This quarter saw the departure of Peter Sutterlin, who returned to Western Ontario after his year's attachment. We look forward to seeing him again in November at the ACD/COGEODATA workshop. Two new students arrived for a year each this quarter, Mike Davies and Gary Symes. They are working with Jean Crow and Mike Elder. Simon Backer left us on 26 September after working for 6 months with John Lewis.
The present staff in the group are:
| D A Byfield | Secretary | (DAB) |
| M F Chiu | Attached to FR80 | (MFC) |
| J E Crow | (JEC) | |
| M Davies | (MD) | |
| 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) |
Restart facilities have now been incorporated in all the major steps of the MUNICH CI package. A new file conversion program has been implemented, interfacing the ATMOL3 system with both MUNICH-CI and MOLECULE-CI programs. Extensive tests on both closed and open shell systems suggest a ceiling of approximately 5500 configurations for the MUNICH system, a limit imposed more from consideration of peripheral requirements and elapsed times than the CPU requirements of such a job. This limit does, however, represent a significant enhancement of the system which, in the MUNICH environment, is limited to approximately 3000 configurations.
After lengthy discussions with Dr B T Sutcliffe (York), the design work has been completed for a configuration selector for the CI program. Based on first order perturbation theory, this technique should provide a quantitative basis for handling systems which formally lie outside the scope of the MUNICH-CI package. With the exception of this work, coding for MUNICH-CI and its interface to the ATMOL3 system is now complete. Whilst the lengthy process of documentation remains outstanding, the program is now ready for release and is in use by certain users, including Dr D Hirst (Warwick) and Dr W G Richards (Oxford).
This is now operational.
This is operational, and features a simplified file structure whereby the various property integrals may be output directly to a specified section of the ATHOL3 dump file. The data input has also been simplified, since the information specifying molecular parameters, which in ATMOL2 was presented on cards, is now extracted automatically from the appropriate dumpfile.
This has been used, in conjunction with Dr R E Overill (Leicester), in a study of NMR coupling constants for a series of first-row hydrides. Some preliminary results were presented by Dr Overill at the Leicester Quantum Theory Conference. Tests on a series of π-bonded systems have revealed an intrinsic divergence characteristic of this type of system, which regrettably presents a severe limitation on the overall use of the method.
The Boys localization program is now operational and appears to iterate very much more rapidly than the ATMOL2 version. This is due to the inclusion of a Newton-Raphson minimization scheme, and to improvements in the old method of 2 × 2 rotations. The latter method is still used to provide a decent starting point for the Newton-Raphson procedure. Procedures for the analysis of Boy's localized orbitals remain unwritten.
Work has been completed, at the request of Professor R N Dixon (Bristol), on the provision of a routine to calculate the J-part of the pseudo potential defined in his valence electrons only SCF method, This has enabled him to interface his program to ATMOL3, as it appears that the Bristol group wish to use the idea of pseudo-potential in the framework of configuration interaction, using MUNICH-CI.
The Monte-Carlo studies of our model are all but complete. Efforts are being made to extract the maximum amount of useful information from our data, in order to gain better insight into the behaviour of the model. In the mean time, Simon Backer has put the time-dependence into the model and has calculated some of the dynamic properties of the system. Using the stored data for the time evolution of the system he was able to create a series of short films which might give some clues to the nature of cooperative effects in this type of system.
This work has been fraught with various difficulties. First we were unsure exactly how the data was written on the tape. Having solved that, there were various problems with the FR80, concerned principally with use of grey levels. After some initial experiments a short piece of film was produced using a 'painting' procedure where atoms far away are relatively dark and those close to much brighter. Unfortunately, the overlap regions were reinforced too much to be useful, and this method may have to be abandoned.
Experiments with hidden line programs were a little more successful but this promises to be rather more expensive.
The Monte-Carlo evaluations of the pair distribution functions of a number of model simple liquids is complete. A search is being made for a propagating collective excitation in the liquid and a criterion for its existence can be formulated in terms of the Fourier transform of the pair distribution function, the so-called structure factor. Unfortunately, this means that one attempts to transform a truncated function immediately giving rise to severe problems.
It is possible to recast the problem introducing either a non-linear integral equation or two linear integral equations. Both these methods have been attempted with little success. A direct evaluation of the structure factor is now under way.
The guaranteed project with the Physics Department of ICST to attempt to simulate a pyroelectric material has now begun. Pyroelectrics are materials which, when heated, pass a small electric current. Their efficiency is not high so we are not contributing to solving the energy problem. These materials are used as infra-red detectors, and the pyroelectric effect is not fully understood.
We are still at the beginning stage, sorting out our basic crystal structures and basic computational techniques, but these are slowly falling into place.
We are still awaiting a couple of manuscripts from speakers at the conference.
Only 6 data sets were collected this quarter, mainly because of the slow rate of submission of films over the summer vacation. Tape problems also slowed the service up a bit, but we were able to make good use of the slack periods by developing and testing new programs. There is now a fully tested version of PROXIMA which is tailored especially for Weissenberg photographs with many weak spots. The darkest film is scanned, processed in the usual way on the 19O6A to index the strong spots, and then the coordinates of the missing spots are generated. This information is returned to the Alpha-16 on magnetic tape (may all manufacturers of computers who write even parity tape marks on odd parity tapes perish painfully) and the complete film pack is then scanned. Thus intensities are produced for weak reflections, but more importantly, the fact that certain reflections are weak is recorded; answering our major user criticism.
This has been used successfully to scan seven more packs of protein films. During program development time the precession program was modernised, and it is now capable of using any of the three scanning rasters, and dynamically allocates core storage depending on the size of the core in which the program is running and the parameters of the film being scanned. This should stop the proliferation of duplicate versions which differ only in minor details, but tend to get out-of-date with respect to each other. The same improvement should be made to the Weissenberg program soon.
In addition, the indexing of precession films can now be performed on the Alpha-16, which saves a bit of 19O6A time and helps the operators decide how successful the scan was without any delay. The routines are not fully checked out, but seem to work satisfactorily, and take surprisingly little time compared with the 19O6A FORTRAN version.
The General Scanner Program runs on the Alpha-16 to write data from the microdensitometer to magnetic tape. It was provided with the microdensitometer, but has recently been extensively modified at ACD. It can now be assembled and loaded on the 19O6A before transfer to the Alpha-16, which facilitates corrections and improvements. The use of different packing modes on magnetic tape has been rationalised, and provision made for listing part of such tape files on the teletype. The system for averaging data from the scanner before output to tape has been introduced so that the scanning aperture can now effectively be changed to suit the film being scanned, lifting the previous restriction to 100 × 100 microns. Work is in hand to allow a sampling, rather than continuous, scan to be used.
The magnetic tape deck gave trouble again, and cost us about a week when we could do little productive work. Sintrom were initially rather slow to do anything about our problem, but now realize how important the tape deck is to the microdensitometer and are giving us rapid attention.
Negotiations are underway with Optronics, Inc to have the two faulty Alpha-16 boards mended in the USA.
Considerable use has been made of the SHEL-X programs this quarter. Apart from the usual scale-merge procedure we have run a number of checks for crystallographers with difficult microdensitometer data sets, made three absorption correction runs for users with inadequate facilities and in three cases completely solved the structures. The SHEL-X system has proved itself to be a valuable alternative to XRAY74, and we have undertaken to make it available to other ICL users since this is easy for us to do, but difficult for Dr Sheldrick (the author) who uses an IBM 370.
A few improvements have been made to the graphical routine ORTEP. The system has been successfully interfaced to the FR80 via SMOG on the 1906A by Eric Thomas which removes the last reason for users preferring to use XRAY74 on the IBM 370/195 rather than on the 1906A. We shall continue to encourage all XRAY users of the 370 (apart from those with remote access terminals to the machine) to use the 1906A, with the added lure that on the 6A they can have good quality graphical output whereas the SD4020 package on the 370 has now become obsolete.
Copies of the system have been sent to ILL in Grenoble, the Leeds 1906A, the Liverpool 1906S and ABSCOR alone has gone to a Turkish university and to the University of Berne.
The protein refinement program has been run successfully on molecules from both Sussex and Oxford.
No further work was carried out on the Trinag-Lysozyme complex refinement because of lack of time. However, it is still of interest to complete this project.
This project has now been successfully completed. A further 250 frames of 35mm film were digitised and the layer line spacings measured using a FORTRAN program on the 1906A. The results of this work were reported by Dr Nwoye at an international conference in Copenhagen, arousing considerable interest. Some plans are being made for modifying the experimental set-up and measuring new data.
Dr Olga Kennard and co-workers in Cambridge have developed a data file of crystallographically solved molecular structures. Richard Feldmann at the National Institutes of Health, Maryland, has developed an on-line interactive and graphical program to access this data base.
Recently the Data Compilation Committee of the SRC have accepted the recommendation that the Feldmann program be tried in England. Potentially it is a powerful tool for crystallographers and chemists and might hopefully be made available to all universities. ACD is at present contributing by implementing and assessing the program.
The CSSR program runs interactively on a PDP10, using display facilities. It was obtained from Richard Feldmann and has been implemented on the Oxford Nuclear Physics computer. The implementation was successful except for a few faulty files which could not be read off tape. These have since been obtained from the US. Both the programs which comprise the system (one written in FORTRAN, the other in SAIL) have been compiled and built from their maintenance files.
Further time was spent using the program, testing, and locating errors. The majority of errors found so far have been traced.
The search program relies on a particular data file structure for speedy access. These files are generated from the Cambridge data tapes by the so-called 'file inversion program'. Work has begun on implementing this program on the ICL 1906A.
Attendance at this congress was a rewarding experience. In between the parallel lecture sessions it was possible to meet a lot of people with similar interests in microdensitometry, crystallographic computing and crystal data search methods. A number of the crystallographic papers presented were based upon the microdensitometer data sets. Problems were discussed with ten of our microdensitometer users, and a number of potential new customers were met (Tasmania, India, South Africa and Sweden, as well as crystallographers in Britain). Discussion sessions were arranged with Prof Abrahamson, who heads a Scandinavian network that is using Feldmann's CSSR system, with Profs Vainstein and Simonov of the Moscow Crystallography Institute to discuss Alpha-16 programs, and with a programmer from Stockholm who is using an Optronics Alpha-16 system similar to our so there is clearly a great deal of interest in film-scanning systems as low-cost alternatives to diffractometers, and the UK service based at ACD is now well-known and beginning to be imitated in other countries.
When our new terminal arrived we found that there were several things wrong with it and some confusion about what sort of port was available for it on the 195. As a result, we are still waiting for IBM to install the correct modem on the terminal and to supply the correct typing heads. The old terminal was out of action for 2½ weeks but, despite the pressure of other work, Marjorie Sherwen has completed the LEARN ATS course, Dianne Byfield has got three-quarters of the way through, and Cynthia Davis has had one or two sessions.
Detailed investigations have been carried out on what improved facilities could be made available on ATS. Compared to the magnetic card typewriter it is not well-suited to producing material containing complex mathematical equations and the 2741 golf-balls are not interchangeable with those on the ordinary electric typewriters used in the Laboratory. In most other respects it is a very useful tool. It is possible to obtain a combination of ATS and the magnetic card typewriter, but this would still not have all the facilities we use. However, IBM claim that they are working on putting in the major missing features.
The FR80 has been used with Roland Brandwood's help, to print ATS documents on various output media from magnetic tape and the ease with which the text can be centred in the frame seems to vary according to which output medium is used, hardcopy being the most difficult. It certainly works but there are some features which need improvement: many characters which are in the FR80 basic character set do not seem to be available to the IBM print program. Investigations are continuing on these matters.
A program to automatically triangulate the flow region and solve Laplace's equation over the resulting grid using Finite Element Analysis has been written and tested.
The next step, to iterate on the free surface profile using Bernoulli's equation, is now being prepared.
The final stages of production of this film have been reached. Colour tests are being carried out and the finished product should be available by the end of October.
Following a period of familiarisation with the operating systems of the PDP15 and 1906A, Mike Davies has written , in collaboration with Shuba Chanda, a new interface between SMOG and PAFEC. PAFEC can now produce pictures on all the FR80 devices as well as those previously available.
A system is currently being written for the PDP15, and should aid the running of finite element packages with interactive graphics. This system will run under the PIGS run-time environment, and is in an advanced stage of preparation.
The big changes expected at the time of the last progress report have not yet materialised. The reprocessing of the raw data has not started, the link to ROE is not available and we have heard very little more from the Belgians except that they seem to be continuing with the project. The reprocessing cannot start until David Carnochan (UCL) finishes the new reference star catalogue, which he has still not quite done. The operation of the link to ROE has been delayed by various technical problems but it genuinely seems likely to be working within a few weeks in a slightly different form from that originally proposed.
During the quarter all the outstanding program development work for the reprocessing has been completed and extensive production testing carried out on the main processing program. Peter Smith has still been heavily involved in running programs, particularly in arranging for David Morgan (ROE) to use the FR80. This was caused by an attack of panic when a notice was read saying that the SD4020 plotter was to be withdrawn at the end of September but it did not say that this only applied to 1906A users. As a result of this the programs found the SCSIM software in an incomplete state on the 195 and numerous bugs were duly uncovered. Bugs in our programs were very difficult to detect since they interacted with the SCSIM bugs.
Turnround time for FR80 output is almost 2 days from the time of job submission; this is because all FR80 tapes are logged and checked manually at RL, which will change, one hopes, when the spooling system is released. At the moment the tapes can easily go astray which also does not help.
The last three months have been ones of alternate progress and stalemate for the G-EXEC team. During August and September Keith Jeffery (IGS) and Elizabeth Gill spent some time at the Institute of Hydrology, Wallingford, putting G-EXEC on the Univac 1108. A total of 40 man-days was sufficient to move all the 40,000 lines of source including the Phase 4 Controller, and to perform some much-needed 'housekeeping' on the source. Some non-standard FORTRAN was removed, routines in BCD or mixed BCD-EBCDIC were converted to EBCDIC, and sundry stray errors and omissions repaired. The Phase 4 Controller is accessible in time-sharing mode from demand terminals (all the current 1108 terminals except the card-reader in the machine room), and generates a second-stage job in the batch stream. The object of this exercise was to allow load-shedding of IGS work from RL at some future date, and to test the machine-independence of the system. It also provided a relatively 'clean' set of source tapes for distribution.
The matter of distribution bes also been under consideration during the period under discussion: ACL and NERC have asked NRDC to look into the possibility of marketting G-EXEC, but no announcements have yet been made at this point. Meanwhile all requests are being held by IGS.
Preparations are going ahead for the Workshop on System Intercommunication to he held in November under ioint ACL/COGEODATA auspices; Marjorie Sherwen has been handling the administration and Keith Jeffery and Elizabeth Gill have been arranging the technical side.
At G-EXEC World Headquarters (an economically-constructed office on the roof of the Geological Museum) progress has been less than spectacular. We were very sorry to lose John Cubitt, to a post at the University of Syracuse in New York, in August; John contributed much of the statistical section of G-EXEC and is the first of the four "founder members" to go. Two other members of the Computer Unit, one programmer and the clerical officer, left within a few weeks. However the production load sustained by G-EXEC is still increasing and users are showing a gratifying tendency to do more steps per job, thus decreasing overheads.
The Phase 4 Controller is now taking over from Phase 3, with a number of new facilities. Its most obvious advantage is that the files on which it operates are resident on a permanently-mounted disk so that first-stage jobs can now run at Priority 12. This effectively doubles turn-round as second-stage jobs can usually run at Priority 8. Users can now get two turn-rounds per day instead of one. Other advances include in-line FORTRAN retrieval as an option, free-format input of commands, much reduced elapsed time as more use has been made of sequential files in building up the second-stage job, and improved accounting statistics.
New plotting facilities are being added by the Edinburgh G-EXEC section, with the adaptation of the GRAPE and ESOPP general-purpose graphics software developed there. This will fill a long-felt want in the G-EXEC package, which has always been short on graphics mainly because of the lack of a suitable plotter. The Edinburgh unit have access to a large flat-bed and will shortly have a fast off-line drum plotter, so that tapes prepared at RL using the G-PLOT package will be able to be plotted in Edinburgh without further processing.
Further plans for long-term development of G-EXEC are being hampered somewhat by lack of hardware, uncertainty about staffing, and most of all by the possibility that NRDC will take over G-EXEC completely and give it to a software house to deal with.
Final corrections are being made to both the program and the manual for Mark II. The package should be ready for release at ACL very shortly.
A copy of most of the Mark II routines has been sent to the University of Lancaster where work has now started on an IBM 360 version of the program.
The Institut "Josef Stefan", Ljubljana, Yugoslavia has now converted a copy of BOON Mark I to run on their CDC CYBER 72-24. This has been in use since the middle of September mainly by economists, students of the 'Electricity' Faculty and by Seismologists.
Unfortunately nearly all their potential users want to use BOON real-time. They are therefore having to rewrite the program in an interactive form.
A request for BOON has also been received from the Free University of Berlin for use on either an IBM 1130 or a CDC CYBER 72.
Since the last progress report work on ALPS has progressed. No major programming has been carried out, but a considerable amount of debugging has been done.
Output from ALPS has been compared with equivalent output from LUCS. This has brought to light a surprising number of differences between the LUCS manual and the MVSL program. Decisions are being made as to which is preferable, and modifications are being made accordingly. In a few cases ALPS differs from both the LUCS manual and the MVSL program by design.
The manual has been written, but corrections are still being made, and alterations made as the 'faults' described above are corrected. The sections of the manual are still too difficult to read for the everyday user (because of forward references) and this difficulty is being given considerable thought.
A few of the table output options have still to be provided.
A version of ALPS may be released internally shortly for anyone wishing to try it out - Volunteers welcomed! The manual will follow in a few weeks.
| Report on visit to USA, 30 June-JO July | P A Machin |
| Interim Report on work with the NIH Crystallographic data retrieval system 21-23 July 1975 | P A Machin, T W Thacker, J Clarke |
| Experience with the Crystal Structure Search Retrieval Program (CSSR)~ 10 September 1975 | P A Machin, M Elder |
| Crystal Structure Search Retrieval Program - File Conversion on 1906A, 30 September 1975 | P A Machin, M Elder |
| XRAY-74 Graphics on the 1906A 8 September 1975 | P A Machin, M Elder, R E Thomas |
| Distribution of XRAY74 to ICL 1906A's 3 September 1975 | P A Machin |
| Talk on the Microdensitometer and Crystallography to Resource Management Branch, 1 July 1975 | M Elder |
| Meeting with NRDC to discuss commercial exploitation held at ACL on 8 July | B Stokoe + others |
| Visit to USA - Discussions with Richard Feldmann at NIH about his Crystal Structure Search Retrieval Program, 31 June-11 July | P A Machin |
| Privileged use of the PDP10 in Oxford Nuclear Physics - for Implementing CSSR 21-23 July, 21-23 August | P A Machin, M Elder |
| Discussions with Prof Allen in Oxford about the use of the PDP10, 30 July | P A Machin, J Howlett |
| Discussions about CSSR with Dr Olga Kennard and O S Hills at State House | P A Machin, J Howlett |
| Tenth International Congress of Crystallography Amsterdam, 7-15 August | M Elder, P A Machin |
| 'Computational Problems in Quantum Chemistry' held at Institut de Chimie, Strasbourg 3-5 September 1975 | V R Saunders |
| 'Beyond the Hartree Fock', conference held at Villiers Hall, Leicester | V R Saunders, W R Rodwell |
| IBM Education Centre, Sudbury, to discuss APL on 23 July 1975 | J W E Lewis |
| Discussions with Prof K Singer at RHC on 24 July 1975 | J W E Lewis |
| Discussions with P J Grout at ICST on 25 July/ 21,28 August / 4 September 1975 | J W E Lewis |
| Data Compilation Committee: meeting on publications at State House on 6 August 1975 | B Stokoe |
| ALPS meeting on 12 August 1975 | B Stokoe, P Kent, A J H Walter |
| University of Surrey to attend seminar introduced by Prof Edwards on "Work of the Research Councils" given during the AGM of the BAAS on 29 August | B Stokoe |
| Discussions with Dr Luckhurst at Southampton University on 2 September 1975/24 Sepemb cr 1975 | J W E Lewis, S A Backer |
| Computational Physics Conference, Glasgow Univ 26-29 August 1975 | J E Crow |
| Talk by ICL at ACL on "Distributed Array Processor" on 11 July 1975 | J E Crow, B Stokoe |
| Visit to engineering firms in Oslo, Norway, 30-31 July 1975 (not officially sponsored but done during holiday) | J E Crow |
| Royal Garden Party, Buckingham Palace 17 July | B Stokoe |
