On 15 April 1987, the Right Honourable Kenneth Baker MP, the Secretary of State for Education and Science, came to the Rutherford Appleton Laboratory and inaugurated the Cray X-MP/48 computer. At the ceremony, he was welcomed to the Joint Research Councils' Supercomputer Unit by the Chairman of SERC, Professor E W Mitchell FRS, who stressed that the computer was being run by SERC on behalf of all the Research Councils. Professor Sir David Phillips FRS, the chairman of the Advisory Board for Research Councils outlined the way in which the provision of the Cray computer had come about and of the compelling arguments advanced by Professor Forty in the report of the working party commissioned by the ABRC, the UGC, and the Computer Board. Mr Baker spoke of the importance attached by Government to the provision of advanced facilities for research. He mentioned his pleasure in meeting Professor Forty for the first time. After the ceremony, a buffet lunch was provided for the guests by the courtesy of Cray Research (UK) Ltd.
In 1974 the SERC Science Board agreed to fund the first of a number of Collaborative Computational Projects. It dealt with Electron Correlations in Atoms and Molecules. Since that time, other projects have been set up in different areas of science. The current ones are supported in the main by the Science Board of SERC through Daresbury Laboratory. One, however, has support from the Astronomy and Planetary Science Board.
Large scale computation is now a well established feature of research in many branches of science. It is often complementary to traditional theory and experiment. It can be an essential component in the interpretation and understanding of experiments conducted on major facilities, such as those used by scientists supported by SERC. Examples of these include the Synchrotron Radiation Source and the Nuclear Structure Facility at Daresbury Laboratory; the Spallation Neutron Source ISIS and the Laser Facility at the Rutherford Appleton Laboratory; the various telescopes installed in Australia. Hawaii, and the Canary Islands; and particle accelerators at CERN in Geneva and at DESY in Hamburg.
The projects bring together people in universities and other research centres to formulate initiatives to collect and develop program codes which will have a significant impact upon research in their areas. In the past the codes have been adapted to run efficiently on supercomputing facilities like the Cray 1-S computer. With the introduction of the Cray X-MP/48 computer, the CCPs have turned their attention to exploiting its special features on these and other newer codes. The secretaries of the CCPs have been invited to give an indication of progress and plans for the transfer of relevant computer programs. We are grateful for their response since this gives a useful insight into aspects of the scientific importance of the Cray X-MP supercomputing facility.
The principal codes supported by CCP1 (GAMESS, CADPAC, and ATMOL) will be implemented on the RAL CRAY X-MP in three steps.
Contact Roger Amos (Chemistry Dept., Cambridge. RDA1 at RL.IB) regarding CADPAC, Martyn Guest or Bob Harrison (MFC or RJH1 at RL.IB) at Daresbury regarding GAMESS, or Vic Saunders (VRS at RL.IB) also at Daresbury regarding ATMOL.
For the longer term, the project is aiming at making good quality Hartree-Fock calculations routinely available (i.e. costing say less than 1 hour cpu time) for very large molecules, consisting of 100-400 atoms. To achieve this goal will require much development of the theory, since existing theoretical methods will not deliver the goods even on a machine as powerful as the CRAY X-MP/48. However, if machines of this power were not available there would be little chance that such a project could reach its aims.
CCP2 has already implemented a number of its major codes on the X-MP, while others are being converted to run on this machine.
Two sets of atomic R-matrix codes have been in-Stalled. These are the 'Opacity' R-matrix program and the Breit-Pauli codes.
The 'Opacity' package consists of modelled versions of the RMATRX program (CPC vol. 4, p. 367 (1978)) together with new asymptotic codes due to Prof. M. J. Seaton (University College, London) for calculating electron collision strengths, energy levels, bound-bound and bound-free data required for the Opacity project. This package will be used at Queens University (Belfast), University College (London) and Royal Holloway and Bedford New College. Trial runs have begun on the iso-electronic lithium, beryllium and carbon ions. The same codes are also being used to investigate the coupling of highly excited states and pseudo-states in electron-atom scattering.
The Breit-Pauli package is a modified version of that published in CPC Vol. 25, p. 347 (1982). and trial runs have begun on the neon-like ions of iron and selenium.
Contact Dr Keith Berrington at QUB for further information on the atomic R-matrix codes.
The molecular R-matrix codes describing the scattering of electrons by diatomic molecules are in the process of being ported to the X-MP and at the present time most modules are working correctly. This package has been developed from the ALCHEMY quantum chemistry program and is being used to study photoionization, dissociative recombination and other processes as well as electronic and vibrational excitation. Contact Cliff Noble (CJN at RL.IB) at Daresbury Laboratory regarding the diatomic molecule codes.
CCP3 is concerned with providing theoretical / computational support in the area of surface science. The CCP3 library contains a number of widely used codes which are currently being mounted on the X-MP at RAL. These include:
The programs listed above model experimental probes of the geometry and electronic structure of surfaces and are routinely used to interpret experimental data.
The other main area of interest within CCP3 is in self consistent field calculations of the electronic structure of surfaces. Several groups are in the process of developing codes using a variety of basis sets and multiple scattering methods. These programs are quickly outstripping available computing resources at other centres, so that in the medium term it is planned to mount and further develop them on the RAL X-MP and thus make them available to a wider community.
In the longer term, we plan to develop codes to attack the chemisorption problem, that is, to determine the electronic structure of a cluster of atoms adsorbed on a surface. This is an ambitious project, which requires as a starting point the computational facilities afforded by the X-MP.
For further information contact Richard Blake at Daresbury (RJB2 at RL.IB).
This project is concerned with the computer simulation of condensed matter, both fluid and solid. Both dynamic (Monte Carlo and Molecular Dynamics) and static (energy minimization with respect to system geometry) techniques are used.
CCP6 is concerned with the theory of atom/atom, atom/molecule and molecule/molecule collisions. Classical, semi-classical and quantal methods are used. In the near future it is hoped to implement the most recent version of MOLSCAT on the X-MP. For information on the progress of this work contact Jeremy Hutson (Chemistry Dept., Durham or JMH2 at RL.IB).
Robert Allen (Daresbury Laboratory) is planning to implement the modified versions of ATMOL and MRD-CI on the X-MP, to allow calculations on charge exchange reactions to be carried out. Contact RJA at RL.IB for further information.
Collaborative Computational Project No. 7 (CCP7) is concerned with the analysis of stellar, nebular and interstellar spectra. As well as implementing existing programs for modelling astronomical objects, CCP7 is concerned with incorporating improved atomic data and algorithms into those programs whenever possible. Users of CCP7 implemented programs have traditionally used File Transfer Protocol to transfer their predictions to a local STARLINK VAX for direct comparison with observation.
Now that "red-book" Job Transfer and Manipulation Protocol (JTMP) is available for the submission of batch work to remote machines, it seems desirable to develop VAX/VMS command procedures for running CCP7 implemented programs on a chosen CRAY (XMP, COS1M or COS2M) directly without having to separately initiate file transfers for input and output. It is hoped that the successful development of VAX/VMS procedures using JTMP, will remove the need for STARLINK users to have a working knowledge of the front-end machine operating system and the connecting network. The peer review committee might allocate time on any of the three CRAYs at its disposal to a prospective grant holder wishing to use CCP7 implemented programs; indeed he might be moved from one CRAY to another, and he would only need to change the value of the CRAY parameter in his VAX/VMS procedure to make the change. Work is currently in hand to develop VAX/VMS command procedures to simplify CRAY batch job submission following the guide-lines described above.
Most CCP7 use of CRAY computers has been for calculating model stellar atmospheres and line formation, without assuming Local Thermodynamic Equilibrium (LTE). When LTE is not assumed (ie. non-LTE is adopted), the statistical equilibrium equations have to be solved simultaneously with the radiative transfer equation to yield ionisation fractions, level populations and the radiation field. Non-LTE calculations are clearly made more realistic by increasing the complexity of model atoms considered, this is especially true for line formation computations. The large additional memory available on the CRAY-XMP/48 is therefore advantageous to non-LTE calculations, and non-LTE codes will therefore be implemented on the CRAY-XMP/48 as soon as possible.
CCP9 envisages that the increased memory (to close to 8 Mwords) will be most beneficial to its work. All the electronic structure codes will be able to deal with more complicated structures, and more accurately, than was previously possible. Examples of potential new applications are:
The recently renewed CCP9 band theory project. based on the relativistic and spin-polarized KKR method (with full spatial dependence of the potential) counts on the availability of the memory and speed of the X-MP for its production work. Contact Walter Temmerman at Daresbury Laboratory (WMT at RL.IB) with queries regarding CCP9s activities on the X-MP.
