Colloquia
Google Calendar of Colloquia
Future Colloquia (Tentative Schedule)
Info on Colloquia Requirements for Students
For students taking the colloquia course, here is some information on requirements for Fall 2008.
Nonconvex compressive sensing: getting the most from very little information (and the other way around)
Date: Friday, October 10th, 2008
Time: 2 pm — 3:15 pm
Place: ME 218
Rick Chartrand
Los Alamos National Laboratory
In this talk we'll look at the exciting, recent results showing that most images and other signals can be reconstructed from much less information than previously thought possible, using simple, efficient algorithms. A consequence has been the explosive growth of the new field known as compressive sensing, so called because the results show how a small number of measurements of a signal can be regarded as tantamount to a compression of that signal. The many potential applications include reducing exposure time in medical imaging, sensing devices that can collect much less data in the first place instead of collecting and then compressing, getting reconstructions from what seems like insufficient data (such as EEG), and very simple compression methods that are effective for streaming data and preserve nonlinear geometry.
We'll see how replacing the convex optimization problem typically used in this field with a nonconvex variant has the effect of reducing still further the number of measurements needed to reconstruct a signal. A very surprising result is that a simple algorithm, designed only for finding one of the many local minima of the optimization problem, typically finds the global minimum. Understanding this is an interesting and challenging theoretical problem.
We'll see examples, and discuss algorithms, theory, and applications.
Bio:
Born in Canada, Rick Chartrand's education was in pure mathematics, receiving a Ph.D. from UC Berkeley in 1999 for work in functional analysis, in a subfield devoid of useful applications. He now works as an applied mathematician at Los Alamos National Laboratory. His research interests are focused on image and signal reconstruction from very incomplete information. He works on developing new algorithms, solving the underlying theoretical issues, and exploring useful applications.
Petascale Computational Science on Roadrunner
Date: Friday, October 3rd, 2008
Time: 2 pm — 3:15 pm
Place: ME 218
Timothy C. Germann
Deputy Group Leader of the Theoretical Chemistry & Molecular Physics Group (T-12)
LANL
Abstract:
I will describe the initial set of scientific applications and computational kernels that have been implemented on the hybrid Roadrunner supercomputer recently constructed by IBM for Los Alamos National Laboratory (LANL). Each Roadrunner "triblade" compute node consists of two AMD Opteron dual-core microprocessors and four PowerXCell 8i enhanced Cell microprocessors, typically utilized with a programming model of four MPI ranks (with one Opteron core and one Cell each) per node. I will describe our adventures first porting, and then drastically rewriting, a molecular dynamics code, SPaSM (Scalable Parallel Short-range Molecular dynamics), which is used for a wide range of scientific studies at LANL, ranging from fluid and materials dynamics to agent-based computational epidemiology. The computation of forces and updates of particle positions and velocities are performed on the Cells (each with one PPU and eight SPU cores), while the Opterons direct inter-rank communication and perform periodic I/O-heavy tasks including analysis, visualization, and checkpointing. The nearly perfect weak scaling measured for a standard Lennard-Jones pair potential benchmark yields 369 Tflop/s (double-precision) floating-point performance on the full Roadrunner system (3060 compute nodes), and is a finalist for the 2008 ACM Gordon Bell Prize to be announced at SC08 in November.
Bio:
Timothy C. Germann is Deputy Group Leader of the Theoretical Chemistry & Molecular Physics Group (T-12) at Los Alamos National Laboratory (LANL). He earned Bachelor of Science degrees in Computer Science and in Chemistry from the University of Illinois in 1991, and a Ph.D. in Chemical Physics from Harvard University in 1995. Following a Research Fellowship in the Miller Institute for Basic Research in Science at UC Berkeley, where he developed parallel algorithms for quantum molecular (reactive) scattering theory, Tim joined LANL in 1997, where he has used large-scale classical molecular dynamics simulations to investigate shock, friction, detonation, and other materials dynamics issues using BlueGene/L, Roadrunner, and other NNSA supercomputer platforms. Along the way, he and his collaborators developed a large-scale epidemiological simulation model and applied it to assess mitigation strategies for outbreaks of either naturally emerging or intentionally released infectious diseases, including pandemic influenza. He has co-authored over 100 peer-reviewed scientific publications, and received a 1998 IEEE Gordon Bell Prize, 2005 and 2007 LANL Distinguished Performance Awards, a 2006 LANL Fellows' Prize for Research, and a 2007 NNSA Defense Programs Award of Excellence.
Roadrunner: A Petaflop/s before its time
Date: Friday, September 26th, 2008
Time: 2 pm — 3:15 pm
Place: ME 218
Ken Koch
Technical Project Leader for Roadrunner
LANL, CCS-DO
Abstract:
The Roadrunner supercomputer was built by IBM for Los Alamos National Laboratory for the Advanced Simulation and Computing (ASC) Program. Roadrunner achieved 1.026 Petaflop/s running the TOP500 Linpack benchmark on May 26th, 2008 breaking the Petaflop/s barrier sooner than the TOP500 data would have predicted. Los Alamos conceived Roadrunner as a way to enable faster, more energy-efficient, and lower-cost computing through the use of acceleration devices in a hybrid computing design, in this case the Cell micro-processor as accelerator to an AMD Opteron. Many believe that the multi-core and many-core future of micro-processors will include the use of a non-uniform mix of devices and/or cores, some of which will have special functionality. Roadrunner is also a platform to prepare for that trend.
This talk on Roadrunner will provide the configuration details of this hybrid Cell-accelerated supercomputer and how it works. It will introduce the modified Cell processor and the hybrid TriBlade compute node developed for Roadrunner. The talk will also cover our applications experiences and the programming approach taken by early applications converted by Los Alamos staff to run on the accelerated Roadrunner machine; two of these applications are finalists for this year’s Gordon Bell award at SC08.
Bio:
Kenneth Koch has worked at Los Alamos National Laboratory since 1985 in the fields of nuclear weapons simulation and high -performance computing. He helped create the ASCI (now ASC) Program and served as the program manager of all ASC simulation codes at Los Alamos for several years. Since 2004 Ken has been involved in high-performance computing in the Computer , Computational, and Statistical Science Division (CCS) at LANL. He has led efforts in advanced computer architectures including one using FPGAs and GPUs for scientific computing. He was the main LANL architect behind the design and implementation of the Roadrunner Cell-accelerated hybrid supercomputer. Ken received a PhD in Nuclear Engineering from Purdue University in 1985 and also holds a Masters degree plus two Bachelors degrees in Nuclear and Mechanical Engineering, also from Purdue.
Implementing Scheme in a Virtual World
Date: Friday, September 12th, 2008
Time: 2 pm — 3:15 pm
Place: ME 218
Lance R. Williams
University of New Mexico
Department of Computer Science
Abstract:
At any given moment in time, hundreds of thousands of people worldwide are immersed in dozens of virtual worlds playing massive multiplayer online games (MMOG's). Second Life is unique among MMOG's because the players themselves create the content of the virtual world they inhabit. They do this (in large part) by means of computer programming and I believe this fact makes Second Life a potentially important resource for computer science educators.
Furthermore, because Second Life supports a programming model where a large number of small scripts execute in parallel and asynchronously, and communicate via message passing, it is an ideal testbed for research in many areas of computer science, including distributed computing, swarm robotics, self-assembly, distributed sensor networks, and artificial life.
In this talk, I first provide an overview of Second Life and its programming model. I then describe and demonstrate a series of evaluators for the Scheme programming language which I have constructed inside the game, including one evaluator where the heap and virtual machine are represented in a completely distributed fashion as a school of swimming fish. In this way, I hope to illustrate Second Life's value to computer science pedagogy and as a testbed for research in distributed computation.
Bio:
Lance R. Williams received his BS in Computer Science from the Pennsylvania State University in 1985 and his MS and Ph.D in Computer Science from the University of Massachusetts at Amherst in 1988 and 1994. His dissertation, in the area of computer vision, was on perceptual completion of surfaces which are only partially visible. After completing his Ph.D., he spent four years at NEC Research Institute in Princeton, NJ where he developed a series of increasingly more general neural models of the process used by the human visual system to compute the shape of object boundaries where they cannot be directly observed. In 1997, Dr. Williams joined the faculty of the Dept. of Computer Science at the University of New Mexico where he is currently an Associate Professor. His research since joining UNM has addressed a range of topics in computer vision, neural computation, digital image processing, and human and computer interaction. Since his first exposure to it in the early 1980's, he has had an enduring interest in the LISP programming language and its implementation.
Home | Colloquia | Research | People
Current Students | Prospective Students | Companies | Visitors
Scholarships | Courses | Degrees | PIBBS
Computer Facilities | Jobs | Links
(505) 277-3112 (phone)
(505) 277-6927 (fax)
Web site questions:
All other questions:
Department of Computer Science
Mail stop: MSC01 1130
1 University of New Mexico
Albuquerque, NM 87131-0001
