High Performance Computing (HPC) has become an essential and pervasive tool used in nearly all disciplines. SHARCnet makes freely available the latest hardware and software including programming, technical, and administrative support. SHARCnet is used daily by researchers not only from its nearly 20 strong member institutions but also by researchers from around the world. The Laurier SHARCNET Research Symposium 2009 (LSRS 2009) is open to everyone interested in High Performance Computing. This Symposium provides a rare opportunity to hear how researchers from WLU are using SHARCNET to broaden and accelerate their research progress putting them at the leading edge in their respective fields. If you are curious about how SHARCnet can impact your research work at Wilfrid Laurier then you are cordially invited to attend any or all of the talks. It will also be an excellent chance to ask any questions during the discussion/refreshment sessions and meet other WLU SHARCnet researchers. All talks will be held in Room BA 111. See Laurier Map for the exact location of the BA (Bricker Academic) building.

Time	Speaker	Title
11:50 - 12:00	Paul S. Maxim, Associate VP Research WLU	Opening Remarks
12:00 - 12:30	James Davenport, University of Bath, UK	So you want to buy a supercomputer?
12:30 - 12:50	Gabriel Moreno-Hagelsieb, WLU	From the pan-genome to the pan-interactome
12:50 - 13:10	Stephen Perry, WLU	The advancement of a neural network model for estimating body motion
		from plantar-surface pressure measurements during gait and gait perturbations
13:10 - 13:30	Angèle Hamel, WLU	Robot Searching with a Moving Target
13:30 - 13:50	Marek Wartak, WLU	Nonequilibrium Optical Gain in Semiconductor Quantum Well Structures
13:50 - 14:10	Refreshements and discussion
14:10 - 14:30	Dmitri G. Goussev, WLU	Chemistry and SHARCNET
14:30 - 14:50	Shohini Ghose, WLU	Quantum Chaotic Dynamics of Cold Atoms
14:50 - 15:10	Max Paliy, WLU	Multiscale coarse-grained models for RNA nanostructures
15:10 - 15:40	David McCaughan, SHARCnet	Supecomputing in a Box
15:40 - 16:00	Refreshements and discussion
16:00 - 17:00	Linda Petzold, UCSB, USA	Multiscale Simulation of Biochemical Systems (May CSASM talk)
17:00	Doug Roberts, SHARCnet	Tour of SHARCNET/WLU Visualization Facilites

Abstracts

• James Davenport, University of Bath, So you want to buy a supercomputer?
  
  The University of Bath (U.K.) recently procured a medium-scale supercomputer (800 nodes). The speaker is the project director for that procurement. He will describe the lessons he, and Bath, learned throughout this process, from inital planning, through procurement to acceptance and user training. He will also draw lessons from comparative procurements across the U.K.
  
  
• Gabriel Moreno-Hagelsieb, WLU, From the pan-genome to the pan-interactome
  
  In this talk I explore the use and limitations of genomic and metagenomic data for the inference of functional relationships of gene products by genomic, or rather contig (as in contiguous DNA sequence), context. Such inference strongly depends on the finding of homologous proteins and simple expectations about what should be observed if the products of genes work together. The basic assumptions behind genomic context are simple, while the difficulty of evaluation varies. The main avenues for context are: (a) gene fusions, where the prediction of a functional interaction arises when two genes are homologs to a single gene in another genome; (b) conservation of gene order, where two genes are conserved together in evolutionarily distant genomes; and (c) phylogenetic profiles where the partnership is inferred from the appearance and disappearance of genes in unison (co-occurrence). I will present the current difficulties and promises from each of these evidences when dealing with genomes, as well as the further difficulties and limited applicability of these evidences for functional interactions when using metagenomic sequences.
  
  
• Stephen Perry, WLU, The advancement of a neural network model for estimating body motion from plantar-surface pressure measurements during gait and gait perturbations
  
  This study was undertaken to determine if an artificial neural network based model could be used to approximate an individual's center of mass (COM) during dynamic movements in upright stance given only pressure data originating from pressure sensing insoles. This type of modelling may provide insight into how the human postural control system uses this sensory information to control balance. The activity was voluntary leaning in four directions (forward, right, left and backwards) all held for just over a second. The model demonstrated good prediction of the COM in the anterior/posterior direction such that the predicted COM approximation was within 10 mm of the measured COM. Extension of this model to 2-D space, incorporating medial/lateral information, has also give a good prediction of the COM location. Pilot work has begun on modeling the COM and it relationship to the base of support during gait using pressure insoles.
  
  
• Angèle Hamel, WLU, Robot Searching with a Moving Target
  
  A well-known problem in artificial intelligence involves a group of p robots searching for a target on m concurrent rays leading from the origin. The robots do not know ahead of time where the target is located and they succeed only if they find it. The problem is to determine an optimal search strategy for the robots. For example, how far should they go on each ray before turning around and trying another ray? How soon should they go back to a ray that has already been explored to a certain distance? The success of a search strategy is measured in relation to the time taken by an offline algorithm that knows ahead of time where the target is and can head straight towards it. The optimal strategy is known for a fixed target; this problem explores the variation where the target is moving towards the origin. The talk will describe this work in progress and also discuss the relationship to another problem involving scheduling jobs on parallel processors.
  This is joint work with Alex Lopez-Ortiz and Spyros Angelopoulos.
  
  
• Marek Wartak, WLU, Nonequilibrium Optical Gain in Semiconductor Quantum Well Structures
  
  Optical gain is a prime quantity in active optical semiconductor nanostructures [1]. Using the newly developed model of a self-consistent nano-scale semiconductor laser model we report on the analysis of optical gain and the role of non-equilibrium carriers [2], [3]. i Our model is based on a combination of Wigner functions for dynamical properties and Green's functions for spectral properties. The method was designed this way because Wigner functions are very convenient for the incorporation of boundary conditions while gain is more conveniently formulated with the Green's functions. In the specific example presented, the Green's functions were calculated in a Luttinger-Kohn model and the Wigner functions with a parabolic model. The nonequilibrium correction term is found from the difference between the carrier density calculated from the Wigner functions and the equilibrium Fermi function distribution used in standard gain formulations.
  
  References
  
  [1] M.S. Wartak and P. Weetman, in .Handbook of Semiconductor Nanostructures and Nanodevices, v. 4, p.409, edited by A.A. Balandin and K. L. Wang, American Scientific Publishers, Los Angeles (2006).
  [2] P. Weetman and M.S. Wartak, J. Comput. Theor. Nanosci. 3, 463.478 (2006).
  [3]. P. Weetman and M.S. Wartak, A Self-Consistent Nano-scale Semiconductor Laser Model Using Green's and Wigner functions in Two Bases., Physical Review B76, 035332 (2007).
  
  
• Dmitri G. Goussev, WLU, Chemistry and SHARCNET
  
  This presentation will highlight recent developments involving hardware and software acquisitions made by the Department of Chemistry for SHARCNET. Typical applications of SHARCNET resources in Chemistry will be discussed and selected examples of quantum-chemical calculations of molecular structure and properties will be presented.
  
  
• Shohini Ghose, WLU, Quantum Chaotic Dynamics of Cold Atoms
  
  The study of nonlinear dynamics in complex systems has led to great progress in the modeling and control of various real-world systems including lasers, biological systems, robotics, signal processing and financial systems. These large-scale systems commonly exhibit chaotic behaviour, characterized by extreme sensitivity of the dynamics to small perturbations in initial conditions. At the microscopic quantum level of atoms and photons, the concept of nonlinear dynamics and chaos is more ambiguous due to the apparent linearity of quantum mechanical evolution. We have explored the characterization and control of chaotic dynamics at the quantum level by analyzing a system of cold cesium atoms. We have performed extensive simulations to study the effects of chaos on quantum dynamics, and develop a fundamental understanding of the nature of quantum-classical correspondence in realistic systems. Our results have been used to design and implement the first ever experiments of such a quantum chaotic system. These studies provide a stepping stone towards the development of a toolbox of arbitrary quantum control necessary for a variety of future applications in quantum information processing.
  
  
• Max Paliy, WLU, Multiscale coarse-grained models for RNA nanostructures
  
  The conventional full atom Molecular Dynamics simulation methods are still too slow for many biomolecular systems, even when using the state-of-the-art computer facilities like SHARCNET. That is why the development of efficient coarse-graining (CG) models for such systems becomes increasing important. In this talk we describe a hierarchy of new mesoscopic models for the study of the molecular dynamics of RNA nanostructures, where the groups of neighbouring atoms are represented by the CG interaction centres - "beads". Such a coarse-grained treatment allows us to reach, for the systems of thousands of nucleotides, a time scale of microseconds and thus, to study the slow conformational motions of the RNA nanostructures. These new models enable simulations of larger and structurally more diverse RNA aggregates than it has been previously possible. We demonstrate that the "3-beads per nucleotide" models, described by a set of just a few universal parameters, are able to describe different RNA conformations and are comparable in structural precision to the models where the detailed values of the backbone P-C4' dihedrals taken from a reference structure are included. These new findings are discussed in the context of the RNA conformation classes based on P-C4' dihedrals. Our multiscale coarse-graining models also allow easy extensions towards sequence-specificity, reactivity and inclusion of the explicit electrostatics.
  This is a joint work with Roderick Melnik (M2NeT Lab, WLU, Waterloo, ON) and Bruce Shapiro (Center for Cancer Research Nanobiology Program, National Cancer Institute, Frederick, MD)
• David McCaughan, HPC Analyst, SHARCNET, University of Guelph, Supecomputing in a Box
  
  In their seminal reference, Gropp, Lusk and Skjellum muse that "to pull a bigger wagon it is easier to add more oxen then to find (or build) a bigger ox". Increasingly, the wagons for many researchers are computational problems and they are pulled by the computers on which we attempt to study or solve them. Fortunately for all of us, computers just keep getting faster...right?
  Parallel computing refers to the solution and/or simulation of problems by large numbers of processors (or computers) working together. Unfortunately, computer programs don't always cooperate as automatically as the beasts that we use to pull a wagon, so it isn't sufficient in most cases to simply add computational resources to the pile and realize faster execution or increased resolution in our programs.
  This talk is intended for those without any significant background in high performance computing and serves as an introductory survey of the essential concepts of computational parallelism. We will begin by motivating the need for parallelism, dismissing the notion that we can just wait for computers to get faster and thus solve our biggest problems. We will then consider highly accessible examples of parallel problem solving, including live demonstrations on a portable supercomputer we've built for this purpose. Time permitting, we will conclude with a high level view of how parallelism can be applied to existing problems in order to provide a level of literacy with the subject matter.
  
  
• Linda Petzold, University of California Santa Barbara, Multiscale Simulation of Biochemical Systems
  May CSASM seminar series talk
  In microscopic systems formed by living cells, the small numbers of some reactant molecules can result in dynamical behavior that is discrete and stochastic rather than continuous and deterministic. An analysis tool that respects these dynamical characteristics is the stochastic simulation algorithm (SSA). Despite recent improvements, as a procedure that simulates every reaction event, the SSA is necessarily inefficient for most realistic problems. There are two main reasons for this, both arising from the multiscale nature of the underlying problem: (1) the presence of multiple timescales (both fast and slow reactions); and (2) the need to include in the simulation both chemical species that are present in relatively small quantities and should be modeled by a discrete stochastic process, and species that are present in larger quantities and are more efficiently modeled by a deterministic differential equation. We will describe several recently developed techniques for multiscale simulation of biochemical systems, and outline some of the future challenges.
  
  

Symposium Organizers

• Ilias Kotsireas
• Doug Roberts