Featured Speakers
2012
Uncertainty Quantification in Computational Models
Wednesday, March 21, 2012
Location: BSF/1007 Darwin Room
Time: 9:00AM
Presented by:
Dr. Habib Najm
Combustion Research Facility
Sandia National Laboratories
Models of physical systems generally involve inputs/parameters that are determined from empirical measurements, and therefore exhibit a certain degree of uncertainty. Estimating the propagation of this uncertainty into computational model output predictions is crucial for purposes of model validation, design optimization, and decision support.
Recent years have seen significant developments in probabilistic methods for efficient uncertainty quantification (UQ) in computational models. These methods are grounded in the use of functional representations for random variables. In particular, Polynomial Chaos (PC) expansions have seen significant use in this context. The utility of PC methods has been demonstrated in a range of physical models, including structural mechanics, porous media, fluid dynamics, aeronautics, heat transfer, and chemically reacting flow. While high-dimensionality remains a challenge, great strides have been made in dealing with moderate dimensionality along with non-linearity and oscillatory dynamics.
In this talk, Dr. Najm will give an overview of UQ in computational models. He will cover the two key classes of UQ activities, namely: (1) the estimation of uncertain input parameters from empirical data, and (2) the forward propagation of parametric uncertainty to model outputs. He will cover the basics of PC UQ methods with examples of their use in both forward and inverse UQ problems, and will discuss methods for estimating the joint posterior density on uncertain model parameters given partial information from legacy experiments. View Lecture Series Poster
Stochastic Multiscale Modeling for Physical and Biological Problems
Friday, March 2, 2012
Location: BSF/1007 Darwin Room
Time: 10:00AM
Presented by:
Professor George Karniadakis
Applied Mathematics, Brown University
Research Scientist, Department of Mechanical Engineering, MIT
Stochastic modeling is required in many applications that involve multiscale phenomena, where information is lost as part of the coarse-graining procedure, e.g. formulating mesoscale dynamics equations from atomistic descriptions using the Mori-Zwanzig approach. It is also required in other systems with draconian approximations, e.g. spatial lumping, where effective parameters are employed to model some of the dynamics, and it can be in additive or multiplicative form. In many of these cases, the solution of the corresponding stochastic PDEs requires treating effectively problems in high dimensional spaces. In this talk, we will present effective new ways of dealing with this "curse-of-dimensionality" and we present demonstrative examples from fluid mechanics, electric networks, and biology. View Lecture Series Poster
Taming Heterogeneous Parallelism with Domain Specific Languages
February 14, 2012
Location: EMSL Auditorium
Time: 1:00PM
Presented by:
Professor Kunle Olukotun
Stanford University
Computing systems are becoming increasingly parallel and heterogeneous; however, exploiting the full capability of these architectures is complicated because it requires application code to be developed with multiple programming models. A much more productive single programming model approach to heterogeneous parallelism uses domain specific languages (DSLs). DSLs provide high-level abstractions which improve programmer productivity and enable transformations to high performance parallel code. In this talk, Professor Olukotun will motivate the DSL approach to heterogeneous parallelism; show example DSLs from the domains of graph analysis, mesh -based PDE solvers, and machine learning that provide both high productivity and performance; and describe Delite, a framework that simplifies the development of DSLs embedded in Scala. View Lecture Series Poster
Memory Systems for Extreme Scale Computing
January 27, 2012
Location: BSF Darwin Room - 1007
Time: 10:00AM
Presented by:
Professor Sun, Xian-He
Chair,
Department of Computer Science Illinois Institute of Technology
Technology advances are unbalanced. CPU performance has been improving at a much faster pace than memory technologies during the last three decades, which has led to the so-called memory-wall problem. In the meantime, newly emerged IT applications, such as computer animation, social networks, and sensor networks, are all data intensive, which has led to the so-called big-data problem. The lasting memory-wall problem compounded with the newly emerged big-data problem has changed the landscape of computing. CPU speed is no longer the performance bottleneck of a computing system, the data access speed is. However, historically computing systems are designed and developed to utilize CPU performance, not data accessing. A paradigm change is needed to support data-centric computing. In this talk we first review the history and concepts of the big-data and memory-wall problems. We then discuss the challenges of design advanced memory systems for extreme-scale computing. Finally, we present some our recent results in understanding and optimizing the performance of memory systems from the data-centric point-of-view.
Professor Sun is the chairman and a professor of the Department of Computer Science, the director of the Scalable Computing Software laboratory at the Illinois Institute of Technology (IIT) and a guest faculty in the Mathematics and Computer Science Division at the Argonne National Laboratory. Before joining IIT, he worked at DoE Ames National Laboratory, at ICASE, NASA Langley Research Center, at Louisiana State University, Baton Rouge, and was an ASEE fellow at Navy Research Laboratories. Dr. Sun is an IEEE fellow and his research interests include parallel and distributed processing, high-end computing, memory and I/O systems, and performance evaluation. He has close to 200 publications and four patents in these areas. View Lecture Series Poster
The PhyloFacts Microbial Phylogenomic Encyclopedia:
Phylogenomic tools and web resources for the Systems Biology Knowledgebase
Dr. Kimmen Sjölander
January 12, 2012
Location: BSF Darwin Room - 1007
Time: 1:00PM
Presented by: Dr. Kimmen Sjölander
Associate Professor
Bioengineering University of California, Berkeley
Dr. Sjölander will present an overview of the PhyloFacts Microbial Phylogenomic Encyclopedia, which was designed to improve the accuracy of functional annotation of microbial genomes through evolutionary reconstruction of gene families, integrating information from protein 3D structure, biological process, pathway association, protein-protein interaction and other types of experimental data to improve both the specificity and coverage of protein "function" prediction. Her talk will cover structural phylogenomics, key computational challenges being addressed in the PhyloFacts project, and a novel approach to ortholog identification. View Lecture Series Poster
2011
Programming Models in the Exascale Era
Prof. Barbara Chapman
Thursday, December 15, 2011
Location: BSF Darwin Room - 1007
Time: 9:30 AM
Presented by: Professor Barbara Chapman, University of Houston Department of Computer Science
Researchers around the world have begun to consider the requirements and implications of future exascale systems. It is assumed today that such platforms will be introduced around 2018 - 2020, and that they will be large clusters with very powerful compute nodes. One of the most urgent open questions is how these systems will be programmed. Will MPI continue to dominate at this level? If so, how will the application developers achieve high performance within each compute node? If not, what will the programming model be? And how will existing applications be migrated to this new kind of platform?
In this presentation, Professor Chapman will discuss the exascale landscape and some ideas on the programming models that might be provided on such platforms. The talk will include considerations for what changes will be needed in the execution environment in order to support efficient application deployment.
View Lecture Series Poster
An Overview of the SciDAC-3 Institute for Performance, Energy, and Resilience
Dr. Bob Lucas
Thursday, December 8, 2011
Location: BSF Darwin Room 1007
Time: 9:00 AM
Presented by: Dr. Bob Lucas, University of Southern California Information Sciences Institute
Over the next five years (2012--2016), computational scientists working on behalf of the Department of Energy's Office of Science (DOE SC) will exploit a new generation of petascale computing resources to make previously inaccessible discoveries in a broad range of disciplines including chemistry, fusion energy, materials science, and physics. The computational systems underpinning this work will increase in performance potential from tens to hundreds of petaflops and will evolve significantly from those in use today: concurrency will scale exponentially; accelerators such as graphical processing units (GPUs) will be utilized; and even the memory hierarchy will change with the incorporation of a new generation of persistent devices (e.g., phase change memory). To ensure that DOE's computational scientists can successfully exploit this emerging generation of leadership-class computing systems, the University of Southern California (USC) has assembled a broad team of computer scientists with the expertise to address their most pressing challenges: (a) end-to-end performance optimization, including single-node performance, interprocessor communication, load balancing and I/O; (b) performance portability for new systems, including heterogeneous processors and new memory hierarchies; (c) management of energy consumption; and (d) resilient computation.
View Lecture Series Poster
Blue Waters: An Extraordinary Research Capability for Advancing Science & Engineering
Friday, February 25, 2011
Location: EMSL Auditorium
Time: 1:00PM
Presentation: Blue Waters: An Extraordinary Computer to for Enable Extraordinary Research
Dr. Thom H. Dunning, Jr., Director of the National Center for Supercomputing Applications and the Institute for Advanced Computing Applications and Technologies, University of Illinois at Urbana-Champaign, will talk about the new supercomputer Blue Waters and its proposed use by the science and engineering community.
A dramatic increase in computing capability has the potential to create breakthrough advances in all fields of science and engineering, including predicting the behavior of complex biological systems, understanding the production of heavy elements in supernovae, designing catalysts and other materi-als at the atomic level, predicting changes in the earth's climate and ecosystems, and designing complex engineered systems from chemical plants to airplanes. However, achieving these breakthroughs requires a computing system capable of answering the most compute-, memory- and data-intensive research questions.
The Office of Cyberinfrastructure in the National Science Foundation is funding the acquisition and deployment of an extraordinary new supercomputing system at the National Center for Supercomputing Applications on the campus of the University of Illinois at Urbana-Champaign. This system, called Blue Waters, is based on the latest computing technology under development by IBM for DARPA's High Productivity Computing Systems Program, including the Power7 processor with a peak performance of a quarter of a teraflop and a new communications hub chip with a total bandwidth of more than one terabyte/sec. Blue Waters will be installed in NCSA's National Petascale Computing Facility in 2011.
Blue Waters will have more than 300,000 compute cores, 1000 terabytes of main memory, 25 petabytes of disk storage, and up to 500 petabytes of archival storage. It will be connected to the nation's research networks at 100+ Gbps. Blue Waters will enable the solution of the most challenging data-intensive problems. Detailed performance projections indicate that Blue Waters will sustain 1 petaflops on many real-world science and engineering applications.
View Lecture Series Poster
Multi-reference coupled-cluster methods: Three different approaches
Monday, February 14, 2011
Location: EMSL Auditorium
Time: 1:00PM
Professor Bartlett will discuss the multi-reference problem in coupled-cluster theory, and describe three different approaches for solving the problem. All are compared for cyclobutadiene's autoisomerization and other multi-reference problems.
Professor Bartlett pioneered the development of coupled-cluster theory in quantum chemistry to offer highly accurate solutions of the Schrödinger equation for molecular structure and spectra. His group is responsible for the widely used ACES program system.
His other research topics include the search for metastable, high-energy
density molecules like N5-; non-linear optics; carbon clusters; NMR
coupling constants; new correlated quantum chemical methods for
polymers and surfaces; ab initio density functional theory; and the
'transfer Hamiltonian' for large scale quantum mechanical simulations of
materials.
View Lecture Series Poster
Architecture-Aware Algorithms and Software for Scalable Performance and Resilience on Heterogeneous Architectures
Monday, January 24, 2011
Location: EMSL Auditorium
Time: 1:00PM - 2:00PM
Presentation: Architecture-aware Algorithms and Software for Peta and Exascale Computing
Professor Jack Dongarra, Distinguished Research Staff member at Oak Ridge National Laboratory and a University Distinguished Professor in the Deparment of Electrical Engineering and Computer Science at the University of Tennessee, will present a seminar titled "Architecture-Aware Algorithms and Software for Scalable Performance and Resilience on Heterogeneous Architectures."
Professor Dongarra will examine how high performance computing has changed over the last 10 years and look toward the future in terms of trends. These changes have had and will continue to have a major impact on software. Some of the software and algorithm challenges have already been encountered, such as management of communication and memory hierarchies through a combination of compile-time and run-time techniques, but the increased scale of computation, depth of memory hierarchies, range of latencies, and increased run-time environment variability will make these problems much harder. He will look at five areas of research that will have an importance impact in the development of software and algorithms, focusing on redesign of software to fit multicore architectures; automatically tuned application software; exploiting mixed precision for performance; the importance of fault tolerance; and communication avoiding algorithms. View Lecture Series Poster
2010
Operating System Resource Management
Monday, September 27, 2010
Location: ETB, Columbia River Room 1103
Time: 2:00PM - 3:00PM
Presentation: Operating System Resource Management
Burton Smith, a Technical Fellow at Microsoft, will give a talk titled "Operating System Resource Management" on Monday, September 27, from 2-3 pm in the Columbia River Room in ETB. Resource management is the dynamic allocation and de-allocation by an operating system of processor cores, memory pages, and various types of bandwidth to computations that compete for those resources. The objective is to allocate resources so as to optimize responsiveness subject to the finite resources available. Historically, resource management solutions have been relatively unsystematic, and now the very assumptions underlying the traditional strategies fail to hold. First, applications increasingly differ in their ability to exploit resources, especially processor cores. Second, application responsiveness is approximately two-valued for "Quality-Of-Service" (QOS) applications, depending on whether deadlines are met. Third, power and battery energy have become constrained. This talk will propose a scheme for addressing the operating system resource management problem.
View Lecture Series Poster
Modeling of physical systems underspecified by data
Friday, September 10, 2010
Location: ETB, Columbia River Room 1103
Time: 1:30PM - 2:30PM
Presentation: Modeling of physical systems underspecied by data
Although it has long been recognized that simulations of most physical systems are fundamentally stochastic, this fact remains overlooked in most practical applications. Even essentially deterministic systems must be treated stochastically when their parameters, boundary and initial conditions, or forcing functions are under-specified by data. Data-driven random domain decompositions provide a novel approach to dealing with the kinds of spatially heterogeneous random processes that typically appear in realistic simulations of physical systems. The method is based on a doubly stochastic model in which the problem domain is decomposed according to stochastic geometries into disjoint random fields. View Lecture Series Poster
A General Computable Methodology for Validation
Monday, August 16, 2010
Location: ETB Columbia River Room 1103
Time: 11:00AM
Professor Roger Ghanem, of the Viterbi School of Engineering at the University of Southern California, will present a seminar titled "A Computable Approach To Validation" at 11:00 am on August 16 in the Columbia River Room in ETB.
Professor Ghanem will discuss uncertainty quantification, verification, and validation and the drive to implement computational resources to reproduce or predict physical reality. The talk will describe research efforts and challenges to understanding the scope of model-based predictions and the necessity of packaging of information so that it can be analyzed and so that reliability estimates can be computed. He will also describe applications in porous media, material modeling, and complex networks. View Lecture Series Poster
Beyond Reactive Management of Network Intrusions
Monday, July 26, 2010
Location: ETB Columbia River Room
Time: 10:00AM
Presentation: Beyond Reactive Management of Network Intrusions
Professor Sushil Jajodia, of the Center for Secure Information Systems at George Mason University, will present a seminar titled "Beyond Reactive Management of Network Intrusions" at 10:00 am on July 26 in the Columbia River Room in ETB.
Prof. Jajodia will discuss issues and methods for protecting systems from malicious attacks, and system survival from attacks that cannot be averted at the outset. He will describe recent research on attack graphs that represent known attack sequences attackers can use to penetrate computer networks, and will show how attack graphs can be used to compute actual sets of hardening measures that guarantee the safety of given critical resources. The session will also include a demo of the working system and an overview of the Center for Secure Information Systems. View Lecture Series Poster
2009
Informatics Role in the Understanding of Infectious Diseases
Wednesday, February 4, 2009
Location: EMSL Auditorium
Time: Noon
Dr. Bruno Sobral and Dr. Chris Barrett, both of Virginia Bioinformatics Institute at Virginia Tech, will present a lecture titled "Informatics and Computational Sciences," at noon on February 4, in the EMSL Auditorium.
The common aspiration that emphasizes properties of systems, as opposed to the characteristics of data, calls for holistic explanations that are inherently complex, the calculation of which are computationally intensive. Biosystems, from molecular information storage and functional prescription to ecologies, exhibit all of the features of systems that can only be understood systemically and require specialized computational support—informatics—to analyze and comprehend. This presentation will provide an overview of molecular, cellular, organism and population-level systems biology of infectious disease. The speakers will start with genomic description and explanation and move toward population-level public health issues, emphasizing the role of informatics support in the scientific understanding of contagion, as an example.
Dr. Sobral is executive and scientific director and professor at the VBI. Dr. Sobral has a long-standing interest in systems approaches to infectious diseases. His research program fully utilizes wetlab and cyberinfrastructure programs. Dr. Barrett is director of the Network Dynamics and Simulation Science Laboratory at VBI. His research interest include simulation of very large systems; theoretical foundations of simulation; interaction-based systems, computing, and dynamical systems; computational and systems biology; computational problems in epidemiology; cognitive science and computationally aided reasoning; computational economics; infrastructure simulation. View Lecture Series Poster
2008
Dr. Larry Smarr
Shrinking the Planet
"How Dedicated Optical Networks Are Transforming Computational Science and Collaboration"
Monday, August 25, 2008
EMSL Auditorium
10:30 AM - 11:30 AM
Larry Smarr, professor at California Institute for Telecommunications and Information Technology, will speak on how dedicated optical networks are transforming computational science and collaboration.
View Dr. Smarr's Lecture Series Poster | Video
Dr. Tony Hey
Corporate Vice President
Microsoft Research
"eScience, Semantic Computing and the Cloud: Towards a Smart Cyberinfrastructure for eScience"
Monday, June 2, 2008
Battelle Auditorium
9:00 AM
Presentation: eScience, Semantic Computing and the Cloud