UK AVS+UNIRAS User Group Annual Meeting 1997

The University of Birmingham

17-18 November 1997

Final Programme

Monday 17th November
10:00 Coffee and Registration
11:00
Visualizing Computational Electromagnetics Data
John Whittle, British Aerospace PLC
11:45
AVS Futures
Mark Mason, AVS (UK) Ltd
12:30 Lunch
14:00 Session A:
The Gsharp Application Seat for Windows
John Stringer, AVS Helpdesk
Mapping the variation of Radon in soil
Margaret Oliver, University of Birmingham
15:00 Annual General Meeting of the UK AVS+UNIRAS User Group
At which the UNIRAS UK User Group and the UK AVS User Group will be formally merged into one.
15:30 Tea
16:00 Session B:
International AVS Centre
Helen Morphet
Module Porting
Paul Lever
19:30 Reception
20:00 Dinner
Tuesday 18th November
09:00 Session C:
An AVS case study
AVS-UK staff
Development of a dynamic visualisation tool for heat and fluid flow analysis
Dr Shian Gao, University of Leicester
A Gsharp case study
AVS-UK staff
Gsharp - discussion session
Lead by John Stringer, AVS Helpdesk
10:30 Coffee
11:00 Session D:
Visualization of Complex Microstructures
James Schumm, University of Manchester
MAVIS/Express: A framework for Molecular Visualization within AVS/Express
Matt Cooper, University of Manchester
Visualising the Phase Separation of a Binary Fluid
Simon Jury, University of Edinburgh
12:30 Lunch
14:00 "THIN" vs. "FAT" Visualisation Clients
Dr Mikael Jern, AVS/UNIRAS Denmark
15:00 Tea
15:30 Close

We will not be running formal parallel sessions this year.
We do, however, wish to encourage informal discussion groups in the adjacent lobby area.


Abstracts

Visualizing Computational Electromagnetics Data
John Whittle, British Aerospace PLC

The paradigms for visualizing CFD are mature and intuitive, and the power of modern workstations has allowed CFD codes to migrate from supercomputers to the desktop. In contrast, Computational Electromagnetics (CEM) is still a supercomputer centred activity, and the development of meaningful visualization techniques has been hampered by the sheer size of the datasets involved. The reasons behind these problems will be discussed, and solutions centred on BAe's involvement in the ESPRIT project INDEX presented.

"THIN" vs. "FAT" Visualisation Clients
Dr Mikael Jern, AVS/UNIRAS Denmark

The widespread popularity of Web technology has created a new information visualization technology model, in which browsers enable the widespread distribution of information using standard HTML, GIF and VRML techniques. The client is effectively reduced to a browser (viewer) of information supported by a server. A true Web client is not capable of program execution unless the executables are downloaded to the client as Java applets. This client is normally referred to as the "thin" client. A thin client, by definition, have minimal software requirements necessary to function as a user interface front-end for a Web enabled application.

The most appealing aspect of the "thin" visualization client to information visualization users is that the overall cost of software and maintenance can be dramatically reduced. The "thin" client allows the application developers to eliminate the notion of software distribution at the client level (no license issue!), eliminate the notion of maintaining local software and supporting multiple operating systems on remote clients.

The concept of a "thin" client, however, raises the issue of client vs. server data visualization rendering. Real-time visual data manipulation doesn't translate well into a "thin" client. While the VRML file format allows distribution of visualization scenes to the Web, the user has no interactive control of the actual underlying data structure. The "mapping" of numerical data into geometry format (VRML) takes place at the server side. The client "3D browser" can only navigate in the 3D world.

Local data manipulation, information drill-down technique, context sensitive menus, object picking and other interactive user interface functions that traditionally have been available on the client are now controlled by the visualization server. In the "thin" client model, nearly all functionality is delivered from the server side of the visualization engine while the client perform very simple display and querying functions.

Web components and Plug-ins are now being used to overcome some of the limitations. Java allows the creation of components "applets" or "JavaBeans", which are automatically downloaded and executed on the local client. These components can significantly increase the data interaction between the client application and user, and allow tasks to be executed on the client.

These Java applets that deliver locally available executables are, however, still dependent on the network bandwidth. Depending on the scope and application, Java applets and its data sets must be downloaded. Java applets are only resident during execution and are therefore removed from the local disk after the completion of the task. As the demand for larger applets and data sets grows, significant download time could be incurred and the network becomes the bottleneck. Keeping commonly used components resident on the client would significant reduce download time, although this practice is counter to the Java applet architecture.

A "Fat" or "Intelligent" visualization client provides local functionality through Plug-ins or locally stored Web components (ActiveX). Visual data manipulation is provided at the client side through these components. Highly interactive user interface tasks are delivered that provide point-and-click navigation through multidimensional data structures. Visual data interfaces such as information drilling, moving a cutting plane through a volume data set etc can be supported. Clearly, a full-featured visual data manipulation has many advantages over the rudimentary offerings of Java applets and HTML query forms. This presentation will explain the concepts and demonstrate several examples of Thin vs. Fat clients in visualisation.

MAVIS/Express: A framework for Molecular Visualization within AVS/Express
Matt Cooper, University of Manchester

This talk describes ongoing work to convert the AVS5-based Molecular visualization application, MAVIS, to AVS/Express. The goal of the work is to produce a suite of modules which, in addition to forming an application with similar functionality within this new visualization environment, will also form a framework within which it is made far easier for the user to add modules making use of MAVIS' internal data structures to add functionality appropriate to the specific work in Computational and Structural Chemistry which they wish to undertake.

Development of a dynamic visualisation tool for heat and fluid flow analysis
Dr Shian Gao, University of Leicester

Unsteady heat and fluid flow simulations often generate a huge amount of data on flow structures and their developments with time, and it is difficult to get the most out of these valuable data and to understand the dynamic nature of the flow using traditional static graphics representation alone. In this project a user friendly software visualisation tool FLOWVIEW has been developed that is capable of constructing 2D images and then creating a movie with the specific attributes set by users. The tool was developed with FORTRAN 90 and UNIRAS subroutines were used extensively to perform necessary data processing and to produce all the basic graphics plotting. The user interface was developed using USEIT with top and side menus and a text dialogue window.

The tool has been applied to analyse different data sets produced from large eddy simulations (LES) of turbulent flow and heat transfer, including transient thermal impinging jets and near wake developments behind a flat plate. It is found that FLOWVIEW, both flexible and easy to use, provides a valuable tool for creating dynamic movie images that help users understand complex heat and fluid flow processes. The development and evaluation details of the tool will be reported in the full paper.

Mapping the variation of Radon in soil
Margaret Oliver, University of Birmingham

At the Uniras conference in 1995 I spoke about mapping the risk of childhood cancer in the West Midlands Health Authority. One offshoot of that work has been an investigation to examine putative links between some kinds of malignancy and radon. Radon, a radioactive gas, escapes naturally from rocks and soil containing Uranium or Thorium. Increasing concern about possible links between emissions of radon and certain types of malignant disease has led to a need to know how radon emission varies from place to place in the soil and in dwellings. The spatial scale at which radon varies is largely unknown, and so efficient sampling schemes cannot be selected to estimate radon precisely.

Two geologically and pedologically distinct parts of the English Midlands were examined. An area of 225km2 was surveyed near to Hereford using an unbalanced multistage scheme with seven stages. The distances between sampling points ranged from 10m to 7.5km. The results showed that more than 90% of the variation occurred within distances of 10m. The other area to the South of Buxton was underlain by three different lithologies. An area of 56km2 was sampled at intervals from 1m to 3.75km. There the variation within the shortest spacing was also substantial, but a fairly large proportion of the variance also occurred over distances of 60 to 240m. These results suggested that geology exerts a strong control on the variation, but that there are other controlling factors. This led to more detailed surveys on one of these lithologies to gain insight into the possible causes of the variation. The kriged map of estimated radon concentrations in the soil showed a striking inverse relation with the relative elevation of the land. Further research has shown that this appears to be associated with the depth of the soil.

These results have implications for designing further surveys to measure radon in the soil, for choosing a method of mapping the gas, and for understanding its spatial variation. These will be described and discussed.

Visualization of Complex Microstructures
James Schumm, University of Manchester

Volume datasets, such as dental biofilms or ice crystals, containing many distinct objects, require special treatment if they are to be effectively visualized. This talk discusses techniques that may be useful in visualizing such data, such as object selection, clustering, and multi-resolution representation.


28th October 1997 - © UK AVS+Uniras User Group and Malcolm Austen, Oxford University Computing Services