InTheLoop | 10.19.2009
The weekly newsletter for Berkeley Lab Computing Sciences
October 19, 2009
NERSC, Argonne to Explore Scientific Cloud Computing
Cloud computing is gaining traction in the commercial world, but can such an approach also meet the computing and data storage demands of the nation’s scientific community? A new program funded by the American Recovery and Reinvestment Act through the U.S. Department of Energy (DOE) will examine cloud computing as a cost-effective and energy-efficient computing paradigm for scientists to accelerate discoveries in a variety of disciplines, including analysis of scientific data sets in biology, climate change and physics.
To test cloud computing for scientific capability, DOE centers at the Argonne Leadership Computing Facility (ALCF) in Illinois and the National Energy Research Scientific Computing Center (NERSC) in California will install similar mid-range computing hardware, but will offer different computing environments. The combined set of systems will create a cloud testbed that scientists can use for their computations while also testing the effectiveness of cloud computing for their particular research problems. Since the project is exploratory, it’s been named Magellan in honor of the Portuguese explorer who led the first effort to sail around the globe and for whom the “clouds of Magellan” — two small galaxies in the southern sky — were named. Read more.
Phase 1 of Cray XT5 Is Delivered to NERSC
On October 12, Phase 1 of NERSC’s new Cray XT5 system, named Hopper after computer scientist Rear Admiral Grace Murray Hopper, was delivered to Berkeley Lab’s Oakland Scientific Facility, where installation is under way. Click here for a slide show of the delivery.
The new Cray system will provide many pioneering features, including the ability for users to customize the operating system for their own codes and to schedule jobs and access their data without logging in to the supercomputer. Cray’s new cooling system and interconnect network technology mesh well with NERSC’s research efforts into energy efficient computing and programming models. When scheduled upgrades are completed in late 2010, the new system will deliver a peak performance of more than one petaflops, equivalent to more than one quadrillion calculations per second.
Lasers Without Mirrors, Designed by Supercomputer
Sometimes it takes a big machine to understand the tiniest details, says a new article in ASCR Discovery. That’s the case with free electron lasers (FELs). The powerful X-rays they generate can probe matter directly at the level of atomic interactions and chemical-bond formation, letting scientists observe such phenomena as chemical reactions in trace elements, electric charges in photosynthesis and the structure of microscopic machines.
To succeed, FEL designs must be optimized to produce and preserve high-brightness electron beams, characterized by high current and low emittance. A team of researchers at Berkeley Lab, led by Ji Qiang and John Corlett, are using Franklin, the Cray XT computer at NERSC, to shed light on these designs. Read more.
Danish Students Get Access to NERSC’s Franklin Supercomputer
Although the CITRIS/LBNL Computational Science and Engineering (CSE) program is based in Berkeley, it’s beginning to have a worldwide impact.
On February 2, 2009, InTheLoop reported that two attachés from the Innovation Center Denmark/Silicon Valley had met with Masoud Nikravesh, Executive Director of CITRIS CSE, and David Skinner of NERSC to explore ideas for collaboration. As a result of that meeting, 15 students from the University of Southern Denmark (SDU) will now have access to NERSC’s Franklin supercomputer.
A post on the Danish blog Silicon Valley Tech Talk, titled “Danish students get access to California Supercomputer,” says that access to Franklin will give the SDU students some practical experience in a computing course that had been very theory heavy. Jim Demmel of UC Berkeley and CRD provided the students with course materials and examples from his popular course “Applications of Parallel Computers.”
VisIt Tutorial Gets Rave Review in VizWorld
A tutorial on the VisIt visualization software given last Monday (October 12) at the IEEE Visualization 2009 conference by Hanks Childs of the CRD/NERSC Visualization and Analytics Group and Sean Ahern of the University of Tennessee and Oak Ridge National Laboratory, got a rave review in the online magazine VizWorld.
A live demonstration of remote visualization and analysis with VisIt over the conference WiFi to NERSC’s SGI Altix, DaVinci, worked flawlessly the first time, “much to everyone’s surprise,” according to the review. The reviewer was also impressed by VisIt’s histogram-based parallel coordinates display, a technology invented by the Vis Group for Cameron Geddes’ accelerator project last year, and presented in the SC2008 technical program. The review concludes:
I haven’t taken a good long in-depth look at VisIt in about two years, but after this I’ve decided I’m going to. The functionality and power is simply too big to ignore. The work they’ve put into support for massively parallel systems and HPC configurations is astounding, and combined with the impressive catalog of data analysis tools it really looks more “useful” than their competitors like ParaView and EnSight.
New Web Page Tracks Computing Sciences in the News
A new web page created and maintained by Linda Vu of the Computing Sciences communications team tracks media reporting on CS research, people, and news. The Magellan project is dominating the news this week. A surprisingly varied group of websites are posting news from CS, including the San Francisco Chronicle, United Press International, Popular Science, MSNBC, Forbes.com, Discovery Channel, Fox Business News, and many more specialized publications. The web page is updated weekly.
Computing Sciences, ITD Recruiting for More Than 20 Positions
Berkeley Lab Computing Sciences and the IT Division are currently recruiting for more than 20 positions, and recruiter Bernadette Cu-Todd has some tips for job seekers in a video on the CS home page. For more information about employment opportunities, click here.
CS Staff Contribute to Performance Analysis Workshop
At last week’s (October 13–14) Los Alamos Computer Science Symposium 2009, two CS staffers made presentations in a Workshop on Performance Analysis of Extreme-Scale Systems and Applications. Harvey Wasserman presented “Extreme Scale Data Intensive Computing at NERSC,” and Lenny Oliker of CRD presented “Parallel I/O Performance: From Events to Ensembles.”
Lab OCFO Is Offering Free Property Pickup November 2–13
The Office of the Chief Financial Officer's (OCFO) Property Management Division is sponsoring a fall cleanup at no cost to requesters between November 2 and 13. Even though free weekly pickups are available year-round, this campaign offers an incentive to get rid of old equipment. Group leads can ask Parisa Farvid (x4965) to create lists of equipment by staff member to justify retention.
Facilities will be accepting electronic work order requests from October 19–30 for the pickup. Request a pickup online and click the “cleanup” box. Include a description of all items and attach a transportation authorization form (TAF) to all items. Property items must weigh less than 250 pounds and cannot be hazardous or universal e-waste. For information about hazardous or universal e-waste pickups, contact Betsy MacGowan (x2826).
Linux Cluster Institute (LCI) Conference Issues Call for Papers
The Pittsburgh Supercomputing Center (PSC) will host the 11th LCI International Conference on High-Performance Clustered Computing, March 8–11, 2010. This year the Conference will focus on the impact of multi-core technologies, including heterogeneous approaches using GPGPUs and Cells, on high end computing; and the power consumption and scaling challenges of peta- and exascale computing.
The conference program committee solicits novel papers, technical presentations, tutorials, and posters on a broad range of topics related to the conference focus. Papers will be selected on the basis of an extended abstract of 5 to 8 pages. Exceptional papers and posters presented by a student and with a student lead author will be considered for best student paper and best student poster respectively, with travel awards. Tutorials can come from a variety of areas but should provide practical information and/or training for the clustered computing community. For detailed information on submitting papers, presentations, or tutorials, please see the conference website.
Berkeley Entrepreneurs Forum to Discuss Future of Moore’s Law
For the last 40 years, Moore’s law, which states that transistor density will double every two years, has provided the framework for innovation in the semiconductor industry. But what will the next 10 years bring for this benchmark of the semiconductor industry and for the participants? And what will be the impact? What does the entrepreneur need to do to be successful in the new environment? And is there a role for government and/or venture capital in this industry?
The Berkeley Entrepreneurs Forum will present a panel discussion on “Moore’s Law: 40 Years Later” from 6:30 to 8:30 pm Thursday, October 22, in the Andersen Auditorium of the Haas School of Business on campus. Moderated by Laura Oliphant, Director of Intel Capital, the panel will include Alexei Andreev of Harris & Harris Group; Gene Meieran, Intel Fellow (retired); and Bruce C. Rhine of Nanometrics.
Logicomix and Autumnal Music This Week at MSRI
The Mathematical Sciences Research Institute (MSRI), in collaboration with the UC Berkeley Departments of Mathematics and Computer Science, will present “Logicomix: The life, loves, and logical struggles of Bertrand Russell and the other great logicians who wrestled with the elusive goal of placing mathematical reasoning on a firm foundation” today (Monday, October 19) at 5:15 pm in Room 100 of the UC Berkeley Genetics and Plant Biology Bldg. (located near the West Gate to campus and close to the intersection of University Avenue and Oxford Street).
The program will be a discussion of the graphic novel Logicomix: An Epic Search for Truth, by Apostolos Doxiadis and Christos H. Papadimitriou, with art by Alecos Papadatos and Annie Di Donna. The panelists will be Christos Papadimitriou, C. Lester Hogan Professor of Computer Science, co-author of Logicomix and of Elements of the Theory of Computation, and author of Turing (A Novel about Computation); Paolo Mancosu, Professor of Philosophy, author of From Brouwer to Hilbert: The Debate on the Foundations of Mathematics in the 1920s; and moderator Robert Bryant, Professor of Mathematics and Director of the Mathematical Sciences Research Institute.
MSRI will also present “A Little Bit of Autumnal Music,” a free concert by women of the UC Chamber Chorus and Marika Kuzma, Director, on Friday, October 23, at 5:30 pm in MSRI’s Simons Auditorium. The program includes Brahms’ Opus 17 for women’s chorus, harp, and horns; plus selections from the German lieder repertoire. The four gorgeous Romantic part-songs of Opus 17 are rarely heard or recorded, so this will be a treat. Brahms chose the horns to evoke a feeling of the forest and the harp to evoke wind and water, and they evoke a pastoral autumn setting.
This Week’s Computing Sciences Seminars
Stochastic Eulerian-Lagrangian Methods for Fluid-Structure Interactions with Thermal Fluctuations
Monday, October 19, 11:00 am–12:00 pm, 50F-1647
Paul Atzberger, UCSB Mathematics
A modeling and simulation formalism is presented for the study of soft materials. The formalism takes into account microstructure elasticity, hydrodynamic interactions, and thermal fluctuations. As a specific motivation we consider lipid bilayer membranes and polymeric fluids. The approach couples a Lagrangian description of the microstructures (lipid molecules / polymers) with an Eulerian description of the hydrodynamics. Thermal fluctuations are incorporated in the formalism by an appropriate stochastic forcing of the resulting equations in accordance with the principles of statistical mechanics. The overall approach extends previous work on the Stochastic Immersed Boundary Method. Simulation studies are presented showing applications of the methodology in the study of lipid flow in bilayer membranes, the shear viscosity of polymer fluids and lipid structures, and the diffusivity of particles in complex fluids.
Nonlinear Tracking Filters, and Topics on Queueing Theory and PT Distributions
Monday, October 19, 1:30–2:30 pm, 50F-1647
Atef Isaac, Agilent Technologies
1) Closely spaced targets not be detected and resolved by today’s radar technology. We will explore statistical estimation techniques based on the maximum likelihood criterion and Gibbs sampling, and addresses concerns about the accuracy of the measurements delivered thereby. A nonlinear particle filter will also be introduced to approximate the targets’ states’ conditional pdf, bypassing the measurement extraction stage, and operating directly on the radar’s sum/difference data, i.e., without measurement extraction. With successful tracking of those targets being achieved with the aid of nonlinear particle filters, the problem of detecting a target spawn will be tackled. Particle filters will be employed as nonlinear tracking filters to approximate the posterior probability densities of the targets’ states under different hypotheses of the number of targets, which in turn can be used to evaluate the likelihood ratio between two different hypotheses at subsequent time steps.
Ultimately, a quickest detection procedure based on sequential processing of the likelihood ratios will be used to decide on a change in the underlying target model as an indication of a newly spawning target. Radar signal processing, data association and target tracking are handled simultaneously.
2) An overview of a linear algebraic approach to queueing theory, modeling of subsystems comprising exponential servers, and the problems associated with power-tailed distributions.
Proofs of Retrievability: Toward RAID in the Cloud
Thursday, Oct. 22, 1:00–2:00 pm, Soda Hall, Wozniak Lounge, UC Berkeley
Ari Juels, RSA Laboratories
With the rapid migration of digital resources into the cloud, users are losing many of the traditional assurances of storage reliability that come with platform ownership and control. In this talk we will discuss a concept called a Proof of Retrievability, POR, an integrity-checking protocol for cloud storage. With communication of just some tens of bytes, a POR enables a client to achieve high-assurance verification of the integrity and availability of an arbitrarily large file in the cloud. In a distributed setting, PORs enable a single entity to amalgamate a collection of low-reliability storage providers into a high-reliability storage-system abstraction, even in the face of malicious provider behavior. The result is, loosely speaking, an analog of RAID (Redundant Array of Inexpensive Disks) in the cloud. Joint work with Kevin Bowers and Alina Oprea, to appear in ACM CCS 09 and ACM CCSW 09.
The Automatic Construction and Solution of a Partial Differential Equation from the Strong Form
Friday, Oct. 23, 9:30–10:30 am, 50F-1647
Joseph Young, Universitetet i Bergen, Bergen, Norway
In the last ten years, there has been significant improvement and growth in tools that aid the development of finite element methods for solving partial differential equations. These tools assist the user in transforming a weak form of a differential equation into a computable solution. Despite these advancements, solving a differential equation remains challenging. Not only are there many possible weak forms for a particular problem, but the most accurate or most efficient form depends on the problem’s structure. Requiring a user to generate a weak form by hand creates a significant hurdle for someone who understands a model, but does not know how to solve it.
We present a new algorithm that finds the solution of a partial differential equation when modeled in its strong form. We accomplish this by reformulating the equation as an optimization problem and then symbolically manipulating the result until we produce a polynomial program. After describing our algorithm, we validate our results by presenting a numerical example.
Link of the Week: Green Flash under the Golden Gate Bridge
David Donofrio of CRD, who works on the hardware design of Green Flash — the prototype ultra-high-resolution climate modeling computer — has found a postcard-worthy image of an atmospheric green flash under the Golden Gate Bridge. The still image is a single frame extracted from a short video clip shot by Paul Kamen, a Berkeley naval architect and yacht racing expert.
About Computing Sciences at Berkeley Lab
The Lawrence Berkeley National Laboratory (Berkeley Lab) Computing Sciences organization provides the computing and networking resources and expertise critical to advancing the Department of Energy's research missions: developing new energy sources, improving energy efficiency, developing new materials and increasing our understanding of ourselves, our world and our universe.
ESnet, the Energy Sciences Network, provides the high-bandwidth, reliable connections that link scientists at 40 DOE research sites to each other and to experimental facilities and supercomputing centers around the country. The National Energy Research Scientific Computing Center (NERSC) powers the discoveries of 6,000 scientists at national laboratories and universities, including those at Berkeley Lab's Computational Research Division (CRD). CRD conducts research and development in mathematical modeling and simulation, algorithm design, data storage, management and analysis, computer system architecture and high-performance software implementation. NERSC and ESnet are DOE Office of Science User Facilities.
Lawrence Berkeley National Laboratory addresses the world's most urgent scientific challenges by advancing sustainable energy, protecting human health, creating new materials, and revealing the origin and fate of the universe. Founded in 1931, Berkeley Lab's scientific expertise has been recognized with 13 Nobel prizes. The University of California manages Berkeley Lab for the DOE’s Office of Science.
DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.