General Monthly Meeting
Visualization of Galaxies, Oceans and Brains
Bernard A. Pailthorpe, VisLab and the School of Physics, University of Sydney,
Date: Wednesday, 5th September, 2001
Time: 6:00 for 6:30 pm
Venue: Search & Discover Room, Australian Museum,
Collins St., Sydney (William St. entrance)
The frontiers of Scientific Visualization now include problems arising with data that scale in size or complexity. New metaphors may be needed to navigate, analyze and display the data emerging from bio-diversity, genomic and socio-economic studies. This talk addresses the challenges in generating algorithms and software libraries that are suitable for the large scale data emerging from tera-scale simulations and instruments. With larger and more complex data sets, moving into the 100 gigabyte to 1 terabyte realm, scalable methodologies and tools are required. The collaborative efforts to address these challenges, currently underway at the San Diego Supercomputer Center and within the National Partnership for Advanced Computational Infrastructure (NPACI), will be summarized.
The ultimate aim of this R&D program is to facilitate queries and analysis of multiple, large data sets derived from motivating applications in astrophysics, planetary-scale oceanographic simulations and human brain mapping. Research challenges in such science application domains provide the justification for developing such tools. Previously planetary-scale oceanographic simulations had resolutions limited to 2 degrees of latitude and longitude. With Teraflop computing resources coming on line, such simulations will be conducted at 10 times (and presently 100 times) resolution, soon yielding multiple sets of a hundred gigabytes numerical output. In mapping the human brain, up to four distinct imaging modalities are used, with data sets already at tens of gigabytes. The immediate research challenge is to composite these images, facilitating simultaneous analysis of structural and functional information. These applications manifest the need for high capacity computer displays, moving beyond the usual 1 megapixel desktops to 10 megapixels and more. Developments in this area will be discussed.
Professor Pailthorpe founded Sydney VisLab, with Australian Research Council funding, to support computational and visualization research. That lab underpins research and teaching innovations across a broad spectrum of disciplines. Its recent work has provided visualization support for high-resolution weather modeling, specifically in the context of the Sydney 2000 Olympics. Recently he directed the scientific visualization program for NPACI and for SDSC, at UCSD. The group efforts there are focused on scalable volume visualization.
Previously, he established and lead a research effort (at the University of Sydney) in computational physics, specifically simulating advanced materials by classical and quantum molecular dynamics. This led to an understanding of the mechanism of diamond formation in carbon thin films. He has wide experience in physics education, including developing new classes in Computational Physics. He has advised Government at senior levels on HPC, including the NSF in the USA, the Australian Prime Minister's Science & Engineering Council (resulting in a new funding program to establish the Australian Partnership for Advanced Computing in 1998) and the New South Wales State Premier (also resulting in a new HPC funding program in 1998).
Report on the General Monthly Meeting
by Matthew Wilmot
Professor Bernard Pailthorpe gave an intensely interesting lecture on the myriad of uses and applications of supercomputers in the world today. He began by highlighting the problem of increasing data size. We were told of the enormous amounts of data that are collected continuously via a huge variety of sources. Satellites rain data down onto the earth every day, and there is a very real problem of how to analyze and how to store these enormous data sets.
The transportability of this data was also highlighted, as the Internet makes it increasingly available to access devices such as the Hubble Space Telescope remotely from all over the world. Within this context, the problems of both interfacing and the transmitting of data appear.
Local examples of these enormous data sets were discussed. These included high resolution weather predictions for the 2001 Sydney Olympic Games. These were in turn used to schedule the games and provide advice to services. Further examples provided by Dr Pailthorpe included ultrasonic images of arteries, visualizations of radar arrays, three- dimensional models of the economy, and schematics of the ozone layer. These were all presented to us in glorious colour projection.
We then delved further into the analysis of these giant data sets as Dr Pailthorpe pointed out that movement is also to be analyzed. The scale of these data sets was indicated when Dr Pailthorpe spoke of the 100 million dollar supercomputers the US currently has, and of the fact that there are data sets which essentially cannot fit into computers such as these.
We were then treated to some of Dr Pailthorpe's own work, in the form of a spectacular movie – a three dimensional rendering of the Orion Nebula. This was created for Hayden Planetarium's VR Dome, situated in the US. Dr Pailthorpe explained how the movie we were shown needed to be created on a teraflop machine – a machine whose power was equivalent to one million laptops. We then explored some of the applications of this high powered computing technology. These included meteorological applications, in the form of studying energy transfer between the ocean and the atmosphere. This involved global analysis and mapping of the world's very large array of currents, and the enormous wavelengths of movement which occur across the world's oceans. Some current medical applications were also mentioned, including the mapping of the brain for study, and to assist in the development of surgical procedures. We were also informed of a recent development by General Electric in the US: a machine which can CAT scan and measure MRIs simultaneously. This machine develops very large file sizes, and these need to be analyzed and stored somehow. Even more specifically, Dr Pailthorpe drew attention to the work being done to analyze and visualize complex metabolic pathways and the binding of molecular switches to DNA.
Finally, we were given a tantalizing glimpse of emerging imaging technologies which are appearing as a result of the proliferation of gigantic, high-resolution data being collated. These fascinating technologies are working to ensure that the way in which we can view complex images and data arrays will improve far beyond that which we can currently utilize. Dr Pailthorpe presented illustrations of giant 10 metre screens created by tiling projectors together. Beyond this, we saw how there are many new experimental physics techniques being used to develop these emerging screen technologies. These studies are resulting in technologies such as electronic paper and organic LEDs that can be unrolled onto a wall like wallpaper.
In his final remarks, Dr Pailthorpe touched upon just one of the effects that supercomputing technology will have upon the Internet. Essentially, we will realize real time, simultaneous, multi-cast meeting through the Internet, enhanced by file sharing and three-dimensional images.
Encompassing a wide array of scientific fields, Dr Pailthorpe delivered a talk which was timely, and very current. As computing technology pervades so many areas of our day-to day-lives, it was exciting to hear what is occurring at the fore-front of this field.