This Newsletter contains interesting articles on Poplog, a portable toolkit for multi-language programming, the ESPRIT CAD*I Project, AI Support for Engineers and Special Interest Groups (SIGs). We also have a pen portrait of the Engineering Computing User Group (ECUG) Chairman, Professor Neil Munro, a report on the first ECUG meeting, a progress report on the Transputer Initiative and many other items of information for everyone.
The next issue of the Newsletter will cover the period July to September, and future publications will be published quarterly. Please keep the articles coming in. The Newsletter is your forum for your thoughts, comments, ideas and discoveries.
Neil Munro was appointed a Lecturer in 1969, Senior Lecturer in 1973 and Professor of Applied Control Engineering in 1979 in the Control Systems Centre, University of Manchester Institute of Science and Technology and he was Head of the Control Systems Centre (1979 to 1981, 1983 to 1985). He is the author or co-author of over 100 papers, and editor/author of 4 books, concerned with the development, implementation and application of computer aided design techniques for multi-variable systems, Honorary Editor of the IEE Proceedings Part D: Control Theory and Applications, and was Chairman of the SERC Control and Instrumentation Subcommittee (1983 to 1985). He had previously spent some 10 years in industry as an electronics technician, development engineer and senior design engineer, working for Barr & Stroud Ltd., Ferranti, Parkinson Cowan Measurements and ICL.
Professor Munro holds several research grants from the SERC. His current research interests are in the creation and development of a new software infrastructure to support current and future CAD facilities for control engineering, the use of expert systems in the development of an adaptive intelligent Man-Machine Interface and in the development of control system design techniques, and the extension of Rosenbrock's design method for multi-variable control systems to large composite systems. In addition, he has a particular interest in the use of graphics workstations and interactive CAD facilities in the pursuit of this research.
Professor Munro is an experienced user of the Perq, Prime and Sun systems of the ECF. This has given him wide knowledge of the facilities available so that he is able to represent the users' interests in the years ahead.
The recently announced Initiative in the Engineering Applications of Transputers has attracted enormous interest from the UK academic community. The mailing list has grown to 530 with many people from non-engineering fields included.
The main mechanism for informing the community of progress will be the regular monthly mailshots, the first of which was sent out at the end of April. Included in this mails hot was information on how to apply for loan pool equipment and an invitation to tender to become the National or one of the three Regional Centres. The establishment of an Interim Facility for Transputer Applications at RAL was also reported. This has been agreed to ensure Centre-like facilities are available from the outset and to overcome the short-term funding problems for setting up the Centres at academic sites.
Loan Pool equipment will begin to be allocated at the end of May and successful applicants should expect to receive their allotted equipment by the middle of June.
The National Centre and first Regional Centre will be operational by 1 October 1987. Decisions on the locations of the other two Regional Centres will be made at the same time, but these can only start to operate on 1 April 1988 provided the continuation of the Programme beyond the first year is funded. There is no new information on the funding of the Programme beyond 31 March 1988.
If you wish to be added to the mailing list and receive the first and subsequent mailshots. please contact me.
Computer Aided Design (CAD) and Computer Aided Machining (CAM) systems are being adopted by manufacturing industry at an ever increasing rate. There is an urgent necessity to connect different systems. different departments and even different companies into a Computer Integrated Manufacturing (CIM) environment. Such a system might consist of a network of information processing and gathering devices distributed over a whole manufacturing plant and covering all information needs based on a common data base and data structure.
The European Programme for Research in Information Technology (ESPRIT) Programme. set up by the Commission of the European Communities, has, as one of its purposes, the support and encouragement of research in CIM. From time to time the Commission invites proposals from interested consortia; as partners, each consortium must have organisations from at least two member states, and when possible, a balance must be struck between commercial and academic institutions. After approval of a project, the Commission contributes 50% of the total cost; the other 50% must be found by the partners, hence commercial companies expect market advantages from their participation.
ESPRIT Project 322, known as CAD Interfaces (CAD*I), is one of the largest initiatives supported in this way. There are twelve partner organisations. from six member states of the Community. They are:
The goals of the Project, which began in November 1984, are to use software technology to develop three main interfaces:
The Project is managed by a Board with a member from each organisation. This Board meets four times a year; two of these meetings are attended by a Review Panel set up by the Commission and consisting of independent reviewers and Commission staff. To carry out its programme, the Project Board has set up eight Working Groups, each concerned with specific tasks and having representatives from the organisations (not only Project Board members). From time to time, two or more Groups hold "harmonisation meetings" to ensure that work is not being duplicated and that any group may have knowledge of other Groups' results.
Rutherford Appleton Laboratory, together with DTH, ERDISA, GfS, LMS and NEH, is involved in developing the last of the interfaces mentioned above. So far, much of our effort has gone into the design and development of a specification for "neutral files". A neutral file is a means of storing information provided by, or needed by, applications programs such as CAD systems, finite element modelling and analysis programs, and for experimental systems (for vibration testing of components and assemblies - even whole cars). It must be possible not only to control experiments but also to analyse and store the results.
The form of the neutral file is such that it is completely independent of the programs which use or produce its data; a typical file looks rather like a program source, and must be interpreted and dealt with in a manner analogous to compilation. The software required to interpret neutral files is produced by using modem techniques for generating parsers and lexical analysers - the same methods which are used to generate compilers for languages such as Fortran, C and Pascal.
The neutral file specification is now in a well-developed state, and we are producing the software for writing and interpreting the files. There is a continual interchange of files between the organisations involved; in this way we are able to check the consistency of each other's software. Each partner is committed to installing the new software at its own site and to demonstrate that it can be used in conjunction with such packages as exist there. At RAL, we shall interface with MEDUSA (CAD), FEMGEN (finite element modelling), NASTRAN (finite element analysis) and FEMVIEW (finite element post-processing).
Since we began, similar initiatives have been started elsewhere, in particular, in the USA and by the International Standards Organisation (ISO). We must take account of these, so we send representatives to meetings in the USA and to the ISO meetings. In many respects, our work is more advanced than their work, so we are having a strong influence on the development of their standards.
The present three year contract ends in October 1987. However, the original proposal was for a five year programme, so we are currently producing a case to continue the work f{)r a further two years. The consortium is considering whether to put forward a proposal for the next phase of ESPRIT, which is about to commence. Perhaps the most satisfying results which have come from the collaboration are the excellent working relationships which have developed between different types of organisations from several different countries with their own mother tongues. We feel that it will be a loss to us all if we cease to work together when the Project comes to an end.
Support for the Engineering Board Prime computers has, until recently, been split between the Informatics Division at RAL and UMIST. UMIST provided the user support and support for the system utilities; Informatics Division supported the operating system itself (Primos). The latter work included accepting new versions of the Operating System from Prime, implementing the SERC modifications and releasing it to UMIST for distribution to sites.
It has been decided that, in future, support for Primos will also be carried out at UMIST. RAL will continue to have local Managers for the various on-site Primes, but Informatics Division will no longer have any extra expertise in the Operating System. Any problems in this area will be referred to UMIST. Management responsibility will remain with RAL.
Informatics Division is currently investigating Primix, the version of UNIX which is hosted by Primos. It is intended to run a trial service on RL.PA shortly. Informatics Division will retain responsibility for this system.
Salford University, under contract to SERC, have developed a version of the Job Transfer and Manipulation Protocol (JTMP) to run on the SERC Prime systems.
JTMP allows users to transfer, and monitor the status of, jobs to remote sites, which support JTMP jobmill facilities. Examples of this are the AMDAHL machines at UMRCC and ULCC, and the Cray X-MP/48 at Rutherford.
For several months, an experimental SERC Prime JTMP service has been offered on the UMIST Prime 9955 to selected users. JTMP has now passed its acceptance tests and will shortly be made available to other SERC Prime sites. Interim documentation is available from User Support at UMIST. The full documentation will form part of a Networks Manual currently in preparation.
Should Poplog, at present part of the Alvey infrastructure, be added to the the SERC Common Base? Originally developed for AI teaching and research at Sussex, it is now sold as a portable, integrated, toolkit for multi-language programming. It includes incremental compilers for Prolog, Common Lisp and POP-II in one environment, on a range of machines (e.g. V AX, M68020, GEC-63) and operating systems (VMS, and various Unix systems including SUN). Incremental compilers are much faster than interpreters, while retaining many of their advantages.
To speed up software development and testing, the editor, help system, compilers and other tools, are combined in one process, avoiding context switching. Programs in different AI styles (e.g. procedural, declarative, object oriented) can be combined with conventional programs and libraries (e.g. NAG). Poplog supports dynamic external linking for programs in Fortran, C, Pascal, etc.
The portable Poplog VM (virtual machine) and machine-specific run-time code generators are driven by a set of compiler-building tools, used by LISP, Prolog and POP-II. Users can add incremental compilers for new application specific languages. Existing users have added SCHEME, ML and object-oriented languages. The development tools are automatically available for new languages.
The library includes parsers, expert system tools, problem solvers, image analysis programs, and a graphical tracer for Prolog, all with full sources. The teaching library, with many illustrative source programs helps users pick up AI concepts and techniques quickly.
There are over 520 commercial and academic licences world-wide, and the number is growing. About 95 UK educational establishments have Poplog licences for teaching and research in AI, cognitive science and computer science. Uses include: expert systems, friendly front ends, compiler development, software tools, word processing, a Unix mail front end, graphics, speech processing, music synthesis, image analysis and interpretation, VLSI design, real-time control, simulation, chemical synthesis, natural language, computer-aided instruction, logic programming, a spread-sheet, and two of the Alvey large demonstrator projects.
The Poplog User Group (PLUG) runs a newsletter, workshops, a software- exchange service and an electronic mail forum (run by HP Labs).
For many, the key feature of Poplog is POP-ll, an incrementally compiled language bridging AI and conventional programming. It is a much enhanced dialect of POP-2, developed at Edinburgh University for AI research. Like FORTH it is stack-based. POP-11 includes facilities with the power of COMMON LISP (and SCHEME), e.g. indefinite precision arithmetic, ratios, complex numbers, arrays, vectors, records, and full lexical scoping. In POP-II but not in COMMON LISP are a process mechanism, a pattern matcher and compiler development tools.
Experienced programmers find POP-11 a useful entry language for AI techniques because of its PASCAL-like syntax, though it is not restricted to AI. In Alvey News, Dec. 1986, a survey reported that POP- 11 was one of the few languages rated good by all users. It is more readable than LISP, enables more compile-time syntax checking, has a richer set of control structures and is more suitable for teaching. Two tutorial books have been published and more are on the way. People who have learnt POP-11 can easily switch to Common Lisp.
To illustrate the syntactic differences, here are two exactly equivalent function definitions (neither of which is intended to be optimal):
(defun member(item list)
(cond ((null list) nil)
((or (eq item (car list)) (member item (cdr list))) t)
(t nil)))
define member(item,list); if null(list) then false elseif item = front (list) or member(item,back(list)) then true else false endif enddefine
Or, using the POP-11 pattern matcher:
define member(item,list); list matches [ == ^item == ] enddefine;
This supports logic programming, list and array processing, number crunching, and sophisticated control structures, including co-routines, active variables and exit actions. All data-types including compiled functions and non-list data-structures, device records, etc. are automatically garbage collected when no longer accessible. A heap-locking facility reduces garbage collection time. VM code-planting procedures allow users to add incremental compilers for new languages. These immediately run on all machines supporting Poplog.
The editor, VED, can be tailored for different terminals, different languages, and different user requirements, including word processing and reading and sending e-mail - an EMACS-like conversion.
Re-use of software is encouraged in various ways. Layered saved images simplify sharing of pre-compiled programs: not always possible with AI languages. Modifiable search-lists for program and documentation libraries, used by the compilers, the editor, and the HELP facility, enable groups of workers to collaborate easily. Local libraries can be added for projects, groups of students or individuals. An online browsing tool for software and documentation is based on the editor. TEACH files explain concepts and provide programming practice, while HELP files summarise facilities. Different categories (including user- defined categories) of search lists are supported - e.g. utilities, packages, help, teach, news, etc.
Version 13 (summer 1987) includes a sophisticated remotely accessible window manager, complete Common Lisp and improved tools for linking external languages. Planned future developments include greater efficiency and new facilities, e.g. more efficient OOP, higher level window-based tools and debugging aids, and tools for producing reduced 'run-time' systems.
Additional language compilers may be added (e.g. Standard ML). Poplog will be ported to new machines, e.g. Transputer and Orion.
Several users have provided tools. Brighton Polytechnic have produced a Rule Based Frame System. The GEC STRATA system and the KEATS system developed by the Open University and British Telecom are being ported to Pop log. An equation simplifier will be available from Amherst, Massachusetts. Aberdeen University will port the window manager for remote use on Atan-ST. Exeter are porting Poplog to Pyramid.
Sussex University provides UK academic licences. Commercial and overseas users are supplied by Systems Designers except for the GEC-63 version marketed by GEC. Information for academic users is available from:
Ms. Alison Mudd,Non academic or non-UK users should contact:
Systems Designers Plc,R Barrett, A Ramsay and A Sloman: POP-11: A Practical Language for AI, Ellis Horwood and John Wiley, 1985, reprinted 1986.
The first meeting of the Engineering Computing User Group was held at Rutherford Appleton Laboratory on Thursday, 2 April 1987, chaired by Professor Neil Munro, and attended by over 100 people.
After welcoming all attendees, Professor Munro highlighted the importance of such meetings as they affected the user community, in particular, they afforded the important opportunity of making their views and requirements known to the policy forming Committees of the Engineering Board.
Included in the main part of the meeting was a series of reports by the members of the Engineering Computing Facilities Executive (ECFE) on the work currently being undertaken, which includes work on graphics, central file servers and the engineering applications of transputers. The morning session ended with a presentation on the use of the Silicon Graphics IRIS workstation in the Strathclyde University's ABACUS Project.
The afternoon was taken up with parallel sessions covering the GEC, Perq, Prime and Sun user communities.
During the final session, special requests for action to the Computing Facilities Technical Advisory Group (CFTAG) were collected by the Chairman.
The next meeting of the ECUG will be held on Wednesday 23 September 1987 at RAL.
Due to the recent increased interest in the uses of Artificial Intelligence (AI) tools and techniques within different branches of engineering, the Computing Facilities Committee (CFC) of the Engineering Board of the Science and Engineering Research Council has awarded a contract to the Artificial Intelligence Applications Institute (AIAI) at the University of Edinburgh. This contract is to provide support for SERC funded engineers who wish to investigate the application and applicability of AI to their work.
AIAI was established by the University of Edinburgh in 1984. It is located in central Edinburgh where it shares premises with the University's Department of Artificial Intelligence. The purpose of AIAI is to act as a technology transfer unit, facilitating the use of AI tools and techniques on real problems by communities that are not AI specialists.
AIAI has a rich collection of academic and commercial AI software including AI programming languages, expert system shells, tools and application software, and a wide range of hardware, including:
AIAI also investigates the use of innovative computer architectures, and hosts examples which can be accessed by potential users.
Support for this contract is provided principally by members of the Knowledge Representation and Expert Systems group (KRES) within AIAI. KRES will collaborate with groups and individuals who wish to apply AI in their work. This can cover a wide range of activities from general advice on hardware and software selection to collaboration in the construction of application systems.
Any project which receives funding from SERC Engineering Board is eligible for support to use these facilities; other projects funded by SERC may also be supported if suitable facilities are available. Preference will be given to projects that are working outside the areas of computer science and artificial intelligence.
If you would like further information, please contact:
Mrs Natalie HoneymanIn accordance with the decision of the site managers to freeze the GEC operating system, the next release of OS4OOO, version 36, will be the last to be installed on the machines listed below. After that installation, no further system development or modifications will take place.
The enhancements will include the final version of mail and hopefully a fix to enable Auto-IPL to succeed.
The following are the sites to which the above will apply:
The installation is expected to be completed by 30 June 1987
SIGs were originally formed in 1977 following open meetings of various engineering communities to advise the Director of the (then) Rutherford Laboratory on the mounting and support of applications software on Interactive Computing Facilities (ICF) computers by staff at RAL. Some SIGs disbanded after making their recommendations; others regrouped as smaller steering committees to oversee the work as it progressed.
A short history of each SIG and its current status is now given.
The lack of a LAN connection between Perq and Sun was discussed at the Engineering Computing User Group meeting on 2 April. Our preferred solution is to use some Newcastle Connection work of Robert Stroud (Newcastle University), but make some small essential changes. Our target is to have a demonstrable connection between Perq and Sun by the beginning of July. After that time, the software for Perq and Sun will be available on an unsupported basis.
An annual open meeting, organised by the Special Interest Group in Electromagnetics (SIGEM), was held at RAL on Wednesday, 8 April 1987. There were 77 attendees, mostly from Universities.
The morning session was taken up with a description of the current trends in computing provided by SERC. Dr B Davies (SERC Director, Computing) outlined the provision of the Cray XMP/48, followed by Dr Mike Jane describing the Transputer Initiative and Julian Gallop reviewing the Single User programme. Prof E Freeman (Imperial College) gave an indication of the effect of the new developments on computing in electromagnetics.
Although some concern was expressed over the planned demise of the multi user mini computers, there was plenty of excitement shown over the potential use of single user systems, such as the Sun range, and the Transputer systems. There was also interest shown in the fact that both TOSCA and PE2D have now been mounted on the Cray at RAL.
In the afternoon session, the presentations were of a more technical nature, starting with a progress report of the tripartite eddy current research project between Bath University, Imperial College and RAL. Both Bath and RAL have working 3D transient codes for use as test beds for further development work. As an aid to data transfer, a Neutral File developed within the project was also described.
Dr R Chignell (ERA, Leatherhead) then gave a brief overview of some of the varied techniques that were necessary when solving higher frequency problems. In particular, he outlined methods suitable for incident waves having wavelengths of the same order of magnitude as the dimensions of a scattering structure. He then distinguished between cases where the size of the structure was less than a few wavelengths, larger than 10 wavelengths, and the resonance region in between.
This was followed by a review of methods suitable for optical frequencies in guided structures, given by Prof B Davies (UCL). Many results were also shown of typical integrated optics devices, solved using the Finite Element method.
As in the morning, many questions were raised, and a lively and stimulating discussion followed.
I would like to thank all those who gave presentations at the meeting, and to all who attended and helped make it a successful event.
The UK Chapter of EUROGRAPHICS in association with the BCS Computer Graphics and Displays Group, ACADE and the GHOST and GINO-F Users Groups is organising the 6th UK Conference on Computer Graphics.
The Conference Programme will consist of papers selected from those submitted together with invited papers.
Papers are especially welcome in the following areas:
Prospective authors should submit three copies of FULL papers by Friday 25 September 1987 to the Conference Secretary, at the address below.
Papers will be reviewed by a Programme Committee drawn from the organisations cooperating in the Conference. Authors may be asked to modify their papers in the light of the referees' comments.
The Ken Brodlie Prize (£50) will be awarded to the author(s) of the best submitted paper based on technical merit and presentation.
Suitably modified papers, selected for the Conference, will be published in Computer Graphics Forum, the journal of the EUROGRAPHICS Association.
For any further information about the Conference, please contact the Conference Secretary:
Dr Adrian ThomasThe Alvey Vision Club Committee (Chairman M B Brown, British Aerospace) are arranging a third technical meeting at Cambridge University, from 15 to 17 September 1987, to discuss research topics and share experiences of computer vision and image interpretation.
The program will contain reports of progress in research from Alvey Vision consortia, several invited papers from experts in the field, and some papers on vision related topics from other authors.
The final program is not yet known, but it is likely to be similar in content to last year's meeting, where papers were presented on the 2.5D sketch, object models their matching and geometry, structure from motion, problems in the understanding of stereo vision, stochastic models for boundary analysis, image descriptions and segmentation, and novel hardware for image processing.
The rapid development of solid state devices for image capture is enabling applications of image processing to spread quickly and cheaply into many new areas of medicine, science and engineering. We hope to learn more of them at our Conference in September.
For further information contact the Alvey Vision Club Secretary: Kate Crennell, Technology Division, RAL