OXFORD UNIVERSITY COMPUTING LABORATORY
ANNUAL REPORT 1960-61
There is no sign of a stationary state in the demands for time on Mercury. From 1 August 1960 to 31 July 1961 the machine was run for a total of 4600 hours, an average of about 182 hours per working day. Towards the end of this period continuous operation, day and night was the rule rather than the exception. The hours were consumed as shown in the table.
|(18)Other university departments||530|
|Programme testing (not itemised)||460|
|Faults (machine or human)||190|
It is clear that the Mercury machine, even if operated to full capacity (representing an increase of 30 per cent on the yearly average) cannot for long meet the computational needs of the university, even if the demand for external work ceases or is unacceptable. The next two years will be rather difficult, but some time will soon become available on machines belonging to the United Kingdom Atomic Energy Authority, and subsequently (1964) on ATLAS, a large machine to be installed at the National Institute for Nuclear Research.
These facilities will be reinforced, particularly for research purposes, by a replacement for Mercury. After much discussion in July 1961 between the university and the U.G.C., a grant of £150,000 was made in August, 1961. This sum will be reinforced by income from the sale of Mercury time, and a KDF 9 machine, made by English Electric Company, was ordered in November 1961, for delivery in the middle of 1964. This machine is at least twenty times as powerful as Mercury.
In January 1961 Miss Joan Walsh obtained the degree of D.Phil. for her work on "Numerical solution of partial differential equations using a high speed computer". Two more mathematicians obtained D.S.I.R. grants, and seven research students are now working for Oxford degrees. Another is being supervised in the laboratory for a London research degree.
Some questions on Numerical Analysis were introduced for the first time in Mathematical Moderations, and for this purpose the Director gave 24 lectures on computational problems in linear algebra. He also gave 8 lectures for special paper in the Final examination for Engineering Science, and an advanced seminar course of 8 lectures on Chebyshev polynomials.
Other special lectures were contributed by:-
Mr J S Rollett:machine-language programming for Mercury (8); advanced Mercury Autocode programming (8);
Mr D F Mayers:numerical solution of integral equations (6); Mercury Autocode courses (three days each in January and April 1961);
Mr D C Handscomb:theory and application of Monte Carlo methods (6);
Mr C Y Lee, of the Bell Telephone Laboratories, New Jersey, spent one month in the laboratory and gave 4 lectures on "Pattern processing and elementary algebraic machine manipulation".
The summer school of 1960, in collaboration with the Delegacy for Extra-Mural Studies, was attended by about 65 representatives from firms, government departments and other universities, and about 15 members of this university. One week was devoted to "linear algebra" and one to "differential equations", and the lectures were given mainly by Mr. Mayers, Mr Rollett and the Director.
A second school was planned for 1961 on a more ambitious and advanced scale, with the solution of differential equations (ordinary and partial) and integral equations as the only topic. The resident lecturers (the Director and Mr Mayers) were assisted by sixteen representatives of A.W.R.E. (Aldermaston), U.K.A.E.A. (Harwell and Winfrith Heath), the Meteorological Office (Bracknell), the War Office (Fort Halstead), the C.E.G.B. (London), London University and Oxford University Engineering Laboratory. The lectures will be collected in one book, "The numerical solution of ordinary and partial differential equations", to be published by Pergamon Press in the spring of 1962.
The Director gave colloquia or lectures at Harwell, London University, College of Advanced Technology at Salford, Cambridge University and the National Physical Laboratory, and was nominated as the first "Visiting Lecturer" at Hatfield Technical College. He was invited to attend and contribute to a symposium on "Matrix methods" at Gatlinburg, Tennessee, organised by the American Society for Industrial and Applied Mathematics, to join the editorial board of the British Computer Journal, and to serve on a Mathematics and Physics Committee of the War Office.
Mr J S Rollett gave a colloquium at Sheffield University, three lectures for the Oxford University Institute of Education, and two in Professor Coulson's Summer School on Theoretical Chemistry. He read papers at a meeting in Glasgow of the X-ray analysis group of the Institute of Physics, and in Cambridge at the Fifth Congress of the International Union of Crystallography; and attended a meeting at Frankfurt of the International Union Commission on Crystallographic Computing.
Mr D F Mayers attended an ALGOL 60 Programming School at the Automatic Programming Information Centre, Brighton, lectured in the Theoretical Chemistry Summer School, and gave a colloquium at the National Physical Laboratory.
Mr D C Handscomb attended the British Association meeting in Norwich.
L Fox. Some comments on the accuracy and convenience of finite-difference processes in automatic computation. From "Premier Congres de l'association Francaise de calcul", Gauthier-Villars, Paris, 1961.
- Numerical solution of ordinary and partial differential equations. From "Frontier research on digital computers", Vol. II, University of North Carolina, Chapel Hill, 1960.
- The British Computer Society, Harrogate Meeting. Nature 187, 655-6, 1960.
D C Handscomb, J R Ehrman and L D Fosdick. Computation of order parameters in an Ising lattice by the Monte Carlo method. J. Math. Phys. 1, 547-58, 1960.
J R Rollett. General programmes for crystal structure analysis on the English Electric DEUCE computer.
- and O S Mills. Programming and crystallographic application of the Ferranti Mercury computer.
(Papers 7 and 9 respectively of "Computing methods and the phase problem in X-ray crystal analysis", Pergamon, Oxford, 1961).
- Noel E Taylor and Dorothy Crowfoot Hodgkin. The X-ray crystallographic determination of the structure of Bromomiroestrol. J. Chem. Soc. 3685-95, 1960.
- and J H Wilkinson. An efficient scheme for the codiagonalisation of a symmetric matrix by Givens' method in a computer with a two-level store. Computer Journal 4, 177-180, 1961.
Professor G E Blackman (Chairman), Dr. N T J Bailey, Professor C A Coulson, Mr E F Jackson, Dr J W Linnett, Dr D ter Haar, Dr L Fox (Secretary)