Contact us Heritage collections Image license terms
HOME ACL ACD C&A INF SE ENG Alvey Transputers Literature
Further reading □ Overview □ 1987 □ 12345 □ 1988 □ 6789 □ 1989 □ 101111a121314151617 □ 1990 □ 181920212223242526272829 □ 1991 □ 303132333435 □ 1992 □ 363738394041 □ 1993 □ 424344454647 □ 1994 □ 484950515253 □ 1995 □ 545556575859 □ 1996 □ 60
CCD CISD Harwell Archives Contact us Heritage archives Image license terms

Search

   
InformaticsLiteratureNewslettersECN
InformaticsLiteratureNewslettersECN
ACL ACD C&A INF CCD CISD Archives
Further reading

Overview
1987
12345
1988
6789
1989
101111a121314151617
1990
181920212223242526272829
1991
303132333435
1992
363738394041
1993
424344454647
1994
484950515253
1995
545556575859
1996
60

Engineering Computing Newsletter: Issue 31,

March 1991

Editorial

Your newsletter contains items on a wide range of topics in this edition. Education and Awareness events will continue to be a significant feature of the EASE Programme - see below. Further activities are planned for the Summer and Autumn and full details will be published either in this newsletter or the Graphics newsletter.

New readers are always welcome! If you have not seen this newsletter before and wish to receive one, please let me know. New contributors are also always welcome!.

Sheila Davidson

EASE Programme Cut by 80%

The Engineering Applications Support Environment (EASE) Programme is the major casualty of the well publicised SERC financial problems. A proposed cut of 80% from 1992/93 onwards was approved by the Engineering Board at its meeting in February. This represents a reduction from an over £3M programme in the current year with 52 staff at RAL to a level of £750K per annum and 10 staff by April 1992.

Such a large and rapid reduction in the programme will have a drastic effect on Informatics Department at RAL. It is possible that the reduction may have to be made earlier than this.

The purpose of this insert in the Engineering Computing Newsletter (ECN) is to appraise you of the situation and how we believe it will impact the engineering community.

Plans have been drawn up on the basis of an orderly rundown of staff during the next year.

The following components of the current EASE Programme will not continue after 31 March 1992 (or earlier where stated):

Other activities may or may not be continued, depending on how the Computing Facilities Committee decides to assign the limited available resources. It is hoped that the next Graphics Newsletter will report on these decisions.

Bob Hopgood, Informatics Department

Future EASE Events

In spite of the cutbacks in some areas of the EASE Programme a series of workshops, tutorials and seminars has been planned for the coming months. Some details are given below:

Engineering Design and Modelling

23 April 1991, UMIST

The EASE Mission Statement is: To stimulate and encourage engineers to use appropriate state-of-the-art software and hardware to enhance their research. This seminar will focus upon engineering design with two presentations from the Engineering Design Research Centre in Glasgow. There will be talks tying design to modelling and examining the software development process. The day concludes with updates from the survey of UK Structural Mechanics Finite Element Usage and Needs and a presentation from Intra University Steering Committee's CADCAM Working Party.

Introduction to Using the X-Window System

24 April 1991, RAL

This tutorial aims to give the non-expert an overview of the X Window System, its capabilities and drawbacks, and where to go for further information. Participants will be assumed to have had some exposure to a window system, though not necessarily to have programmed in such an environment.

The Programmer's Hierarchical Interactive Graphics System (PHIGS)

8-9 May 1991, RAL

PHIGS was published as an ISO/IEC standard for computer graphics in 1989. It attempts to combine a modelling system and a viewing system in a single standard. Implementations of PHIGS are appearing on the market, particularly on workstations (Sun, Dec, IBM, HP, etc).

This tutorial, spread over eight hours of lectures, gives an introduction to the ISO/IEC standard for computer graphics, will cover the essential features of PHIGS, and will conclude with a review of the limitations of PHIGS and the steps taken by ISO to extend PHIGS. The final lecture will also cover the status of other projects in ISO.

he tutorial is aimed at applications programmers and users of graphics systems. Some familiarity with GKS would be an advantage but is not essential. Both Bob Hopgood and David Duce have been BSI delegates to the ISO Standardisation meetings that developed PHIGS. They are currently editors of the revision for GKS and represent BSI at Computer Graphics Reference Model standardisation activities.

Fortran to C Conversion Course

17-21 June 1991, RAL

A five-day, intensive, hands-on programming course for Fortran programmers wishing to convert to the programming language C, covers the basic features of C, using examples from Fortran to ease the conversion process. The presenter, Dr S K Robinson, Head of Software Engineering Group, Informatics Department at RAL has acted as a consultant in regard to the development of C compilers. He has recently become a member of the BSI/S Programming Languages Committee and has already presented this course to industry and academics several times already.

Computational Structural Mechanics Finite Element Codes

UK Intra University Steering Committee Survey (IUSC) of Usage and Needs in HEIs

The purpose of this Survey, which has been funded by the Computer Board of the Department of Education and Science, is to assess the current usage of and future UK needs for Computational Structural Mechanics (CSM) Finite Element (FE) codes in the academic community.

The Survey form was distributed in Summer 1990 using EASE/IUSC contacts and the National Agency for FE Methods and Standards (NAFEMS) BENCHMark distribution. 157 replies from staff and research grades in Higher Education Institutes (HEIs) have been received - between 10% and 20% of the active post-graduate (CSM) FE user community. The full implications of the results for this community will be discussed by the rusc and the Computer Board in early 1991 and a final report on the Survey made available in late Spring to the developers as well as the community. Some important results and implications can already be seen.

47% of the community choose US commercial codes, 44% choose UK commercial codes and 9% use in-house codes. The users' first choices are dominated by three codes: PAFEC (26% of all users), ABAQUS (24%) and ANSYS (11 %). LUSAS is a first choice of 4% of all users followed closely by a range of codes with 3% or less. There are no mainland European first choices.

On closer study, the success of PAFEC (a UK code) and ABAQUS (a US code) is due to wide availability and meeting complementary market niches.

PAFEC is available at academic rates through a centrally negotiated HEI deal. It is usually accessed in a PAFEC pre/post environment (pIGS) and supported by PAFEC. The code is used for research, teaching and student projects. The access made available to a lot of the Source Code is a positive feature used by research groups (eg to model creep damage mechanics at Nottingham University). It is, however, very limited in non-linear facilities - being unable to deal with more than one non-linear feature (eg geometric and material) at a time.

ABAQUS is not yet available through a central HEI deal but can be accessed at the National Centres or by direct licensing at academic rates with the developer or UK agent. It complements the quasi-linear PAFEC capabilities for the community and dominates usage by research groups dealing in highly non-linear problems. Only minimal support is guaranteed and there is little indication of its use in teaching or project work at undergraduate level.

Users were invited to criticise the codes as they stand and to itemise their wish list. The trends are summarised briefly here. Main criticisms centre on the infra-structure of codes rather than on functionality. Key infrastructure targets are inadequate documentation and support, poor pre/post processors, difficulties with error estimation, unavailability on PC or supercomputers, high costs of training and lack of access to source versions. The key lack of functionality was the failure of many of the codes to model the very wide range of materials (particularly composites) cited by users. The user 'wish lists' tended to seek correction of these failings.

There is a need to make PAFEC, ABAQUS and the other major codes more widely available in the HEIs, to extend code support and to encourage more commercial contacts between HEI groups and developers particularly in the fields of benchmarking, documentation, training and algorithm development.

R G Anderson, Manchester Computing Centre

Testing Software with EASE

During the Summer of 1990, using funds awarded by the Computing Facilities Committee, Napier Polytechnic of Edinburgh undertook the survey and evaluation of quality assurance software and tools suitable for engineering research.

The results of this activity will shortly be available as an EASE Technical Report.

Many of you have taken part in this survey on Validation, Verification and Test (VV&T) of software, one of the findings of which is that there is a need within the research community for a software testing centre.

As many software projects within the community are of a limited duration and lifespan, the feeling is that acquiring and using VV&T tools imposes an unacceptable level of overhead in terms of cost and time. A software testing centre can offer the following facilities with cost spread among the users within the research community.

I am interested in the views of the EASE community or expression of interest. Dependent upon the level of interest expressed, it is proposed to hold a workshop on this subject where hands-on use of the tools will be available.

Teo Goh, Napier Polytechnic

Open Meeting of CTACC

The Real Time Control Transputer Applications Community Club (CTACC) held its Second Annual Open Meeting on 10 December 1990.

The meeting was organised by Dr David Holding, Dr Geoffrey Carpenter and Mrs Brenda Phillips at Aston University. It was chaired by Professor George Irwin from the Queen's University of Belfast. Fifty-two people registered in advance but extremely bad weather reduced the number attending to thirty-three.

Neural Systems Engineering

Professor Igor Aleksander of Imperial College gave the guest lecture on neural systems engineering. The talk included an overview of neural systems engineering, a description of activities at Imperial College and a look at future directions.

Current implementations of neural networks use silicon chips while alternative photographic and holographic strategies also exist. Four pillars of neural computing can be identified: single layer perceptrons, multi-layer perceptrons, auto-associators and unsupervised nets. Neural systems can be classified at three levels equivalent to a single perceptron, a very small area of the brain and a complete neural basis for cognition. Working systems at the third level will incorporate natural language and will allow soft man-machine interfaces.

Work at Imperial College has concentrated on RAM neurons which incorporate probability in place of a weight based approach. One system is used by the Home Office for identifying intruders and another by British Telecom to identify parts of the face.

It will be possible in future to develop a generalised neural unit, using auto-associators with variable degrees of connectivity, which matches biological systems well. A possible UK plan has the following application goals:

Real Time Control of a Hexapodal Mobile Robot

Professor John Gray, Dr David Barnes and Mr Clive Downes of Salford University reported on a mobile robot demonstrator developed with the help of a Computing Facilities Committee (CFC) grant obtained through CTACC.

The robot design is that of a mechanical insect measuring approximately 30 em by 15 em. It contains three identical linked segments, each controlled by a transputer and possessing two legs, together with a head possessing feelers and infra-red sensors. The body segments can undertake limited operations independently. The robot's task is to track and follow in real time the motion of a moving object over rough terrain within a laboratory environment.

Transputers are used to control concurrently the robot's 12 degrees of freedom, to process internal and external event sensors concurrently and for inter-segment communication. They satisfy the important engineering constraints of size, weight, cost and self-containment. The control software has been implemented in occam with the separate sensor-driven activities of stance, gait, avoidance and tracking programmed in parallel.

CTACC Chairman's Report

Professor George Irwin identified some of the highlights of CTACC's first 17 months of operation.

CFC funding for developing demonstrator projects was the most significant item. Two six month projects have been approved: one of these was the hexapodal mobile robot. The original two man years funding has now been halved because of SERC's financial problems.

The committee had a strong involvement in the TA90 conference: eliciting, encouraging and refereeing papers, recommending how they should be organised into a programme and chairing sessions. CTACC cosponsored an IEE colloquium which attracted much interest from industry and academia.

The CTACC newsletter has run to four issues and is a successful activity. The e-mail broadcast facility has been less widely used but delegates agreed that the intended purpose of the broadcast facility - to help members seek advice and share useful information is worthwhile.

An activity and interest directory and a bibliography on transputers in control are being compiled.

Discussion of Future Plans for CTACC

The following points were discussed and agreed:

Work at Aston University

Winding up a most interesting day, Dr David Holding spoke about transputer based work at Aston University and led a tour of the laboratories. The Aston group investigates reliable, safe and secure systems in telecommunications, signal processing and distributed control using formal methods for design. A prototype packaging sub-system, consisting of an arbor drum and a slider each controlled independently by transputers, was demonstrated. Similar systems in industry use mechanical linkage for control but a transputer approach will lead to increased speed and flexibility.

Raymond Fawcett, RAL

EASE Technical Reports

Production of the technical reports resulting from various Engineering Applications Support Environment (EASE) funded activities has commenced and these are being distributed to the academic community.

It has been agreed that copies of all reports will be sent to the Directors of Higher Education Institute computer centres, EASE HEI site representatives, and members of SERC committees etc. In addition single copies can be made available on request by members of the academic community free, but any requests for multiple copies will be charged at a rate of £5 per copy.

Copies of all reports can also be made available to non-academic and industrial organisations at a cost of £50 including VAT. The current list of reports includes:

ETR 1/90
An Evaluation of Three User Interface Management Systems
ETR 2/90
Assessment of Superworkstations
ETR 3/90
EASE Workstation Assessment, June-August 1990
ETR 4/91
Sun Workstations: System Administrator's Guide
ETR 5/91
Sun Workstations: An Introduction to SUNOS
ETR 6/91
Sun Workstations: A Guide for New Users

Reports on other topics are also in process of preparation.

For information, ETR 4/91 is meant to be a simple introduction to system administration for novice administrators responsible for tasks such as the installation and maintenance of system software, and the management of filestore, hardware devices, networking services, etc. It also covers some UK-specific issues which are not covered in the Sun manuals, particularly in the field of networking. ETR 5/91 was written for users with little or no experience of SUNOS, Sun's implementation of the UNIX operating system. ETR 6/91 is primarily intended for readers who have not used a Sun workstation before. It offers a general introduction to using a workstation and describes the Sunview graphical user interface to introduce the basic concepts involved in using a typical window system. It also indicates some of the software available to EASE Sun workstation users.

Geoff Lambert Rutherford Appleton Laboratory

SENAC - Computer Algebra Driver for NAG

The Symbolic Environment for Numerical and Algebraic Computation (SENAC) is a Computer Algebra package with the ability to solve resulting numerical problems using the Numerical Algorithms Group (NAG) Fortran and Graphics libraries.

The package, which is built using the Logic Programming language LISP, gives the ability to make complicated calls to the NAG libraries without creating a lengthy Fortran program.

When NAG library calls are made SENAC generates, compiles and links Fortran code to use the routine selected. The ability to convert user defined functions prepared in symbolic form into Fortran is available and when the NAG library routine requires derivatives SENAC both automatically forms symbolic derivatives and codes them into Fortran. Since the numeric computation performed by SENAC uses optimized Fortran code large numerical problems can be performed far faster than is possible with other Computer Algebra packages.

SENAC has its own data structures that are used both to pass numerical input to the NAG routine being called and also to capture the resulting output, which can subsequently be processed with further calls to library routines. A simplification for the NAG user is also obtained owing to SENAC storing sets of environment variables with default values for many of the input para-meters to NAG routines. These parameters are documented comprehensively in SENAC's User Manuals along with their default values, which can be overridden when required before a NAG call.

The SENAC package is modular being split into three separate modules, which are Sencore the interactive symbolic host language, Numlink the symbolic/numeric interface to the NAG Fortran library, and Graflink the symbolic/numeric interface to the NAG Graphics library. There is also a fourth module which can be used instead of Numlink, known as Wslink, that allows a symbolic/numeric interface to the NAG workstation library.

Sencore

A computer algebra language with the ability to call NAG Fortran and Graphics library routines. Sencore is an interactive module, with an easy to use help system and a comprehensive User Manual, it is used to input and manipulate symbolic expressions and perform symbolic differentiation. Sencore is also responsible for handling the numeric data structures, which can be input as list, matrix and sparse matrix data types. Sencore also has the ability to translate user defined functions into Fortran programs to be used by SENAC for calculations or for external use.

Numlink

Numlink provides the interface between Sencore and the NAG Fortran library. Its ability to automatically generate error free Fortran code in itself can save valuable time especially with the automatic formation of derivatives of supplied user functions as well as the automatic coding of both user functions and derivatives into Fortran. The processing of input data contained in SENAC data structures and capturing output from a NAG library call into a SENAC data structure is also automatically performed by Numlink. Apart from passing the names of data structures through to NAG, Numlink is also responsible for another two types of parameters. Problem parameters, which need to be defined by the user such as input function name and endpoints of function domain and algorithm tuning or analysis parameters which are all given preset default values, defined in the Numlink Manual, and can be overridden before calling a NAG routine if necessary.

Graflink

Graflink is the interface module between Sencore and the Graphics library. This module is modelled on Numlink with user defined functions being automatically translated into Fortran and algorithm tuning or analysis parameters being preset to default values. Calls to plot routines are significantly simplified.

The University of London Computer Centre (ULCC) holds the European distributorship for SENAC.

David Winstanley & Minaz Punjani, University of London Computer Centre

Club Mmed

Modelling and Management of Engineering Data Community Club (Club Mmed)

On Wednesday 19 December 1990, Club Mmed held a tutorial entitled, An Introduction to Information Modelling. Forty-one people from industry and higher education attended the meeting which was held at UMIST. The aim of the day was to introduce delegates to some of the concepts and techniques used in modelling of engineering concerns.

The morning sessions were given by representatives from Leeds University: Dr Susan Bloor set the scene by first defining some basic concepts, then giving an overview of the special characteristics of engineering data, focusing on the great complexity involved and the different views of the data that must be maintained.

Brad Harris talked about entity-relationship-attribute (ERA) modelling and normalisation, stated some guiding principles, and showed how these are realised in a high-level graphical modelling language (IDEFIX).

Stuart Roberts gave a description of three established database paradigms (hierarchical, network and relational), and indicated their inadequacy for capturing semantics and behaviour. After describing the features of a Semantic Data Model he concluded that although leading to intelligent databases and so highly desirable, there is at present no definitive Semantic Data Model and the field continues to evolve.

After lunch, Tony McClelland (Strathclyde University) gave a detailed description of the graphical modelling language NIAM (Nijssen Information Analysis Method), which he has used extensively in his work modelling ship structures.

he rest of the afternoon was concerned with the textual information modelling language Express, part of the proposed STEP standard. Two talks and a 'hands on' session were presided over by Hilary Kahn and Alan Williams of Manchester University.

Hilary Kahn gave a thorough review of Express constructs, relating these to the ideas presented earlier in the day. Express can describe (part of) the real world in terms of Schemas, which consist of a number of typed items, entities and their attributes, and rules governing the state of the information model.

Alan Williams presented the use of Express at Manchester in a variety of applications, concerned mainly with the electronics industry. Express is used because it is precise, provides documentation for review and criticism, and removes unnecessary implementation considerations. He . described his way of starting to write an Express model, emphasising the use of a divide and conquer approach and the need for comprehensive commenting.

The tutorial ended with a lively session of hands-on Express. Delegates were given a modelling problem to solve in groups, and asked to present their solutions to the meeting.

Mike Mead, RAL

Use of Superworkstations

One objective of the EASE programme has been to make available top end computing kit to the Higher Educational Establishments. This mechanism allows academics to sample technology that might otherwise be unavailable.

One such machine was the Ardent (now Stardent) Titan, the first of a new generation of high performance 3D graphics computer without a specialised 3D graphics transformation/clipping/shading pipeline. The Titan is a symmetric multiprocessor computing system capable of supporting up to four CPUs. Each processor contains an integer unit, separate instruction and data caches, and a pipelined scalar and vector floating point unit. Each processor, on the evaluation machine, is rated at 10 Mips and from 1-2 Mflops, depending on the sequence of operations. The graphics subsystem provides double buffered 24 bits per pixel with a 16- bit Z-buffer and four overlay planes.

ABACUS, the Architecture and Building Aids Computer Unit of the Univerisity of Strathclyde, made a successful bid to RAL for the loan of the Titan for a six-month evaluation period. ABACUS's interest in the Titan was stimulated by two factors, its vector processing capability and the DORE graphics environment. These facilities matched the needs of two of our applications, a Digital Illumination Model (DIM) which is computationally intensive, and ROVE (Real Time Object Visualisation Environment), a 3D animation package which currently relies on a geometry engine and makes heavy use of a machine's native graphics. The object of this exercise were twofold; firstly, to ascertain if the Titan's vector processing power could be usefully harnessed to resolve the computational overheads of the radiosity based lighting model, and secondly to explore the novel graphics subsystem.

The Titan's processing performance comes from its use of vector operations as opposed to the more conventional scalar instructions. A scalar variable is a single data value in a program, most computers perform operations combining a few scalar values to produce a new scalar value. A vector unit treats a whole sequence of scalar values as one data unit, a vector, and can perform a single operation on all the vector's elements at once. For this reason those parts of the code that can be vectorised run up to ten times faster than their scalar counterparts.

On the graphics front, most advanced graphics workstations provide additional hardware to boost the performance of 3D graphics, (eg the Silicon Graphics geometry engine), the Titan relies on its main CPU and the efficiency of vector processing. To complement this architecture Ardent developed a high level 3D graphics environment for the Titan called the Dynamic Object Rendering Environment (Dore). Dore was an object orientated environment providing all the primitives, attributes and hierarchical management of current graphics libraries.

The outcome of this loan will be a report assessing the computational and graphical performance of the Titan. This has been achieved by way of experience gained through writing software, with similar functionality to that currently implemented on conventional machine architectures. The objective has been to make full use, where possible, of the vector processing capabilities and to compare and contrast the relative merits of the Titan graphics subsystem with the alternative fixed function graphics pipeline hardware.

Mike Grant, University of Strathclyde

Nominations for Top IT Award

The British Computer Society (BCS) announces the opening of nominations for its 1991 Awards.

There are three premier Information Technology (IT) awards given by the BCS each year. Ten outstanding IT projects are chosen as finalists from the nominees. These go forward to the final and three are chosen as overall winners by a panel of judges drawn from industry and academia, chaired by H R H The Duke of Kent.

In making the choice of a winner, the Judging Panel considers not only technical achievement but also the contribution made towards extending the understanding and acceptance of computers by society generally, of the effects of the work itself on society and the extent to which the application of that work would assist in improving the overall quality of life taking into account the ways in which computing is itself changing.

There have been three winning categories: Applications, Social Benefit and Technical. Examples of winners:

From 1991 these distinct categories are being dropped but there will still be three Awards which are now open to the widest possible range of IT projects.

The BCS Awards are now in their 19th year and since the initial launch, each year's winners have gained much prestige as a result of their success. Winners also find that the Awards are an ideal platform for the promotion of their work in IT.

The final date for nominations is 25 May 1991. Those who know IT projects which could be nominated or want more information should contact me

Frances Evans, The British Computer Society

AI for Engineers

EASE and the SERC Financial Crisis (and AIAI)

The announcement on the front of the January 1991 issue of the newsletter about support to Engineering Applications Support Environment (EASE) events at the Artificial Intelligence Applications Institute (AIAI) has caused a lot of worry, and many people have contacted us asking if they can still come on our courses. The short answer is yes.

In the past, EASE covered all the costs associated with an approved SERC attendee. What they are now unable to cover is the cost associated with a person actually attending a course, ie the training manuals, coffee, etc. The major costs relating to course preparation and presentation continue to be supported by EASE. So we will now have to charge all students a small registration fee which will vary according to the particular course:

1991 Course Schedule

KBS Skills, 8-12 April

Fundamentals of knowledge representation are explained and the programming paradigms practised. Highly practical based course for students to gain experience of both rule based systems and object oriented systems.

Planning and Scheduling, 14-17 May

Explains and demonstrates the advanced AI techniques involved in knowledge based planning and scheduling systems. It describes the benefits and limitations of existing systems, enabling students to assess the applicability of planning and scheduling to their areas of interest.

Reasoning with Constraints, 5-7 June

Explains how constraint logic programming can be used to solve problems such as scheduling and timetabling. By focusing on specific applications, it assists students to evaluate these techniques for use within their own applications.

Common Lisp, 29 April-3 May

Teaches the core parts of Common Lisp, enabling students to understand and apply a representative subset of the language to a range of programming problems. Aspects of the Common Lisp Object System (CLOS) are also described.

Introduction to Prolog, 16-19 April

A popular course for beginners giving them a sound grounding in elementary Prolog. Specifically developed as an introductory course to the language to encourage non-programmers to make a start.

Advanced Prolog, 7-10 May

An intensive, practical course intended for students with some experience of Prolog but who wish to extend their knowledge. Discusses the use of Prolog for building expert systems and introduces constraint logic programming.

Terri Lydiard, AIAI

Key Enablers for Advanced IT in the 1990s

8-12 April 1991, Manchester

The Post Experience Vocation Education (PEVE) Unit is now offering a new five day hands-on course in key enabling technologies for advanced IT in the 1990s. The technologies covered are Formal Methods, Object-Oriented Design and the Application of Parallel Computer Architectures. The course is supported by the DTI and is aimed at technical management in IT companies.

Course Objectives

After completion of the course delegates should understand:

Martyn Spinks, University of Manchester

Forthcoming Events

⇑ Top of page
© Chilton Computing and UKRI Science and Technology Facilities Council webmaster@chilton-computing.org.uk
Our thanks to UKRI Science and Technology Facilities Council for hosting this site