The gViz Project Research Overview Review of the Project - Impact, Relevance to Beneficiaries and Futures

The project has succeeded in its aim of developing visualization middleware for e-science.  Software generated by the project is being incorporated in production versions of both IRIS Explorer and pV3, and other software will be made available to the community as open source.  The research begun in the project is being further developed in a number of e-science projects and industrial applications.  International recognition of the project has come through presentation of a project paper at the premier visualization conference, IEEE Visualization 2004 in Austin, Texas – the only UK paper to be accepted this year and the first paper in that forum to address issues of Grid-based visualization.

The key objectives of the original proposal, bar one, were successfully achieved.  Robust additions to IRIS Explorer now allow it to be used in a Grid environment, and will be key to release 5.4 of the product from NAG; together with the outreach activity, this delivers the objective of work package 1.  Work package 2 has delivered a Web service version of pV3, already being used within Rolls-Royce, and this is being pursued through a proposal to the DTI Inter-Enterprise Computing initiative.  Work package 4 delivered on its aim of studying XML for visualization: the skML language was conceived, designed and implemented – and is being used in a number of new e-science projects, including Open Overlays (a project in the EPSRC Fundamental Computer Science for e-Science programme) and EPSRC Advanced Environments for Enabling Visual Supercomputing.  Progress was also made in managing different data formats.

Two unexpected pieces of research emerged during the project:  work package 1 identified limitations in computational steering when simulation code is embedded in a visualization system, and this prompted a major activity – the development of the gViz library to link separate visualization and simulation software.  This has proved to have broad utility and is now being used in the Integrative Biology pilot project.  Work package 4 identified the need for a visualization ontology and work has begun at an international level to pursue this further.

Work package 3, on geometry compression, an independent piece of work to be contributed entirely by Streamline Computing Ltd, was not achieved.  This was for reasons outside the control of the investigators of the project, and simply reflected the difficulty of Streamline in recruiting suitable staff and changing commercial priorities for the company.  However Michael Rudgyard of Streamline Computing engaged fully with the overall direction of the project through participation in meetings and workshops, and influenced the direction of the research described in work packages 2 and 4. Indeed, as a result of this project, a proposal has been submitted to DTI under the Intra-Enterprise Computing initiative for follow-on work to develop, amongst other things, a commercialised version of pV3 using Web services.  In addition, Streamline has directly exploited some of the lessons learned in this project through the exploitation of SOAP and Web services technology in their new generation of software development tools, the Distributed Debugging Tool (DDT) and the Optimisation and Profiling Tool (OPT). These will be amongst the first commercial software products that use GRID technology to emerge from the UK. Streamline are now actively investing in R&D for geometry compression as 1) they perceive the market for this technology to be more mature, 2) their existing product base has gained a level of maturity and 3) they are able to map resources to this project.    

Dissemination took place through journal and conference papers and posters, with full details on the project website (www.visualization.leeds.ac.uk/gViz). In addition to the IEEE Visualization paper mentioned above, it is worth noting the tutorial presented at Supercomputing 2003, the State of the Art presentation on Distributed Collaborative Visualization at Eurographics 2003 (also published in a journal), four All Hands papers and many live demonstrations.  In all, 16 refereed publications (papers, posters, tutorial notes) have resulted from the project.

The work has proved highly relevant to the intended beneficiaries: the middleware becomes available to the general community through software suppliers and open source; the heart modelling community in particular are appreciative of the work, and are beginning to exploit it in their research.

Finally the project enjoyed a good spirit of collaboration throughout.  Although each work package had a leader, all academic partners contributed constructively across the full programme of work.  (The academic leads for Grid-enabled IRIS Explorer and the gViz library were Leeds; for pV3, Oxford; for conceptual models and skML, Oxford Brookes; for data transfer, CCLRC; and for ontologies, Oxford Brookes and Leeds.) The industrial partners likewise all contributed in a significant manner, and we have been grateful for their active participation and encouragement.