OXFORD UNIVERSITY COMPUTING LABORATORY

ANNUAL REPORT 1958 - 9

  1. Premises and equipment.
  2. The only major event to report under this heading is the final installation of the Mercury high-speed computer, which passed its tests in March 1959 and was formally handed over to the university on 16 Mach. The maintenance of the machine is being performed, under contract for one year with Ferranti Ltd., by an engineer employed by them and who now occupies a room on the ground floor adjacent to the computer.

    All the ancillary equipment, comprising tape-editing machines, punched-card machines, and the various desk machines, is maintained by 'on-call' arrangements with the respective makers.

  3. Use of Mercury.
  4. The one-shift maintenance contract with Ferranti permits the use of the machine for forty hours each week, of which approximately two hours per day are involved in routine maintenance. The remaining hours, reduced in number by inevitable machine 'faults', were insufficient to meet demands from paying customers, university departments, and the research of the laboratory. Ferranti's therefore agreed, subject to a small charge, to allow use of the machine in 'unmaintained' time of about twenty hours each week.

    In the first six months of operation the total 'useful' time so obtained was about 1,160 hours, of which 245 were used by thirteen extra-mural organisations, including the universities of Birmingham and Glasgow, 445 by twelve university departments in Oxford, and about 430 by the Computing Laboratory, the remaining hours coming under the category of 'miscellaneous', including demonstrations to visiting groups.

    The level of efficiency of the machine, measured as the ratio of 'total useful maintained time' to 'total maintained time' (excluding routine maintenance), has been reasonably high, averaging 87.2 per cent. for the first six months, and 91.2 per cent. if two very bad weeks, one in April and one in June, are excluded.

  5. Research students and visitors.

    Of the three research students mentioned in the previous report one has completed a thesis, for the Degree of B.Sc., on 'Commercial applications of high-speed computing machines'. The other two are studying various aspects of the numerical solution of partial differential equations.

    As a result of the final degree examinations in June two mathematicians were awarded D.S.I.R. grants, and will study respectively 'the numerical solution of eigenvalue problems' and 'numerical methods in the analysis of crystal structures', for the Degree of D.Phil.. A third new B.Sc. student is supported by one of the firms who hire machine time, and will apply Monte Carlo methods for solving partial differential equations.

    Dr. Th. Busk, from the University of Aarhus, Denmark, spent three summer months in the laboratory, Dr. R. A. Sparks, National Science Foundation Fellow, is extending his research until the end of 1959, and Dr. T. Haavie, from the Atomic Energy Establishment in Norway, is studying here for one year under the auspices of the British Council. Mr. D. E. Conway also worked for three months on programmes for the calculation of interfacial angles, for inclusion in the Barker Index of crystals being prepared by Dr. Porter of the Department of Geology and Mineralogy.

  6. Teaching

    University lectures on Numerical Analysis (40) have been given by L. Fox, and on Programming (16) by J. S. Rollett. Seminars on Crystallographic Computation (J. S. Rollett and R. A. Sparks) and on Atomic Structure Calculations (D. F. Mayers) were held in the laboratory in Hilary Term.

    Concentrated courses on the Autocode method of programming were given by the laboratory staff in April and September, and were attended by some 120 members of the university and about twenty extra-mural representatives.

    The success of these courses is indicated by the fact that some forty university members now write programmes, others use existing programmes for their problems, and knowledge is steadily increasing, in all scientific departments, of the kind of work for which this machine is valuable.

  7. The Mercury library.

    A large part of the work of the laboratory has been the writing of general-purpose programmes. Routines and sub-routines are now available, and have been sent to other .Mercury installations, for the following problems. Each category may have several different programmes, the present number being given in brackets, according to the nature and size of the problem. The time taken to construct such programmes varies from. a few days to several months.

    1. Linear simultaneous algebraic equations, and inversion of matrices (11).

    2. Latent roots and vectors of real symmetric matrices (4).

    3. Regression analysis, curve fitting, other statistical problems (3).

    4. Operations with Chebyshev series (4).

    5. Interpolation (3).

    6. Input and output (5).

    7. Ordinary differential equations (2).

    8. Crystallography (10).

    Certain programmes have also been received from other Mercury centres, and the complete collection is available, in the Mercury 'library', for machine users.

  8. Other activities

    In April Dr. Fox accepted an invitation to participate in a conference on 'Boundary-value problems' at the University of Wisconsin, and gave a paper on Numerical experiments for eigenvalues and eigenfunctions of ordinary differential equations'. He also gave an invited series of lectures on ordinary differential equations at a summer conference on 'Frontier research in digital computing' at the University of North Carolina, Chapel Hill, and attended the annual Americal meeting of the Association for Computing Machinery at M.I.T.

    Dr. Mayers attended a conference on electronic digital computers at the Birmingham College of Technology, and the annual meeting at Cambridge of the British Computer Society. With Dr. Rollett he contributed to a seminar, in the Unit of Biometry, on 'Statistical methods on Mercury', became a consultant to A.W.R.E., Aldermaston, and pursues research. on atomic structure calclliation.

    Dr. Rollett visited Geneva, to discuss the type of ancillary equipment most suitable for the problems of the European Organization for Nuclear Research (CERN) and nuclear establishments in general. He is a member of the scientific and engineering sub-committee of the British Computer Society, and represents the laboratory at most meetings of the ancillary equipment sub-committee of the Inter-University Computing Committee.

    In general, the laboratory has taken full part in periodic group discussions on programming, equipment, and other allied topics, and maintains close relations, including participation in 'opening' ceremonies, with other coring centres.

    The laboratory was officially declared 'open' on 10 June by Sir Douglas Veale, at a Party attended by senior members of the university, and of other computing establishments, and which included short demonstrations of the use of Mercury on scientific problems.

  9. Publications.

    Fox, L., 'Minimax methods in table construction'. Reprint from Proceedings of Symposium on 'numerical Approximation', Mathematics Research Centre, United States Army, University of Wisconsin, Madison April 1958.

    'Scientific uses of Digital Computers', Bull. Inst. Phys., 185 - 9, August 1959.

Professor C. A. Coulson (Chairman), Dr. N. T. J. Bailey, Professor G. E. Blackman, Professor D. G. Champernowne, Dr. J. W. Linnett, Dr. D. ter Haar, Dr. L. Fox (Secretary).