The fourth Information Technology Awareness in Engineering Workshop, organised as part of the EASE Programme, takes place at Clyde Hall, University of Strathclyde on 30-31 March 1995. The topic was selected from a number of possible topics to reflect its importance at this time, particularly as the new Innovative Manufacturing Initiative (IMI) is currently formulating its research programme under the various sectors. Representatives from all 6 IMI Sectors will be participating in the Workshop and there will be ample opportunity to interact with them during the two days.
The Workshop programme will provide delegates with an up-to-date view of Object Technology from both the practical applications in engineering and the IT viewpoints. Speakers, from academia and industry, who have already exploited this technology, will give an account of their experiences and explain how it has benefited them in their particular field. Other speakers will cover the computer science aspects and, in one case, give a view of the future of this technology. This part of the programme will ensure potential engineering researchers understand the potential of Object Technology and how such systems can be used properly and appropriately.
Speakers will include:
The remainder of the programme concentrates on the IMI, with presentations from each of the Sector Managers or their deputies, followed by a break-out into a number of sector-based parallel streams where the discussion will focus on how Object Technology can impact the research programme of an individual sector. In these discussions it will be important to identify any specific actions which need to be taken to make the technology more acceptable to a sector. This will enable the Sector Manager to assess whether or not there is a requirement for a particular research focus to be followed to ensure such actions are carried out. This means that delegates at this Workshop will have a real opportunity to influence the research strategy of the various IMI Sectors.
Speakers for IMI will be:
All speakers, apart from those presenting the details of the IMI Sectors, will be providing a paper which will form part of the Workshop Proceedings and will be available to delegates when they register.
ECN readers should have already received their personal copy of the workshop programme and registration form. If you have not received this and would like one please contact me and I will arrange to send one to you by return.
The latest ITE Seminar was held at University of Salford on 26-27 January 1995. It provided a good overview of the achievements to date, application potential and future possibilities of some key IT technologies. As we have come to expect from this series the arrangements were smooth and efficient, with the local attraction this time being a meal at the Yang Sing Restaurant. This article is a set of opinions rather than a full report as the Virtual Reality (VR) scene has already been set in a detailed article by Bob Stone (Intelligent Systems Solutions Limited) in ECN issue 53 November 1994 and copies of the seminar proceedings are available from RAL.
In a way that seems characteristic of many IT technologies there is much hype and hard-sell attached to VR and associated techniques, particularly from North America. The Far East provides a contrast: notably the Japanese seem to be carefully exploring the use of VR as just another tool available to computer specialist and engineers - a tool that offers most potential when it is integrated with other techniques and is suited to the application domain. The take-up of expert systems/knowledge-based systems is perhaps a warning from which we can learn? In a well balanced seminar programme, the speakers explained carefully what we can do (and, perhaps more importantly, cannot do!) at present with VR, where are the most likely breakthrough points for commercial application and some intriguing glimpses of the future. There was some difference of opinion as to the futures being dreams of paradise (individually tailored of course) or inescapable nightmares.
The talks were supported by a good selection of demonstrations and the opportunity to visit Intelligent Systems Solutions Limited. Strong messages came over from more than one speaker about the importance of matching the technology to the application, delivering a timely and functioning solution, and understanding the contributions that the underpinning disciplines can offer. Surely we cannot rely on a pure technology push, based upon slightly flaky hardware and software? It is a mistake to try and re-invent wheels: for example, we can use existing databases but front-end them with usable tools; we can learn a lot from film and television (especially in an animation context) as to conventions, codes and symbols - the environments with which people feel comfortable. The second day of the seminar had parallel streams following on from presentations from the EPSRC/BBSRC/ESRC Innovative Manufacturing Initiative (IMI). Reports back from the parallel sessions indicated that those present at the seminar were sure that there was great potential for VR in the three current sectors of IMI: aerospace, construction and responsive processing. Rapid prototyping and visualization seemed to fall by the wayside: they may not be so glamorous as VR, but are more mature and thus more readily applicable technologies. VR in IMI might be taken forward as components of sector projects or as one of the generic technologies that the Initiative will support.
There was also a parallel session on human factors aspects of VR. At present this seems to be largely unknown territory although we have some good hints, almost a mapping out of the explorations that are needed, from disciplines such as ergonomics. As was brought out in the October 1994 EPSRC workshop we need to build upon existing knowledge and it seems that many of the critical issues for VR centre on soft science problems of health and safety, usability and psycho-social issues. VR is too important to be left as a solitary vice for computer nerds. This mix of disciplines with an interest in applicability in the broad sense is one area where the UK may have a lead: perhaps this is where we should be funding some forward-looking R&D to build our future industries and enhance our competitiveness?
There was a very strong response to the first call for proposals in the three initial sectors (Aerospace, Construction and Responsive Processing) of the IMI. There were over three hundred outline proposals submitted, representing projects with a total value of some £175M with industrial partners prepared to pay around half the cost of projects overall. The outline proposals are being considered by the appropriate Sector Programme Managers and their Sector Target Advisory Groups to agree which proposals should be invited to proceed to full proposal stage. Applicants will be informed of the decision on their outline proposals shortly after 20 February 1995.
At its January 1995 meeting, the IMI Management Committee approved the Road Vehicles aspect of the Land Transport Sector. The focus for this area was provided by a report from Dr Peter Davies, Director of the Advanced Technology Centre, University of Warwick. Copies of the report have been circulated to those on the IMI Land Transport mailing list. There is a Community meeting for this sector on Thursday 9 March 1995 in London. For further information contact Vince Foley.
For general information on IMI please contact the IMI Support Group (EPSRC Swindon Office).
Our Word 6.0 courses have been extended to three levels:
A variety of courses on Unix, C and Fortran 90 have already been scheduled in the diary. We will also arrange an Introduction to Fortran 77 and CMS to VMS conversion courses if they are requested.
We have a rolling programme of PC courses on various topics: DOS, Windows, Word for Windows, Access Database, Excel etc. Please contact the courses administrator for details.
We are willing to arrange courses on other topics should there be sufficient demand.
With effect from 1 April 1995, all participants of all courses will be charged, to reflect the actual cost of running the course. We have been able to hold the cost of PC courses without change, and although the costs of more technical courses have risen, they are still excellent value. Courses will be run only if there is sufficient demand to make them viable. The cost of courses is currently £120 per student per day for PC Courses and Fortran 90. The cost for other programming courses is £180 per student per day.
If you have been trying to telnet over to CERN to retrieve their WWW software you may be interested to know that a few months ago they separated their WWW and telnet servers. This was due to the phenomenal interest in WWW, and the consequent heavy load on the machine.
The new addresses for WWW and telnet are: http://info.cern.ch/hypertext/WWW/TheProject.html and telnet.w3.org.
It was announced in December 1994 that W3C (the WWW Consortium) is to be split into two co-operating and coordinating branches. One branch will be known as MIT-W3C and be based at MIT in the States; the other branch will be known as Euro-W3C and be based at CERN. The final details of membership to Euro-W3C have yet to be finalised.
The first Chief Executive of the Council for the Central Laboratory of the Research Councils (CCLRC) is to be Dr Paul Williams, announced David Hunt, Cabinet Minister for Science, on 31 January 1995.
Dr Williams is the current Director of the Daresbury and Rutherford Appleton Laboratories. He will then take up his appointment on 1 April 1995 when, subject to the approval of Parliament, the new Council will be established. The post of Chief Executive was filled by open competition, following the announcement by Dr Hunt in October last year.
Welcoming the appointment, Mr Hunt said:
I am confident that establishing the new Council will create the right environment for this major resource of UK Science and Engineering base to flourish. It has an important part to play in carrying forward our White Paper objectives of wealth creation and improving the quality of life.
I am delighted that Dr Williams will be the first Chief Executive of the new Council. His academic knowledge, managerial skill, and high international standing, together with his intimate knowledge of the UK Science and Engineering scene make him admirably suited to lead the Council. He inherits two excellent laboratories, with a skilled and dedicated team of scientists and engineers. I am sure they will all play their part in making the new body a success.
Dr Williams said:
I am looking forward to the challenge of leading these laboratories into an important new phase in their development. As components of the new Council, their world-leading capabilities will be more accessible to all the UK's Research Councils, and to UK industry. I am confident that the Council is well equipped to play its part in responding to the aims of the White Paper Realising our Potential.
Dr Williams is a physicist by background and has had a wide and varied career, in the UK and overseas, as both a research scientist and also as a manager of significant science and engineering facilities. He is a fellow of the Institute of Physics and is currently the Director of the Daresbury and Rutherford Appleton Laboratories.
On 12 January 1995 DRAL hosted, on behalf of the Department of Trade and Industry (DTI) and the European Commission (EC) DG-III, a conference on the first IT Call in Framework IV. The conference was in fact a video conference, supported by UKERNA using SUPERJANET and so allowed participation from DL (Daresbury Laboratory) as well as RAL. The event was organised by the Information Dissemination Group of the Computing and Information Systems Department.
David Hartley of UKERNA gave a welcome address, explaining the technology. He was followed by Dr Gordon Walker, the Head of RAL, who welcomed the participants to DRAL and introduced Dr Ian Eddison of DTI who chaired the event. The EC presentations, from Rosalie Zobel on Multimedia and Jean Francois Omnes on HPCN (High Performance Computing Networking) were very informative. Margaret Dennis of the DTI Esprit Unit gave valuable advice on the procedures.
Questions from the audience and from others were first sifted by a group of three experts (ably supported by Julian Thompson from the DTI Esprit Unit), and then answered by the panel of speakers. The answers revealed subtle policy issues in the EC and were a help to potential FPIV participants. The general feeling of the audience in both physical locations was that the conference was a success, both from the point of view of useful information and the utilisation of video conferencing technology.
The DTI was sufficiently pleased with the success of this event that it is currently considering running a similar video conference event at RAL in June to coincide with the Third FPIV Call. Full details of this will be announced in a future ECN and elsewhere as soon as they are known.
One of the requirements of participating in the Commission-funded programmes taking place under the Fourth Framework Programme is the need to form a consortium for joint research with organisations from different European countries. In order to help organisations identify potential partners, the RTD-Partners team is now focusing on the Fourth Framework Programme.
The RTD-Partners database is provided by the Commission as part of CORDIS (Community Research and Development Information Service). Although it can be used for other types of partnerships, one of its main functions is to help organisations identify partners for participation in Commission-funded programmes. A major advantage of the database is that it covers all types of organisations and all programmes and sectors of research. This allows an organisation to search for partners in information technologies, telecommunications, materials and manufacturing, energy, environment, medicine, quality of life, biotechnology, agriculture or marine technology.
The RTD-Partners database can help you identify suitable partners for these programmes in two ways. By placing an entry on the database, your details and research interests will be available to the wide range of organisations which access the database on a regular basis. Alternatively you can access the database yourself to identify partners with complementary interests. To place an entry on RTD-Partners, it is necessary to complete an entry form. An electronic version of the form can be obtained by sending a message to the e-mail address: cordis-cp@lcd.co.uk (there is no need to include any text in the mail message as the system will respond automatically).
The completed form can then be returned to CORDIS by e-mail for rapid inclusion in the RTD-Partners database. A paper version of the form can be obtained by sending a fax to the address given below. This form, which has been specially designed for ease of completion, allows organisations to describe themselves and specify in detail what collaboration they are seeking. This includes which programme(s) they are interested in, details of the specific project areas and coded keywords relating to their research interest.
The RTD-Partners form also allows organisations to specify the type(s) of partner they are seeking. This includes the prospective partner's type and size of organisation, geographic location, and nature of its activities. The form also asks for full contact details including phone and fax number so that interested users can contact the relevant person directly.
Once the RTD-Partners form has been sent to CORDIS, it will be processed rapidly and the information included in the RID-Partners database. Your details will then be available, using both free-text and coded searching, to the large number of organisations who currently search RTD-Partners on their own or on their client's behalf.
Many organisations have already found the RTD-Partners to be an extremely useful tool for finding partners, particularly for calls for proposals. According to feedback, a significant number of organisations have already formed some kind of partnership as a result of their entries on the database.
At present there are over 15,500 entries on RTD-Partners. Many of these records have been received recently and relate specifically to the programmes taking place under the Fourth Framework Programme. The database is currently accessed for more than 500 hours a month. Placing an entry and accessing the database are free of charge.
Further information on RTD-Partners, including entry forms, can be obtained from me.
Turbulence is the key to predictive realism of CFD for most practical flows in Mechanical, Aerospace, Hydraulic and Process Engineering. It is also the most challenging phenomenon to represent in CFD, for it is as ill understood as it is important, and requires, for virtually all practical applications, closure models which are amalgams of rational principles, mathematical concepts and intuition.
This linked sequence of two courses on turbulence and its modelling cover a major proportion of past and present approaches to the representation of turbulence effects within numerical algorithms for the prediction of flow and transport phenomena.
The courses are independent and self-contained, the first addressing basic issues and modelling practices currently being employed in engineering CFD, while the second deals with advanced concepts which are still subjects of research, but are beginning to enter CFD practice.
This article is the second half of a discussion on C and Parallelism and focuses on the many proposed extensions to C++, to give it a parallel semantic. The original article appeared in ECN 54. It is noteworthy that virtually all the proposed language extensions have been to C++ rather than C. In part this can be explained politically: C++ is perceived to be the more modem language; and in part technically: the class mechanism of C++ can be hi-jacked to good effect.
It would be impossible within the confines of this article to present a survey of the astounding number and range of available systems, but it is hoped that at least a flavour will be communicated. A categorisation scheme will be introduced and three illustrative systems described.
I only investigated systems for which related papers were easily available on the Internet. The total number identified in the search was around 30. Some systems provide informative WWW resources. Most of the systems arc public domain, and network accessible via ftp.
There are more of these systems than there are technical approaches, which is another way of saying that a rationalisation is required. Certainly, for the programmer the diversity is hardly conducive to the longevity of application software, even when the daunting choice has been braved. Though standardisation is what is required, run-time efficiency is the problem. No single approach can be guaranteed good performance across the range of applications. Also, different and valid programming models arc provided, and it is premature to lose this diversity.
Parallelism may be introduced by a number of different mechanisms:
This article examines Language Extensions to stay focused. Extensions account for the majority of the systems and are of greatest relevance to the application programmer, entailing novel paradigms for parallel programming. Class libraries can be understood with a grasp of C++, and adequate system documentation. Parallelising systems are mostly programmer-transparent. Note that message passing systems are not addressed here as these tend to assume a more language independent form, and do not reveal the special characteristics of C++.
The categorisations (and representative systems) are shown in the accompanying diagram. The major distinction between C++ extensions is whether the application consists of a number of different threads with communication and/or synchronisation (code parallel, eg CC++) or whether one thread of control is multiply instanced when operating upon aggregate data structures (data parallel, eg C**, Mentat, pC++).
C++ offers greater opportunity for data parallel extensions. In contrast, classic parallel programming involving semaphores, monitors and messages is code parallel.
Within data parallel systems, the parallelism can be explicit (more usual eg C**) or implicit as with Mentat.
One can further divide the explicit data parallelism - some support the geometric decomposition of data that is generally possible with scientific and engineering calculations. pC++, which comes from Indiana University, is an example of such a system. However, as pC++ is so heavily based on High Performance FORTRAN (HPF), it will not be addressed further in this article.
The par construct concurrently executes proc1 and proc2, and terminates when both components have terminated.
par {
proc1();
proc2() ;
}
Iteration is more natural than recursion to support arbitrary levels of concurrency, so there is a parfor construct.
parfor (int i=O; i < N; i++ )
{
p1(i) ;
}
The termination criteria for par and parfor are a hindrance for some applications so spawn executes an arbitrary CC++ expression in a new thread of control; while the call to spawn itself terminates immediately, eg:
spawn x + y + g(z)
Two new declaration qualifiers (mutable and sync) are introduced:
const int x;
// cannot be assigned
sync int x;
// assignable once
mutable int x;
// assignable many times
// (default for int)
A process which reads a sync variable will be suspended until initialisation has occurred. A class member function can be declared as atomic. Only one atomic function per class instance may executed at a time. This is a mechanism for enforcing mutual exclusion in executing certain sections of code and is a higher level way of presenting monitors, used in traditional parallel programming.
A special aggregate class is introduced, which consists of an ordered collection of values (or aggregate elements). For example:
class matrix {float v;)[] [];
new matrix [100] [100];
Note that a matrix is an object containing a 2D collection of floating point values NOT a 2D array of objects.
Aggregate member functions are applied to the entire collection of elements, but member functions declared as parallel allow elements to be acted upon concurrently (unlike objects in an array).
The power of C** can now be revealed. The following parallel friend function performs the concurrent updating of matrix elements (each with co-ordinates (#1, # 2 )) . Note that parallel functions require a parallel argument.
friend void stencil
(parallel matrix A) parallel
{
A[#1] [#2].v =
( A[#1-l] [#2].v
+ A[#1+l] [#2].v
+ A[#1] [#2-1].v
+ A[#1] [#2+1].v) / 4.0 ;
}
New values are entirely a function of the old matrix. As with a Single Instruction Multiple Data (SIMD) parallel machine, the operation can be viewed as entirely simultaneous. Since parallel invocations cannot communicate, a typical programming style is to repeatedly invoke parallel operations.
Parallel values can be returned as in the following code for matrix multiplication which returns a matrix, although at the code level a scalar (row #1 * column #2) is returned.
friend matrix operator*
(
parallel matrix A,
matrix B
) parallel
{return (A[#1] []*B[] [#2]);}
With C** there is no worry about the interaction of many threads of control, and the parallelism is easily identified. The race free properties of SIMD are present, without the limitations of such hardware.
The granule of parallelism is the Mentat class instance. The keyword MENTAT precedes the associated class declaration. Member function invocations automatically proceed concurrently whenever data dependencies permit. The programmer is not concerned with scheduling decisions, communications or synchronisations. For example:
x = matrix_op.mpy(B,C);
A = matrix_op.mpy
(
x,
matrix_op.mpy(D,E)
) ;
Despite the code, (B*C) and (D*E) will be implemented in parallel.
There are restrictions: Mentat objects have disjoint address spaces and Mentat class member functions use call by value semantics.
The caller of a Mentat object function does not block waiting for the result. Dataflow analysis allows continuation until the result is needed. If the return value is not used locally then dataflow is transferred elsewhere.
The return to future function rtf() is analogous to a C++ return() within a Mentat class member function but there are differences:
For example:
rtf(DM.read(record));
The member function read need only have been executed if (and when) the returned value is actually used. The burden of parallelism is shared. Although the compiler only performs a limited form of automatic parallelisation, the programmer as a result has less to do. Is this merely disguised (and discredited) dataflow or a firm step towards eventual complete automation of parallelisation?
The three systems described represent much of the variety in approach. They attempt to capture parallelism controllably within a few simple constructs echoing those of C++. Whilst the elegance of the best of these mechanisms is undeniable, one might question the value of the overall exercise. On a practical level, are the mechanisms general purpose enough and can they yield the desired performance?
If you are interested in following up anything from this article contact david@inf.rl.ac.uk or why not access the internet using the following addresses?
The Transputer Consortium (TTC) is pleased to announce that the World Transputer Congress 1995 (WTC '95) will be held on 4-6 September 1995 at the Harrogate International Centre, North Yorkshire, England. WTC '95 is the leading international transputer conference and exhibition and is the third in a series sponsored by and run under the overall management of TTC.
WTC '95 is also sponsored by the Commission of European Union.
The Conference will be held over 3 days, 4-6 September. The format will be similar to that in previous years with both Plenary and Parallel Sessions. Confirmed speakers include Rosalie Zobel (European Commission), Eugen Schenfield (NEC Research Institute, US), Mike Delves (NA Software, UK) and John Elliot (Smith System Engineering Ltd, UK). Submitted papers, accepted following the refereeing process, will be presented in Parallel Sessions.
An associated exhibition attracting the world's leading suppliers of transputer-based and other relevant hardware, software and application products will be held in the Exhibition Hall. Opportunities also exist for posters and demonstrations of academic achievements.
The tutorials will be held on 2-3 September 1995 in the Royal Baths Assembly Rooms and will cover the fundamental principles underlying transputer technologies, the design paradigms for exploiting them, and workshops that will focus directly on a range of specialist themes (eg real time issues and image processing).
There will be a Welcome Reception on Sunday 3 September. The Conference Banquet will be held on Monday 4 September, with the Exhibitors' Reception held on Tuesday, both based on a Yorkshire theme.
The PPECC Steering Group met on 12 December 1994 to review the progress of the club in its first year and plan the schedules of events for the forthcoming year.
The meeting welcomed two new members into the Steering Group: Dr M Sabin and Dr G S Cooper (University of Salford). These new members further enhance the expertise for the Group. Further expansion of the Group is anticipated within the next year when it is hoped that more Industrial members may be added to the Group.
A review of the first series of events of the Club concluded that our current range of seminars, courses and workshops were providing a focus for engineers interested in parallel processing.
The Group discussed repeating the newly established PPECC courses in the second year and considered further developments to these events.
Seminar topics suggested by members were discussed and a provisional programme of courses and seminars laid out. A future edition of the PPECC mailshot was discussed and members' views and articles for inclusion are welcome.
The Steering Group still welcomes input and suggestions from members on all aspects of the PPECC events and activities.
The British Computer Society (BCS) has a Special Interest Group (SIG) focused on parallel processing (PPSG).
his Group has a membership of roughly 150 parallel processing specialists/companies within the UK. The group organises open meetings on related topics throughout the year and produces a magazine Parallel Update.
For details on the Groups activities contact the PPECC Secretary who maintains a collection of membership application forms.
An electronic mailing-list has been created for the PPECC at the Newcastle Mailbase facility. This provides a discussion point for members and information source on Parallel events and courses.
The list is currently used as a primary electronic source for information, announcements and registration forms on the PPECC activities.
The address of the list is ppecC@mailbase.ac.uk.
The mail list is open for membership and engineers wishing to join simply need to send an e-mail message to the address mailbase@mailbase.ac.uk containing join ppecc name surname where the <name> and <surname> fields contain your information.
A number of interesting responses were received taking issue with the ECN 54 article on C and Parallelism. The following code fragment was intended to show a difference between C and FORTRAN. In C it was claimed the value of i is predictable (ie 6) but in FORTRAN it is not.
integer i ... i = 5 ff(i,3) i = i + 1
Two valid counter-examples in C were received.
extern int *p; p = &i; ff(i,3)
In the code body of function ff, the contents of address p (and hence i) may be changed
#define ff(o,t) o = o + t
If ff is a macro defined function then local arguments seemingly passed by value may indeed be changed.
Arguably the macro function mechanism is a facility of the C preprocessor rather than the language itself. However, in practice the use of the two is inextricable. The pre-processor is used to give more power to C. It is interesting to note that some languages intrinsically do not need recourse to this mechanism.
The point of the article is that unless control is explicitly relinquished by the C programmer, then the values of variables are predicable. Putting the address of a local variable in a global address pointer is rather a wild (though permissible) action.
Many thanks for these inputs from keen eyed readers of the ECN that helped me refine the point.
The Academic View
This is a comprehensive introduction to computational fluid dynamics, a hands-on five day residential course held in the delightful setting of The Cosener's House, Abingdon, Oxfordshire.
The content covered a broad range of CFD topics, applied particularly to industrial, aerodynamic and environmental flows. This included, for example, basic discretization schemes applied to model equations including the finite volume method, the design of high resolution shock capturing schemes, implicit time-marching methods, models of turbulence, the use of pressure correction methods, combustion modelling, grid generation techniques and the impact of developments in computer hardware.
The breadth of the course was matched by the panache of the presenters, Derek Causon (Manchester Metropolitan University), Steve Fiddes (Bristol University), Bassam Younis (City University, London) and Bill Jones (Imperial College) who, at the very last moment, was asked to fill in for Jim McGuirk (Loughborough University), absent because of family bereavement. Besides tea breaks, the talks were regularly interspersed with practical classes designed to drive home the key points of the lectures. What I gained personally was some insight on current CFD trends over a broad interdisciplinary range of topics - ideas that I could possibly use in my own area of interest on environmental free-surface flows - and also in making several useful contacts. What was also useful was a 5 cm thick wad of lecture notes and comprehensive lists of landmark references in CFD. If there were criticisms of the course then I think these could only be levelled at the quality of some of the interactive computer graphical displays (which, for instance, plotted only in black and white and only one dependent variable, when there were several of interest) and of one piece of FORTRAN source code which was poorly written. There were, of course, all-singing, all-dancing (as the developers may have us believe!) commercial software such as FLUENT, PHOENICS and FLOW3D, made available for use during the practicals.
Another feature of the course that I enjoyed was a series of research lectures given by the presenters on their areas of research interest and which culminated with a comment by Derek Causon that doing CFD at the research level is difficult. It would be much easier to take up knitting or crocheting. This left me with a ponderous image of Professor Causon had he decided to take up the latter!
In summary, I thought the course was well organised and well worth attending not just by novices but by intermediates as well. The food and catering was excellent with a nice touch provided by the final dinner: a three course meal with a variety of wines. Perhaps all that was lacking was some live chamber music!
The Industrial View
I believe that the course Introduction to CFD is very useful for my work. It covered all relevant items of physical modelling and numerics in CFD. Although it was an introduction to CFD, this intensive course explained in detail why and how model assumptions are made and which are the numerical techniques used to solve the founding systems of nonlinear partial equations.
Even though I had already been applying CFD for years, it was very useful for me to get an overview of the most relevant CFD concepts, their history and their theoretical implications. The lectures were well structured and the way they were presented made the subjects interesting and understandable. They also mentioned practical CFD applications from different areas as well as research topics. The practicals helped to understand some of the key concepts and encouraged discussions about CFD.
This article describes two recent events in which the Visualization Community Club has been involved.
In January, AGOCG (the Advisory Group on Computer Graphics) with support from the Community Club, ran a postgraduate course on Graphics and Tools: Techniques and Tools. Lecturers were from the Universities of Bath, Leeds and Manchester (where the event was held) and technical support was provided by BBSRC, RAL and University of Manchester.
The 23 postgraduates, who have a practical need for graphics and visualization in their work, spent a week learning about presenting data, use of colour, principles of 3D and visualization. They also learnt Uniras and AVS.
50% of the time was spent in practical sessions, using a teaching set of Silicon Graphics Indigo workstations in the Department of Chemistry. In addition to laboratory sessions, the students were set (and they completed) a 2D presentation challenge and a 3D flow visualization challenge. For the latter, each student group was required to complete a video sequence which showed what they had managed to learn from the data.
We hope the students have learnt some techniques and tools that they can apply in their work and meanwhile the lecturers and organisers have to decide whether to hold a 4th such event in 1996.
In November 1994, the Community Club held an event at the University of Leeds on the Visualization of Experimental Data.
The event covered a wide range of applications which included chemical reaction dynamics, particle physics, protein structure, medicine, combustion and the photoelastic measurement of stress.
The event highlighted many of the problems of experimental data: noisy data (most speakers); voluminous data; highly stochastic, multidimensional data (particle physics). Among the techniques described were principal components (conventionally used in remote sensing, but here applied to medicine); the use of configurable 2D icons to visualize multiple parameter data; and the use of multiquadric functions for stochastic data. As may be expected there was wide use of image processing.
This event made me more aware of the variety of problems arising from the need to analyse experimental data. New solutions continue to be needed and no single software product dominated. However, configurable software such as Khoros and AVS allowed the researchers to experiment readily with different methods of analysis.
For the first time in a Visualization Community Club event, there was a call for abstracts and this enabled postgraduates from within the UK research community to present their work to others. It seems to me that this is a valuable function of a Community Club and would be worth encouraging in future events.