early machines, was not altogether reliable, but the usual rule was that if a fault were inconsistent the hardware was suspect, whereas if it were consistent one's program was at fault. However this was not always true. As Professor Hartree's pupil, I designed a program to calculate atomic wave functions using the Runge-Kutta routine written by Stanley Gill. My program calculated the solution of a set of simultaneous second-order differential equations (the Hartree-Slater-Foch equations) step by step. One of the equations also calculated the normalization coefficient, given by |P2dx, the value of P being the solution of one of the other equations. At the origin P was zero, so that p2 was also (very) zero, and its value was certain to remain very small for several steps. The program calculated P2 from the current value of P and fed it to the Runge-Kutta routine as required. The program sequence for this consisted of four instructions: (1) load the multiplier register with P. (2) multiply the contents of the multiplier register by P. and place the result in the accumulator, (3) scale the result by a left shift (i.e. multiplying by a factor of two, the operation being necessary because floating point was not provided), and (4) store the scaled result ready for the R-K routine to use.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 27% of the total text.

Page 1 of 3

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

Copyright ©; 1979 by the American Federation of Information Processing Societies, Inc. Used with permission.

Anecdotes Some Memories of EDSAC I: 1950-1952

Sandy Douglas

    (Image Omitted: © 1979, American Federation of Information Processing Societies, Inc. Permission to copy without fee all or part of this material is granted provided that the copies are not made or distributed for direct commercial advantage, the AFIPS copyright notice and the title of the publication and its date appear, and notice is given that copying is by permission of the American Federation of Information Processing Societies, Inc. To copy otherwise, or to republish, requires specific permission. Author's address: The London School of Economics and Political Science, Houghton St., London WC2A 2AE, England. Key words: FDSAC, R. A. Brooker, A. S. Douglas S. Gill D, R. Hartree, J. C. P. Miller, E. C. Page, M. Phister, V. E. Price,

D. J. Wheeler. CR category: 1.2. © 1979 AFIPS 0164-1239/79/020098-03

Anecdotes Some Memories of EDSAC I: 1950-1952

Sandy Douglas .00/0)

The EDSAC, like most early machines, was not altogether reliable, but the usual rule was that if a fault were inconsistent the hardware was suspect, whereas if it were consistent one's program was at fault. However this was not always true. As Professor Hartree's pupil, I designed a program to calculate atomic wave functions using the Runge-Kutta routine written by Stanley Gill. My program calculated the solution of a set of simultaneous second-order differential equations (the Hartree-Slater-Foch equations) step by step. One of the equations also calculated the normalization coefficient, given by |P2dx, the value of P being the solution of one of the other equations. At the origin P was zero, so that p2 was also (very) zero, and its value was certain to remain very small for several steps. The program calculated P2 from the current value of P and fed it to the Runge-Kutta routine as required. The program sequence for this consisted of four instructions: (1) load the multiplier register with P. (2) multiply the contents of the multiplier register by P. and place the result in the accumulator, (3) scale the result by a left shift (i.e. multiplying by a factor of two, the operation being necessary because floating point was not provided), and (4) store the scaled result ready for the R-K routine to use.

When travelling by train from Barnstaple to Bude, I think it was -- the line is now defunct -- Douglas Hartree did some hand calculations (to two decimals) which showed that the normalization of the Ps so far produced was out by about one percent, a small but significant departure. As I had had some nonsense out occasionally, but with the result later becoming consistent, I checked the program and could find no error which would cause such a discrepancy. By chance I was watching the screen which displayed the store contents while carrying ou...