2012 NRAC Allocations

Each year, Compute Canada awards a significant fraction of its computational resources through a grant competition to Canadian researchers; the 2012 award announcement can be found here on this site, and as a PDF here.

The largest, national-scale awards are described below; a summary of all of the awarded resources is available here.

The ATLAS Collaboration

ATLAS is an experiment at CERN’s Large Hadron Collider (LHC) in Geneva which studies proton-proton and heavy-ion collisions at very high energy.  At the very moment of collision, the unprecedented energy densities created allow ATLAS to examine the structure of matter at a much finer scales than was possible before, to extend investigations of the fundamental forces of Nature, to understand the origin of matter and to search for physics beyond the Standard Model. Decades of experimental and theoretical development have led us to the brink of understanding the transition in the evolution of the universe when particles of massive matter emerged fractions of a second after the Big Bang.  The ATLAS group is confident that they will discover new physics, likely in the form of new particles, at the LHC.

But to make such discoveries requires that the enormous amount of data collected by ATLAS is be analyzed on an international network of high-performance computing centres linked together by Grid tools, the Worldwide LHC Computing Grid (WLCG). The availability of Compute Canada resources is crucial to our continued contribution to the groundbreaking discoveries coming from the Large Hadron Collider, the most sought-after of which is the Higgs boson, the particle thought to be central to the mechanism that gives subatomic particles mass. The enormous energy densities resulting from the collisions will also enable searches for physics beyond the Standard Model of particle physics, such as supersymmetry, quark substructure, and extra dimensions, to name a few.

  • Compute time Allocated: 17,733,000 processor-hours
  • Storage Allocated: 2,820 TB
  • Value of award: $5.16 million

André Bandrauk, Université de Sherbrooke

Time Frequency Spectra The response of molecules to ultrashort, intense laser pulses is the source of a new science – Molecular Photonics – with the aim of controlling the electrons in matter, with applications from chemistry, to biology, to materials science and even quantum information. A major thrust of this new science is dynamic imaging and control of quantum phenomena from the femtoseconds (10-15 seconds) for atomic rearrangements to attoseconds (10-18seconds – a billionth of a billionth of a second) for electron motions themselves. (For an overview of using this technology to view the unfolding of atomic systems, see “Quantum Dynamic Imaging“, co-edited by Prof. Bandrauk.

But to truly understand relativistic quantum effects in molecular systems, one must consider even shorter timescales – zepto seconds, or 10-21 seconds (a millionth of a millionth of a billionth of a second). Progress in this science require the numerical solutions of multidimensional time-dependent Schroedinger (TDSE) and Dirac (TDDE) equations with different time scales- femtosecond for nuclear motion, attosecond for electron motion and finally zeptosecond for relativistic effects. These equations are finally coupled to the laser Maxwell equations to include multiparticle collective effects due to pulse propagation. A major effort involves developing efficient numerical codes for Maxwell-TDSE, on large memory parallel machines based on in-house high level split-operator methods.These methods are now being generalized and developed for the relativistic4-component Dirac equations in intense laser fields with future applications to laser-induced nuclear fusion. Finally the TDSE and TDDE codes will be transported and rendered to a CFI funded CAVE (Computer Assisted Virtual Environment) in order to make MOVIES-Molecular Optical Visualization Imaging of Electrons-a new method of doing research and teaching in Molecular Optics.

  • Compute time Allocated: 73,247,000 processor hours
  • Storage Allocated: 141 TB
  • Value of award: $7.13 million

Yoshua Bengio, Université de Montréal

In recent years, Artificial Intelligence research has been heavily driven by the availability of large quantities of data and by the progress of Machine Learning algorithms designed to exploit it. These developments have led to numerous and ubiquitous applications, such as the search engines we use every day, recommendation systems for books or movies and speech recognition systems. Many learning algorithms require good representations of the data. Learning representations has been one of the central elements in the recent research in the area of Deep Learning. Deep learning algorithms discover multiple layers of representation, with each layer’s representation being defined in terms of the layer below.

Yoshua Bengio’s lab has been among the primary creators of of this new sub-field, with its own workshops and a vibrant community. Prof. Bengio’s team has contributed better understanding of these methods, new Monte-Carlo Markov Chain methods for training them, and completely new principles for learning new representations. In 2011, they won the Unsupervised and Transfer Learning Challenge, an international machine learning competition. Several important scientific questions remain open with respect to all these questions, and we will continue to expand the scale and scope of the application of deep learning methods. With AI-type tasks, scale and computational resources matter a great deal in achieving superior performance, requiring even bigger models (with tens or hundreds of millions of free parameters to be tuned) and bigger datasets (with millions or hundreds of millions of examples). Our current experiments involve loosely-coupled parallelization. These algorithms have important practical application of interest with several industrial partners, in particular Ubisoft, with an NSERC chair. They also allow Prof. Bengio’s group to advance the state-of-the-art in artificial intelligence and machine learning research.

  • Compute time Allocated: 1,964,000 processor hours
  • Storage Allocated: 19 TB
  • Value of Award: $210,000

Alan Evans, McGill University

Medical imaging research necessitates extensive data processing in order to explore the scientific questions of interest. This need has led to the development of an extensive array of data processing tools and complete processing workflows. However, these tools have been developed across a variable set of software environments, such that the execution of these tools may require specific operating system knowledge or software library configurations which limits the adoption rate of the tools and leads to frequent recreation of similar tools at different sites.

The Evans group has deployed a distributed middleware layer which encapsulate both the software environment and the specific processing tools such that sites can execute processing codes within their ideal environments without having to reconfigure their own computer systems for each set of codes. The CBRAIN web-interface currently interacts transparently with 11 remote clusters and servers across the country and in Europe. The platform has been used by approximately 100 collaborators distributed over 27 cities in 9 countries. Application domains vary from normal brain development to Alzheimers, early onset dementia and schizophrenia.

  • Compute time Allocated: 3,682,000 processor hours
  • Storage Allocated: 27 TB
  • Value of award: $383,000

Clinton Groth, University of Toronto

The Groth group at the Institute for Aerospace Studies develops new, advanced, computer algorithms to study the combustion of conventional and alternative fuels in practical devices, such as gas turbine engines. Combustion is inherently a multi-scale process that involves a wide range of complicated physical and chemical phenomena happening together on a wide range of length- and time-scales, making it extremely challenging to study; yet its importance to industrial processes and transportation is vital. Numerical predictions of burning flows rely heavily on reduced mathematical modelling and sophisticated methods to represent the underlying physics and make the problems more manageable. Unfortunately, current techniques and solution algorithms lack the detail of modelling needed to help design the next generation of quiet, high-efficiency, low-emissions combustors. Professor Groth’s group is remedying this situation by applying new and innovative mathematical models and computational tools to improve the understanding of combustion phenomena, and using this new knowledge in the design of more efficient and green engines burning both conventional and alternative fuels, such gaseous and liquid bio-fuels,syngas, ethanol, and bio-diesel. In the longer term, the research is expected to be beneficial to Canadians by enhancing energy security, reducing the environmental impact of combustion, and strengthening competitiveness of key industries in the energy and transportation sectors.

  • Compute time Allocated: 21,915,000 processor hours
  • Storage Allocated: 30 TB
  • Value of award: $2.12 million

Andriy Kovalenko, National Institute for Nanotechnology/University of Alberta


The Kovalenko Group develops of methods to predict properties of realistic nanoscience systems, and applies those methods and systems to nanotechnology problems of high industrial importance for Canada. Coupling computational and quantum chemistry with 3D statistical-mechanical molecular theory of solvation for electronic structure and nanocatalysis in solution; theories of electronic structure and quantum transport in molecular electronics nanostructures and nanodevices; statistical-mechanical molecular theory for electrochemistry in nanoporous materials; non-equilibrium statistical-mechanical molecular theory mass and charge transport at nanostructured surfaces and in nanoporous materials; molecular dynamics and coarse-grained dissipative particle dynamics simulations coupled with 3D molecular theory of solvation for solvent environment effects and ligand binding processes in chemical nanostructures and biomolecular systems.

The methodology is being applied to problems inspired and driven by the collaborators of the Kovalenko Group in Canada and worldwide, in academia, national laboratories, and industry.

  • Compute time Allocated: 16,155,000 processor hour
  • Storage Allocated: 3 TB
  • Value of award: $1.54 million

Richard Peltier, University of Toronto


Professor Peltier keeps his eye on planetary climate – from 750 million years ago to far into the future. As founder of U of T’s Centre for Global Change Science, Peltier is known worldwide for his research on global climate change. Using sophisticated mathematical concepts, he has developed powerful models to depict what happened to our climate over the past 750 million years and what is likely to happen in the future—if human behaviour does not change. His models of ice-age climate variability are considered the gold standard for research on climate change.

In 2010, Professor Peltier received the prestigious Bower Award and Prize for Achievement in Science in recognition of his ongoing contributions to the understanding of Earth systems. The first Canadian to receive the $250,000 award, he joined an impressive roster of previous recipients which includes Marie Curie, Thomas Edison, Albert Einstein and Stephen Hawking.

Among the most highly cited earth scientists in the world, Professor Peltier is also a dedicated mentor and teacher. More than 30 doctoral students have received their PhD degrees under his supervision.

  • Compute time Allocated: 12,710,000 processor hours
  • Storage Allocated: 370 TB
  • Value of award: $1.67 million

Ue-Li Pen, University of Toronto/Canadian Inst. for Theoretical Astrophysics

Pen’s group at CITA uses Compute Canada resources both to perform extremely large-scale simulations — such as investigating why the Milky Way’s black hole is so quiet — and data-analysis, examining enormous datasets from radio telescopes such as the Green Bay Telescope or the new Canadian CHIME Project.

Pen’s group aims to undertake some of the largest computer simulations ever in this field to examine the Baryonic Acoustic Oscillations – the Universe’s original ringing from the Big Bang – which will shed light on the original density and evolution of Dark Energy.

  • Compute time Allocated: 34,011,000 processor hours
  • Storage Allocated: 345 TB
  • Value of award: $4.46 million

Régis Pomès, University of Toronto/Hospital for Sick Children


The Pomès group specializes in the development and application of computer simulation techniques to the study of biomolecular processes. Using approaches grounded in statistical mechanics, they examine the structure, function, and dynamics of biomolecules; the folding, solvation, aggregation, and binding equilibria of proteins; and the transport of ions across biological membranes. These vital biomedical processes could not be examined at this level of detail without the availability of high-performance computing, and the efficient sampling algorithms developed in the group.

The group’s methodological advances have put them in a unique position to address challenging problems of protein self aggregation, to clarify the basis of protein solvation, and to expand the scope of simulation studies of membrane proteins up to and including the characterization of rare events essential for proper physiological function. Together, these studies will help bridge the gap between microscopic and macroscopic scales of important biological processes and provide meaningful new insight into the molecular basis of human health and disease. Ultimately, these advances will facilitate the development of biomimetic materials such as artificial skin and vascular grafts, as well as new therapeutic approaches for the treatment of numerous pathological ailments including pain, epilepsy, bacterial infections, and neurodegenerative diseases.

  • Compute time Allocated: 116,515,000 processor hours
  • Storage Allocated: 296 TB
  • Value of award: $11.4 million

Chris Pritchet, University of Victoria / CANFAR


The Canadian Advanced Network for Astronomical Research (CANFAR) supports forefront scientific discovery by Canadian astronomy researchers. CANFAR is an operational research portal for the delivery, processing, storage, analysis, and distribution of very large astronomical datasets. The portal is a collection of web services and middleware systems that connect computing and storage resources with data management and workflow management tools in support of collaborative projects in astronomy.

An innovative but challenging new feature of the research portal is the operation of services that channel the onslaught of telescope data through Canadian networks to the computational grid and data grid infrastructure (components of Compute Canada). CANFAR is currently used by many astronomy projects with users located at many institutions across Canada. These projects are using data generated by peer-reviewed allocations of a significant amount of observing time on three of Canada’s telescopes: the Canada-France-Hawaii Telescope, the James Clerk Maxwell Telescope and the Herschel Space Observatory as well as data from other facilities such as the Hubble Space Telescope and the MACHO project. Several of these projects are included in the Scientific Justification part of this request. Development of the CANFAR research portal came from a $2.29 million grant to the University of Victoria from the CANARIE Network-Enabled Platforms Program with support from the National Research Council’s Herzberg Institute of Astrophysics (NRC-HIA). NRC is committed to the long-term sustainability of the research portal as a tool and service for astronomers. The portal is operated and maintained by the Canadian Astronomy Data Centre (CADC). CANFAR eliminates the need for project teams to create project specific infrastructures and enable new projects that may not otherwise be pursued, given the challenges of creating a project infrastructure.

  • Compute time Allocated: 4,383,000 processor hours
  • Storage Allocated: 1,000 TB
  • Value of award: $1.65 million

Peter Tieleman, University of Calgary

In the past few years biomolecular simulation has become a critical component of a broad variety of research. Recent advancements through research in biomolecular simulation have significant impacts on many practical problems affecting day-to-day life and include compelling commercial and industrial applications. This applied side of biomolecular simulation is in addition to its traditional importance in fundamental research. The interactions between atoms and molecules are at the basis of every physical, chemical, and biological process.

Molecular simulation uses high performance computing and sophisticated models to understand and predict these interactions. Once primarily the area of statistical mechanics and theoretical physics, molecular simulation has become a powerful and universally used method in physics, chemistry, engineering, material science, biology, biotechnology, drug design, neuroscience and other areas. The Tieleman group are primarily interested in processes involving the cell membrane, the thin layer that envelopes every cell. This membrane plays a key role in biological processes including signalling, transport of nutrients and drugs, cell growth, neurotransmission, and bioenergetics.

  • Compute time Allocated: 39,446,000 processor hours
  • Storage Allocated: 150 TB
  • Value of award: $3.93 million

David Zingg, University of Toronto


The aircraft industry is committed to reducing overall CO2 emissions from aviation by 50% by 2050, which will require a great deal of technological progress. Improving aircraft fuel efficiency can address both rising fuel costs and the need to reduce CO2 emissions and is therefore a high priority in future aircraft design. The research of the Zingg group is aimed at drag reduction through novel unconventional aircraft configurations, innovative aerodynamic concepts, and flow control. It spans both the development of algorithms for aerodynamic and multidisciplinary optimization and their application to unconventional configurations and flow control. The long-term objective is to contribute toward the design of the next generation of aircraft with reduced environmental impact.

The Zingg group’s Compute-Canada powered computational research will apply novel computational and optimization methods to develop and evaluate unconventional aircraft concepts, and completely new ways of reducing the drag that planes experience.

  • Compute time Allocated: 28,577,000 processor hours
  • Storage Allocated: 2 TB
  • Value of award: $2.7 million