There have been no changes in the group this quarter: Mike Chiu and John Lewis have relinquished their FR80 responsibilities, and Jean Crow is now known as Jean Nunn-Price. The present composition is:
| D A Byfield | Secretary | (DAB) |
| M F Chiu | Attached to FR80 | (MFC) |
| J E Nunn-Price | (JEN) | |
| M Davies | (ND) | |
| M Elder | (ME) | |
| E M Gill | (EMG) | |
| M F Guest | (MFG) | |
| P Kent | (PK) | |
| J W E Lewis | ½ attached to FR80 | (JWEL) |
| P A Machin | (PAM) | |
| V R Saunders | (VRS) | |
| P F Smith | (PFS) | |
| B Stokoe | (BS) | |
| G Symes | (GS) | |
| A J H Walter | (AJHW) |
Many members are now working in areas not reported on here; it is assumed that a description of their work appears in other reports. We expect this to be the last Applications Software Group report in this form, with the impending reorganisation of the Atlas Computing Division.
Work has commenced on the project designed to investigate the competitiveness of valence bond calculations in relation to other methods. Dr Gerratt (Bristol) visited the Laboratory for discussions on the project and supplied a source version of the valence bond program developed by Dr Raimondi at CECAM. This program has been implemented on the 360/195 and the results of a test calculation of Raimondi successfully reproduced. Inspection of the code however, suggests the need for a substantial rewrite of parts of the program before the competitiveness of the method can be seriously gauged. The present version is limited to very small systems since all integral manipulations are performed assuming the full list of two-electron integrals is resident in core!
The first draft of the write-up of the bonded function CI program has now been completed and is at present in the course of revision.
More users have expressed interest in obtaining results from the CI package and collaborative projects are at present in progress with:
The calculations on the electronic spectrum of H2S which were undertaken as a full-scale test of the capabilities and limitations of the CI program have now been written up in draft form and will shortly be ready for publication.
Work has been comp1eted, at the request of Prof J N Murrell, on the interfacing of the Sussex valence electrons only program to the ATMOL3 system. This entailed a complete rewrite of the cumbersome I/O system, with the incorporation of the ATMOL3 dynamic core feature and file structure, although many of the routines have been left unaltered to aid future developments of the system by the original author. (I G Vincent, Sussex).
A version of tle ATMOL2 graphical analysis program is now available in ATMOL3, which provides compatibility with the FR80 and incorporates the ATMOL3 dynamic store feature and file structure. The possibility of interfacing the Archuleta 3D surface plotter is being investigated as a replacement for the somewhat cruder 3D-histogramming package SCAT3D. Early results are encouraging although the core required at present by the new routines is unacceptable for large grids. Suitable rewriting of the packing routines in 360 Assembler should ease this problem.
Much work has been put into the ATMOL version of the Stevens' Slater integrals program, with a view to generally bringing this system up to the standard of the Gaussian integrals program. Most of the work is now complete and has involved the incorporation of new data input routines (which allow the use of the ATMOL3 geometry generator), a general cleanup of the coding (which involved the coding of two new driver routines, one for the one-electron integrals and one for the two electron integrals) plus the incorporation of improved numerical quadrature procedures which we hope will cure a number of vexing problems which have faced us ever since the program was first used by us (about three years ago).
Considerable work has been put into the ATMOL3 write-up, although much work remains to be done before completion of the manual (perhaps about 2/3 is presently complete).
Eighteen data sets were collected this quarter, which makes 35 in the past six months; a rate of 70 per year, the highest yet achieved. The second full year of the film scanning service ended on 31 March, and, looking back, was just as successful as the first. There were fewer scanner and peripheral maintenance problems this year, and the number of processed structures was increased by 35% from last years 43 to this years 58. We receive no feed-back from some users, but have been informed that 41 microdensitometer data sets have been solved satisfactorily to date. For those that have been fully refined the R-factor range is 4.7-16.0%, mean 7.5%, median 6.9% from a total of 30 structures. Weissenberg films continue to predominate, but we receive regular batches of protein precession films. Both the PROXIMA programs have run satisfactorily and the recent modifications have proven very successful. We received a set of oscillation photographs from a poor crystal of a mercury salt, and a combination of scanner, 1906A and FR80 produced a satisfactory set of data for the crystallographer.
The programming work involved in interfacing the X-Y recorder to the PROXIMA programs is substantially complete. The plotter can now be used to display the spots which have been located on the film, and to outline the integration boxes. This proves very useful for dealing with difficult films, when it is necessary to track down the source of the spurious spots that scanning produces, or the cause of overlapping integration boxes. The display also enables the operators to choose an origin spot for indexing purposes without waiting for a preliminary 1906A run. Thus the plotter is already justifying the reason for its purchase by increasing the rate at which film packs can be processed.
The SHELX system has been used to solve and fully refine an unusual 5-coordinate copper complex, starting from a densitometer data set, as an extension of the usual service. Supervision of an AERE scientist who is finishing the crystallographic part of his PhD is proving fruitful. With some assistance he has solved three structures, two from densitometer data sets, in the past few months using SHELX.
The work has progressed smoothly this quarter, greatly assisted by Dorothy Froud who has been borrowed from Resource Management Branch. The full set of inverted files from the tape of bibliographic information have been produced on the 360/195 and transferred to the Oxford PDP-10 on 7-track tape (with Mike Chiu handling the tricky 32-bit word + EBCDIC to 36-bit word + ASCII interface) . These files have been fully checked out and are now error-free, as nearly as we can determine. A number of errors present in the original system have been located and corrected. We have added a new retrieval command to search for 4-letter fragments in the titles of structures, and added a facility to check initials to the author's name retrieval. Work on the more difficult inversion of the tapes of crystal data is now underway.
ATS has been fairly extensively used during this quarter but the secretarial staff have not used it at all. A mistake by IBM means that the replacement 2741 terminal which arrived last autumn, and which was only got working after many delays, was in fact the wrong one. If we want ATS with all the facilities we originally asked for (to be compatible with ACL House Style) the terminal will have to be changed yet again. The new one will have to work in 'correspondence code' mode and will not be available to ELECTRIC even when ATS is not running. There are alternative facilities: the text-layout system in ELECTRIC is much more limited th an ATS but has been improved recently and more improvements may be in the pipeline. The Magnetic Card Typewriters in the typists' office offer most of the facilities of ATS without any of its drawbacks.
This is completed satisfactorily, and glossy colour brochures advertising the film have been designed and printed. These will be sent to prospective purchasers, and also to the Press and others invited to a special pre-release Press showing of the film in mid-May. A number of committees of the Engineering Board have visited the Laboratory and have seen the film. Comments have been favourable, and interest has been shown by some visitors in making a similar film in their own field of research, for which SRC assistance would be sought.
The package to run on the PDP15 to provide an interactive graphics facility to finite element packages running on the 1906A has provided its first successful runs with PAFEC. However, facilities are still limited. An example of this is that after a node has been added to the finite element mesh, it cannot be deleted, but work to solve this problem as well as a few other minor problems is well under way and should be completed shortly.
Work has started on a Rutherford "library" of laser computer codes for use in connection with the Laser Centre currently being equipped. Initially no new programs are being written, but copies of existing programs are being transferred to the IBM 360/195 with some modifications due to the transfer.
The CASTOR 2D code, and the MEDUSA ID code are now held at Rutherford, as well as the home-grown ZIP ID code, which also runs on the 1906A.
In February, the ABRC visited the Laboratory and a short demonstration film was made in colour on the FR80 directly from the ZIP program, showing a simple 1D compression calculation. The film titles were animated by Dave Ralphs and Peter Dewar using SMOG.
The second processing of the raw data from S68 is going smoothly and is keeping up with the schedule; it should be completed by the end of the year. Most of the jobs are submitted at the RJE station at ROE. There have been some difficulties with the link but not enough to hold up the flow of work. Some assistance is being provided to ROE and UCL staff in debugging one-off jobs.
The Belgian half of the project is producing a catalogue of bright ultraviolet stars. It would be possible to print the catalogue from plates prepared from microfilm produced on the FR80. David Carnochan is producing experimental samples of various layouts; a decision whether to proceed with this should be made in the next quarter. A 'catalogue-on-a-fiche' may be produced in any case.
A new phase of the project has now begun; the system is being installed at places outside NERC with arrangements being handled by the Scientific and Technical Services division of NERC Headquarters. The first of these new installations is in a Cyber 72 in the Hague, for the Soil Survey and Geological Survey of the Netherlands. Because CDC do not support standard direct-access methods, extra time was required to alter the Controller and the low-level I/O modules to use the local version of direct access. At the end of a week the library was up and the Controller beginning to generate jobs, so that the local team were able to take over testing. It is planned to spend another four days in May clearing up any snags and helping to get the system into full production. We find that delays are often caused by an inability to get enough details of the operating system beforehand to 'tailor' the Controller to the installation, and are taking steps to refine the installation procedure.
Another new installation, much further afield but still within NERC (wearing its ODM hat) is at the Geological Survey of Indonesia, where Steve Henley is putting the final touches to a full G-EXEC system on their 370/153. Dispatches from the front indicate that the system was up and running in a week. Several IGS people are working in the area and have been G-EXEC users at home, so the local staff will have some experienced users to advise them.
A similar installation is underway at Harwell, on their 370/168, for use by IGS Edinburgh under a Department of Energy contract. This was an easy move in the first stages, as we were able to unload a PDS from a 3330 disk to tape and load it onto a 3330 at the other end - once we had discovered that (a) the RL and Harwell I/O routines are incompatible and (b) that FORTRAN H and H extended are also incompatible. Implementation is currently slowing down because private disks are 'bad news' at Harwell and turnround is correspondingly slow.
Future installations in the pipeline include one for the 360-370 complex at Newcastle (for a joint project the G-EXEC team are to undertake with ERCC) which the team will do themselves; one for the Manchester Regional Computer Centre, (on their 1906A/ Cyber complex), and one for the Rothamsted Soil Research Institute (on a System 4), both the latter being at present in the hands of local teams, although the G-EXEC team expect to be called in for advice during both projects.
Two of the team attended a seminar on Implementing Relational Databases at Southampton recently, and were pleased to find that G-EXEC already has most of the relational operators available in other guises.
The G-EXEC system on the UNIVAC 1108 at the Institute of Hydrology is now being used regularly from IGS, for a hydrology data bank preprocessing project, making use of the editor and interactive processing facilities. A proposal has been put forward that it should also be used for the Monthly Work Return and Costings processing for IGS.
There is still no news on the possible take-over of G-EXEC by NERC Headquarters, although possible. staffing arrangements of a group to look after it have been discussed.
The magnetic tape containing a CDC version of BOON Mark 1 has now arrived from Ljubljiana, Yugoslavia. This should be ready for distribution to CDC sites shortly.
The conversion to use SMOG instead of SPROGS has now been made. The complete set of BOON Mark 2 files is now in the pseudo directory :BOON2 in :NPIN02. The files are:
BOON2 the BOON2 macro GPLOT the FR80 macro CLPLOT the CALCOMP macro FR80F the FR80 graphics FORTRAN source CALCOMPF the CALCOMP graphics FORTRAN source FORTOT1 ) the BOON FORTRAN source FORTOT2 ) PLANTOT the BOON PLAN source
A number of requests for BOON Mark 2 have already been received. Distribution of the source will start shortly.
Lancaster have now received all the BOON Mark 2 files and have edited most of the routines to make them acceptable to the IBM 370 system at Daresbury.
Subroutines have been written and tested to act as substitutes for the ICL routines FMOV, GETARRAY, PUTARRAY, GETPART, PUTPART, ALOG2 and NINT. Routines are being written to act as substitutes for the ICL routines COMP, COPY, COPY8 and DATE. As the IBM 370 is considerably faster than the ICL 1906A the DUMP and RESTART facilities will not be included in the initial version.
Small differences were observed in the frequency response figures produced by the command FWTS in the 1906A and 370 versions. Possible causes of a loss in accuracy on the two machines are being checked.
M E Claringbold tried to get this system to run on the 360/195 but found that none of the three magnetic tapes containing the program were now readable. Oxford have now returned the card deck of the program to us and we are in the process of compiling the routines.
Kate Crennell has produced samples of the Bibliographic Index on the FR80 to demonstrate the quality we can obtain. From this preliminary evidence Dr Kennard's group at Cambridge feel that the quality is sufficiently high to justify pursuing the project further. The following technical details were demonstrated:
Italics were not demonstrated, but an italic font could be defined in the same manner as (b) above. Recent work in collaboration with staff at Rutherford has produced a semi-automatic method of rendering software GROATS fonts in FR80 "hardware", which may accelerate the availability of new fonts.
A suite of 360 Fortran routines has now been written which will
It is envisaged that the routines should be readily modified to produce an application-specific program (ie a program each to do the Bibliography, Author and compound indexes). The design has attempted to keep the requirements of the Structure Index (mixed text and graphics) in mind.
The work so far described involves programming on the 360 only and although "fine tuning" of the application-specific programs may take some time and require feedback from Dr Kennard's group, we should be able to produce "proofs" on the FR80 for the first 3 indexes in about 2-3 man-months. The experience gained in this period should enable us to give a realistic estimate of the time required for the Structure Index.
Enhancements to the FR80 software and hardware could lead to the following (without requiring extensive changes to the 360 programs):
The quarter has-seen considerable progress in the calculation of S(k,w) the vital quantity in a neutron scattering experiment. The computational effort has not been as severe as first feared. This project is beginning now to achieve its original aims, ie search out the collective nodes in simple fluids and relate them to the intermolecular forces.
A great deal of effort is being expended in getting the various components of the required molecular dynamics program to run almost perfectly. Close attention is being paid to the convergence of various series involved in these calculations. Serious production should begin sometime in the next quarter.
This quarter has been a period of consolidations results are being written up and a few final runs have been made. Two manuscripts are in active preparation and a third is projected.
Some time ago some Monte Carlo calculations were carried out in conjunction with Dr Hammersley and Professor Rawlinson of Oxford on a one-dimensional analogue of the penetratable sphere problem. This project has been revived in 3D and active preparation of the program is underway.
| Applications Software Group Meeting, 6 January and 12 March | ASGR |
| ALPS Progress Meeting at ACD 13 January 1976 | B Stokoe, P Kent, A J H Walter |
| Meeting with J E Hailstone to discuss the GEC 4080: User Interface Group, 2 February | P Kent |
| Meeting with GEC to discuss the 4080 | P Kent |
| Visit to ISVR Southampton to discuss minicomputers and the interactive facility, 19 February | P Kent |
| BOON Progress Meeting held at State House on 27 February 1976 | B Stokoe, P Kent |
| User Interface Group meeting held at ACD on 4 March | P Kent |
| Library Users Group meeting held at ACD on 4 March 1976 | P Kent |
| Meeting with Cranfield to discuss the interactive facility held on 23 March | P Kent |
| Implementation of G-EXEC on CYBER 72 at IWIS, The Hague, Netherlands on 22-26 March | E M Gill (IGS) |
| Implementation of G-EXEC on 370/ 168 at Harwell on 9, 11, 12 March | E M Gill (IGS) |
| Relational Database Seminar held at Univ of Southampton on 29, 30 March | E M Gill (IGS) |
| Geological Society of London, Information Group Committee held on 14 January and 4 March | E M Gill (IGS) |
| SRC Course on letter/minute/report writing (Senior Course) held at the College, Swindon 27-29 January | P F Smith |
| S2/68 Joint Executive Committee meeting held at Institut d-Astrophsique, Liege, Belgium on 3-4 February 1976 | P F Smith, B Stokoe |
| Visit to Leicester and Nottingham Universities Engineering Departments with User Interface Group members on 11 February | J E Nunn-Price |
| Visit to Imperial College and Hatfield Polytechnic with User Interface Group members on 17 February | J E Nunn-Price |
| Talk by John Hammersley , London on 27 February | J E Nunn-Price, B Stokoe, J W E Lewis |
| Attend CET committee at State House to show FE film and answer questions on 4 March | J E Nunn-Price |
| Visit to North London Polytechnic for talks on X-ray data processing 22 January | M Elder, P A Machin |
| Meeting on 'Interactive Computer Graphics in Crystallography', Cosener's House, 29-30 January | B Stokoe, M Elder, P A Machin |
| Meeting on 'Profile Analysis in Powder Diffraction', Rutherford Lab, 4 March | P A Machin |
| Meeting at ACD with ICL to discuss DAP, held 5 January 1976 | V R Saunders, J W E Lewis, M F Guest, M Elder |
| Meeting to discuss possible uses of FR80 for publication purposes, held at ACD on 22 January | B Stokoe, M F Chiu |
| Meeting of the Data Compilation Committee held in State House on 26 February | B Stokoe |
| ALLC talk held at ULCC on 17 March | B Stokoe |
| ATMOL2 - Appendix to User Notice 11 Graphics on the FR80 | M F Guest, M F Chiu |
