Message from the Chair of the Board

Innovation2: SHARCNET and HPC host inaugaural forum

Safeguarding your water

Executive Director's Message

A simple competitive fact: Interview with SHARCNET SD

Notice of SHARCNET Annual General Meeting 2004

Second to None:
SHARCNET receives $20M, to become one of Canada's most powerful HPC centres

As the result of a grant from the Canada Foundation for Innovation (CFI), SHARCNET may soon be one of Canada’s most powerful High Performance Computing (HPC) institutes.

In March, SHARCNET was awarded a total of $19.3M from the CFI, 100% of its funding request and the largest allocation nation-wide. SHARCNET has applied to the Ontario Innovation Trust (OIT) for funding to match this award. A positive decision from the OIT, in addition to vendor and institutional contributions, will bring the breadth of the SHARCNET expansion to just under $50M.

With this new equipment fully deployed, SHARCNET will be Canada’s largest HPC institute, with the greatest number of partners, and some of the nation’s most powerful computing facilities.

The CFI is an independent corporation established by the Government of Canada in 1997 to strengthen the capacity of Canadian universities, colleges, research hospitals, and other non-profit research organizations to carry out world-class research and technology development.

SHARCNET currently helps to accelerate the research result of some of Canada’s pre-eminent academics; from increasing the understanding of outbreaks of diseases such as SARS and Foot and Mouth to the development of new models to manage financial risk.

Hugh Couchman, SHARCNET Scientific Director believes that SHARCNET’s strength lies not only in its leading-edge equipment, but also in its globally-competitive research community.

“Through the provision of world-class computational infrastructure, made possible through agencies like the CFI, SHARCNET enables research and innovation which would otherwise not be possible,” confirms Couchman. “Our research community is second to none, and their support was central to this latest award.”

With this funding, SHARCNET will expand its community by more than two-fold, welcoming the Universities of Waterloo, Brock, York and Ontario Institute of Technology to the consortium. It will expand it computational infrastructure to meet the growing demand for HPC in both established and emerging fields, from Astrophysics and Chemistry to Economics and Financial Mathematics.

SHARCNET will also focus its attention on the development of innovative solutions in storage and visualization over a distributed network. SHARCNET plans to link its 11 partner sites via 1 Gb/s and 10 Gb/s connections, through its dedicated connection to the ORION network, which currently connects clusters at Western, Guelph and McMaster.
State-of-the-art visualization systems and specialized facilities for grid research, computational chemistry and interconnect topology will also become part of SHARCNET’s leading infrastructure.

Bioinformatics is just one of the burgeoning fields that will be supported by this new phase of SHARCNET. The ability to analyze exponentially increasing data with more and more sophisticated computational methods is expected to enable dramatic breakthroughs in health and medicine.

Dr. Ming Li, a Professor and Chair of Bioinformatics at the University of Waterloo, is extremely enthusiastic about these new capabilities. Li’s work on genome-genome comparison and protein 3D structure prediction will be greatly enhanced with the use of SHARCNET.

“SHARCNET provides the immense computing power necessary for these intensive calculations,” says Li. “HPC has and will continue to enable impressive advances in Bioinformatics and many other areas.”

Couchman agrees. “This new infrastructure will allow our researchers to compete at the highest levels. The potential to ‘compute tomorrow’s solutions’ has never been more attainable.”

HOME

Microscopic research, major impact
BY ANTHEA ROWE

Microscopic transistors mean microscopic devices. And microscopic devices have the potential to revolutionize everyday life as we currently know it.

SHARCNET Chair Erik Sorensen’s research stems from his fascination with materials and their different states of matter. His work in Condensed Matter theory seeks to understand what phases occur in systems and why materials change from one state to another (e.g. from liquid to solid or gas). With his numerical experiments at extremely low temperatures, he foresees the potential for discovering previously-unknown states.

As Sorensen puts it, the discovery of “new and exciting phases [and their respective] properties” would essentially represent the discovery of new materials previously unknown to the modern world. New materials, and their unique properties, would have significant applications to any number of industries, including computers, electronics, aviation, aerospace, automobiles, and more.

Originally from Denmark, Sorensen has earned academic degrees and completed postdoctoral work at a variety of institutions, including the University of Aarhus, Denmark; the University of California, Santa Cruz; Indiana University, Bloomington; and the University of British Columbia. In 2001 Sorensen left a professorship at the Université Paul Sabatier in France to accept positions as an Associate Professor in McMaster University’s Department of Physics and Astronomy and SHARCNET Chair in Computational Materials

Sorensen’s research could have implications for spintronics, the study of electron spin. Spin properties could be exploited to produce infinitesimally small electronic devices. Sorensen remarks that “the most exciting aspect of spintronics is the fact that microscopic transistors built with such technology would dissipate much less heat than ordinary transistors. Hence, one could pack orders more transistors together,” thereby creating incredibly powerful – and incredibly tiny – devices.

Using SHARCNET’s high performance computing facilities, Sorensen models the electrons, bosons, and quantum spins that describe the materials he is investigating. He has discovered that, rather than behaving independently, the electrons, bosons, and quantum spins often interact with each other and thereby correlate their behaviour: “These strong correlation effects can give rise to many extraordinary effects and completely new phases at low temperatures.”

Sorensen acknowledges that his research would not be possible without SHARCNET support.

“What attracted me [to SHARCNET] was the unique possibility of a professor position at a major university combined with access to state-of-the-art facilities and personnel. You might find the facilities elsewhere, but what I particularly appreciate is the cross-disciplinary research environment SHARCNET has helped to foster and the excellent system administrators, programmers, and administrative staff that SHARCNET has attracted. The fact that we now have a HPC community within SHARCNET is extremely valuable.”

HOME

Synchrony and SARS
McMaster researcher uses mathematical models to predict – and help control – epidemics.
BY ANTHEA ROWE

David Earn, Associate Professor of Mathematics at McMaster University and CIHR New Investigator, works in the field of Mathematical Biology. Using SHARCNET he develops and analyzes mathematical models of biological systems, primarily for applications in epidemiology, ecology, and animal behaviour, with impressive results.

One of the main themes of Earn’s research is the understanding and controlling of extinctions. In his work on endangered species, Earn uncovered a relatively simple but nevertheless significant concept: synchrony promotes extinction. Earn analyzed “conservation corridors” – the connective lands that allow animals of a given species to move between habitation areas – and found that “corridors have the potential to synchronize population fluctuations, and consequently to cause local extinctions to happen at the same time everywhere.” He recommends that, before developing the corridors that are intended to help protect at-risk species, conservation bodies take his research results into consideration.

Earn applied this discovery to his research on epidemiology. He created and then implemented mathematical models that simulate the transmission of infectious diseases such as measles. His research produced groundbreaking results in 2000 when, as Earn says, “We discovered that it’s possible to predict changes in the frequency of measles epidemics from birth rates and vaccination levels.” And prediction is simply the precursor to prevention.

Earn’s work on the prediction of measles outbreaks led him to the development, using SHARCNET, of computational methods for the design of more effective vaccination strategies. The strategies, derived from Earn’s modeling of synchronicity, call for coordinated mass vaccinations of entire populations, a proposal in direct contrast to the current practice of inoculating individuals at a given age (thirteen months, in the case of measles).

Having established himself as a mathematician who knows his way around infectious disease prediction, Earn was called upon by various academic and governmental bodies to study the 2003 outbreak of SARS. Following collaborations with Health Canada, the Ontario Ministry of Health, and individual physicians at the heart of the outbreak, Earn is now working with a research team on a study of the diagnosis and epidemiology of SARS. Headed by McMaster’s Professor Mark Loeb of Pathology and Nuclear Medicine, the project represents a $1.7 million dollar research project funded by the Canadian Institutes for Health Research (CIHR).

HOME

Parallel Priorities
SHARCNET offers graduate course at University of Guelph
BY ANTHEA ROWE

Parallel programming cannot be learned in a day. Such was the maxim that educator and HPC Consultant David McCaughan was invoking when he decided to create a full-fledged university course on the topic.

Parallel programming is premised on parallelism: a strategy for performing large, complex tasks quickly by breaking them up into smaller tasks and assigning the smaller tasks to multiple workers who work simultaneously. Currently, nearly fifty percent of SHARCNET operations rely on parallel processing.

McCaughan, a SHARCNET HPC Consultant at the University of Guelph, regularly assists SHARCNET researchers in the development and maintenance of parallel computing processes for the production of research results. His experiences as both a university professor and an HPC consultant convinced him that “parallel programming concepts are extremely difficult to master – even for very experienced computer programmers. Their bases in parallelism – as opposed to the more linear serialism upon which the computing discipline generally relies – represent a conceptual stumbling block for many researchers and programmers.”

So in the fall of 2003, McCaughan, in conjunction with the University of Guelph’s Department of Computing and Information Sciences (CIS), offered the graduate course “Parallel Processing Architectures.” The twenty-six individuals who participated in the course came from a variety of disciplines, including CIS, Engineering, and Mathematics and Statistics. The majority of the participants were graduate students commencing SHARCNET-related research work, but a number of faculty members – SHARCNET-related or otherwise – sat in, too.

The course represented a vital resource for any researcher planning to use SHARCNET’s computing facilities because it not only taught the concepts behind parallel programming but also “provided students with the opportunity to explore the unique challenges of parallel design, implementation and analysis in a
structured setting.” In fact, for many students, parallel processing became so significant that, rather than merely employing it to further their research, they abandoned their current research in order to pursue work on the topic of parallel processing itself.

In the case of graduate student and SHARCNET Fellowship recipient Vimal Sharma, who works under Guelph CIS Professor and Acting Associate Dean of the Faculty of Environmental Science David Swayne, course assignments evolved into a ground-breaking project affiliated with Environment Canada. The team’s research and computing efforts integrate lake and atmospheric models in order to arrive at more accurate climate models for Canada. “Incorporating lakes in climate modeling is extremely important to a country with so many large bodies of water,” says Swayne. “It never would have occurred to us to attempt the modeling using parallel processing; however, the availability of both SHARCNET’s funding and its superb parallel processing facilities made the project entirely feasible.”

Dave McCaughan sees this example as evidence that the parallel processing course is worthwhile.

“A more in-depth approach to user education is invaluable to our researcher community and their students,” McCaughan remarks. “Researcher response to [the Guelph graduate course] has been overwhelmingly positive.” Both he and SHARCNET management hope to soon see the course available in a variety of formats at multiple SHARCNET partner institutions.

HOME

Safeguarding your water
Waterloo CRC Chair depends on HPC for groundwater research

Four years ago, the water tragedy in Walkerton, Ontario, reminded North Americans of just how crucial water safety is, and more, how increasingly complex it is to protect this precious resource.

Dr. Edward Sudicky, a Canada Research Chair (CRC) in Quantitative Hydrogeology and Professor of Earth Sciences at the University of Waterloo, is using High Performance Computing to do precisely that.

Sudicky and research group, who use computational models to describe fluid flow and the migration of contaminants in subsurface water (groundwater), are internationally recognized as leaders in Hydrogeology.

Hydrology is the science of water occurrence, movement and transport. Hydrogeology is the part of hydrology that deals with the occurrence, movement and quality of water beneath the Earth's surface.

According to Sudicky, groundwater is the largest reservoir of fresh water on Earth.

“In Canada alone, it has been estimated that about 20% of the total municipal and rural water consumption, including that used for domestic use, for irrigation or for other industrial activities, is derived from groundwater.”

However, he reveals, our supply of potable water is in serious jeopardy.

“This resource is under serious threat due to contamination from industrial, agricultural and urban activities. Moreover, groundwater discharges into surface water bodies and other environmentally sensitive ecosystems such as wetland areas. If the groundwater is contaminated, it threatens these natural assets.”

Ground water is not analogous with an underground river, but in fact is usually held in porous soil or rock materials, much the same way water is held in a sponge. Because hydrogeology deals with this complicated subsurface environment, it is a complex science.

“The migration rates and spreading patterns of toxic chemicals in groundwater are strongly influenced by the extreme heterogeneity of Earth materials,” Sudicky confirms. “All of these complexities have a profound bearing on issues such as the prediction of future impacts and the design of suitable groundwater-quality monitoring networks.”

This is where HPC comes in, and, according to Sudicky, its impact is enormous.

“Our recent strides in computational hydrogeology have positioned us to explore water safety issues in unprecedented detail,” he states. “I have been working in this area since the late seventies and have seen the immense impact that the increase in computational power has had on hydrogeological and hydrological computing.”

The models developed by Sudicky’s group over the past 20- years are used worldwide at universities, research institutions, government agencies and in the private environmental consulting industry.

His continuing research has wide-ranging implications, all of which, he confirms, are dependant on HPC.

“The computational models we are developing are becoming evermore complex, thus the demands for adequate computing resources are increasing exponentially. If our HPC computing demands are not be met in a timely fashion, research on topics such as the impact of climate change on water resources, or the safe disposal of high-level radioactive wastes, could not be met.

HOME

Innovation2
SHARCNET and HP host inaugural forum on public-private collaboration

The first annual SHARCNET-HP Academic-Industry Forum, held at Western on April 28, brought together public and private sector innovators from across Southwestern Ontario, to share cutting-edge research and development initiatives and explore opportunities for public-private sector collaboration.

The day long event, attended by nearly 70 participants, focused on relationship-building and opportunities for academic-industrial collaboration in the areas of Materials and Manufacturing, Electronic Device Automation and the Life Sciences.

While academic speakers focused on their work pioneering HPC techniques for potential industrial applications, private sector speakers highlighted current R&D successes and challenges, and future opportunities for collaboration with the SHARCNET research community.

Linda Gowman, VP of Research and Development at Trojan Technologies, a London-based water treatment technology firm, deemed the event an important educational opportunity for both communities.

“Many private sector companies are not fully aware of the potential for HPC-driven research to advance their R&D initiatives; and likewise academic researchers are not aware of the particular needs of these organizations,” said Gowman. “Events like this Forum are an excellent first step in facilitating potentially lucrative relationships.”

Trojan employs HPC to simulate large-scale water treatment systems, which are deployed for residential, municipal and commercial clients worldwide.

The event featured two keynote speakers; Mr. Richard Kaufmann from HP who spoke on the topic of Innovation in a Commodity World, and Dr. Mark Poznansky, President of Robarts Research Institute, who addressed the unique R&D challenges faced by Canada in Moving to a Knowledge-Based Economy…Will we succeed?

Lecture topics ranged from the development of nano-devices and lighter aluminums, to strategies for combating Mad Cow and Foot and Mouth Disease.

“This event was a tremendous opportunity to display research capabilities and initiate crucial relationships with private sector representatives,” confirms Eugene Kim, Assistant Professor of Physics and SHARCNET Chair of Computational Materials at the University of Windsor. Dr. Kim presented Fabrication and Modeling of Nanoscale Molecular Electronics Devices. Some of his current work involves fabricating and modeling novel electronics devices which utilize molecules. Future goals are to utilize these devices to study biological systems.

The event was generously sponsored by SHARCNET’s premier vendor partner, HP, as well as by the London Economic Development Corporation, the UWO Faculty of Science
McMaster Faculty of Science, Waterloo Faculty of Engineering, UOIT Faculty of Science, and Brock Faculty of Science.

Innovation2 will continue on an annual basis.

HOME

SHARCNET Campaign EXCELerates research

The Academic-Industry Forum was the final event on the 2004 EXCELerate calendar.

EXCELerate, an annual campaign which this year spanned the months of March and April, is designed to enhance SHARCNET’s outreach efforts to new and existing users, (academic, business and industrial) and educate a broader community on the uses and importance of High Performance Computing.

This year’s campaign included interactive training workshops, research colloquiums, and open houses. Workshops topics included Linux and Software Libraries, Parallel Programming and MPI, and were held at SHARCNET partner institutions across South Central Ontario, from the University of Windsor to The University of Ontario Institute of Technology in Oshawa.

In total, EXCELerate events attracted over 350 attendees, including nearly 300 at technical workshops, seminars and research colloquia.

Workshop and seminar slides are available at http://www.sharcnet.ca/pastworkshops.php

HOME

A simple competitive fact:
Newly appointed Scientific Director discusses SHARCNET, its challenges and its future

In December 2003, SHARCNET’s research support services received a tremendous boost with the appointment of Dr. Hugh Couchman as Scientific Director. Couchman, a Professor of Astrophysics at McMaster University and Fellow of the Canadian Institute for Advanced Research (CIAR) Cosmology programme, is also one of SHARCNET’s founders.

Of the highest quality and impact…

Couchman’s primary role as Scientific Director is to provide leadership and coordination for scientific activities within SHARCNET, which includes both outreach to the researcher community and strategic planning for technical development.

Fittingly, his focus is squarely on enabling innovative science.

“A fundamental part of this role is to ensure that SHARCNET's resources are positioned to enable and promote HPC-related research of the highest quality and impact.”

HQP: An overwhelming value…

In complement to this research focus, Couchman is also concerned with the development of an HPC culture in Ontario. He believes that in order to facilitate competitive research, both hardware and people need to be maintained and nurtured simultaneously.

“We must build a deep well of knowledge and expertise in the community if we are to have the necessary uptake of HPC for Ontario and Canada to succeed in contemporary research and among modern economies. This will occur by enabling collaboration, by building communities, by running training workshops, and by providing fellowships.”

As HPC evolves…

“HPC is developing an inexorable ubiquity in all aspects of research and will, increasingly, be relevant in non-academic settings,” he predicts. “SHARCNET must play a central role in helping people rise to the challenge of using HPC as it evolves.”

And the challenge, according to Couchman, is not purely academic and technical, but also economic.

“It is of vital importance that we maintain and grow the pool of expertise if Ontario and Canada are going to be competitively placed among the leading modern economies. This is a simple competitive fact.”
Professor Couchman completed his graduate work at the Institute of Astronomy, Cambridge. After receiving his Ph.D., he completed a postdoctoral position at the Canadian Institute for Theoretical Astrophysics followed by faculty positions at the University of Toronto and the University of Western Ontario before moving to McMaster University as a full professor. Couchman’s research focuses on the numerical modelling of the formation of cosmic structure. He is deeply involved in the effort to develop HPC communities and resources in Canada and is currently a member of the author’s panel for the C3.ca HPC Long Range Plan.

HOME

© 2003 The Shared Hierarchical Academic Research Computing Network.
To receive print copies of SHARC BTYES, please contact the Editor.