A-Z Index | Phone Book | Careers

InTheLoop | 08.06.2012

August 6, 2012

Researchers Create Carbon-Dioxide-Separating Polymer

Using supercomputers at NERSC, researchers from Haverford College have come up with a new type of two-dimensional polymer, PG-ES1, which allows, in theory, for highly efficient separation of carbon dioxide from the exhausts of power plants. Based on simulations, PG-ES1 is predicted to be more than 100 times as permeable to carbon dioxide as the best existing materials, while maintaining a rejection of nitrogen and methane gases that meets or exceeds the best existing materials. This allows it to act as a molecular filter that lets the carbon dioxide to pass through easily, while preventing other gases from escaping. Read more.

Twice-Stuffed Permafrost

Researchers from Pacific Northwest National Laboratory, using NERSC’s Hopper supercomputer, suggest that hydrates—water-gas compounds found in ocean permafrost—can provide energy and store it, too—and then trap carbon dioxide. Read more.

ASCR Will Soon Post Openings for Computer Science and Applied Math Program Managers

The Department of Energy, Office of Science, Office of Advanced Scientific Computing Research (ASCR), will be seeking a Senior Program Manager for Computer Science who will help to reinvent the field with respect to operating and runtime systems for extreme scale supercomputing. There will also be an opening for a Program Manager for Applied Mathematics. Both positions will be in the Computational Science Research & Partnerships (SciDAC) Division.
Neither position has been officially posted yet, but interested candidates should monitor the DOE Office of Science jobs site and begin gathering the required materials. Read more about the positions.

Bell Presents Two Talks at International Symposium on Combustion

John Bell of the Computational Research Division presented two talks at the 34th International Symposium on Combustion held in Warsaw, Poland, July 29–August 3, 2012. The first talk was part of the 13th International Workshop on Premixed Turbulent Flames and was titled “Simulation of Lean Premixed Low-Swirl Hydrogen Flames.” The second talk was part of the main symposium and was titled “Simulation of Nitrogen Emissions in a Premixed Hydrogen Flame Stabilized on a Low Swirl Burner.”

Winning CS Summer Student Posters

Students who have spent their summer working in Computing Sciences presented posters describing their work on Thursday, August 2, in Perseverance Hall. The three posters judged best poster, most innovative, and most original were:

  • Best Poster: “Preparing Next for NAND Flash: A Very Large-Scale Storage Class RRAM Memory System,” Myoungsoo Jung, Pennsylvania State University (supervisor: John Shalf, CRD FTG)

  • Most Innovative: “Fitting Polygons to Fluorescence Microscopy Images of Peptoid Nanosheets,” Davina Smith, Las Positas College (supervisors: Babak Sanii, Charles Verboom, IT, Molecular Foundry NanoBio floor)

  • Most Original: “CRT Building Digital Archive,” Nicholas Fong, University of California, Berkeley (supervisor: Elizabeth Bautista, NERSC Computer Operations and Network Support)

Reminder: Aug. 8 Town Hall Meeting on DOE’s Early Career Research Program

Employees interested in submitting proposals for DOE’s 2013 Early Career Research Program are strongly encouraged to attend a Town Hall meeting with Lab Director Paul Alivisatos from noon to 1:30 p.m. Wednesday, August 8, in the Bldg. 50 Auditorium. For more information on the program, visit the Lab’s Early Career Research Program website.

The Early Career Research Program, now in its fourth year, supports the development of individual research programs of outstanding scientists early in their careers. The Office of Science is inviting proposals for support under the Early Career Research Program in the following program areas: Advanced Scientific Computing Research (ASCR); Biological and Environmental Research (BER); Basic Energy Sciences (BES), Fusion Energy Sciences (FES); High Energy Physics (HEP), and Nuclear Physics (NP).

Preproposals are required and must be submitted by 5 p.m. Wednesday, September 6, 2012 (Eastern time). The preproposal should be created as a PDF file and submitted electronically through the DOE Office of Science Portfolio Analysis and Management System (PAMS) website. Read more.

David Bailey: Algebra Is Essential in a 21st Century Economy

A few days ago, Andrew Hacker, an author and former professor of political science at Queens College in New York City, created quite a stir with a New York Times op-ed entitled Is Algebra Necessary?, in which he argues that it is no longer necessary to expect the vast majority of K-12 students to study algebra, geometry, or calculus.

In their Huffington Post blog, David Bailey of Berkeley Lab’s Computational Research Division and Jonathan Borwein of the University of Newcastle, Australia, argue that algebra is still relevant, especially for disadvantaged students. And both authors wonder whether their careers in mathematics would have been possible if their schools had “tracked” them into mathematically less challenging courses. Read more.

This Week’s Computing Sciences Seminars

Charm++: An Asynchronous Parallel Programming Model with an Intelligent Adaptive Runtime System
Monday, August 6, 2:00–3:00 pm, 50F-1647
Laxmikant (Sanjay) Kale, Parallel Programming Laboratory, University of Illinois at Urbana-Champaign

Parallel programming challenges at the extreme scale include the need for strong scaling, support for composing multi-module (often multi-physics) codes, heterogeneity and load imbalances. I will describe how the “Migratable-Objects” execution model, embodied in Charm++ and AMPI, can address these issues. In this model, the program is expressed in terms of the work-units and data-units of the application without reference to which processor they are housed on. These units, which can be objects, user-level threads, or data blocks, can be migrated among processors under the control of an adaptive runtime system (RTS). The RTS adapts to variations in the application configurations/evolution, as well as to variations in the machine environment. The RTS (a) monitors multiple aspects of an ongoing execution, (b) takes corrective actions in order to optimize multiple criteria such as performance and power, and (c) effect recovery from component failures. Message-driven execution is an inherent part of the migratable-objects model, which leads to several performance and productivity benefits. In particular, it strongly supports parallel composition of multiple, independently developed modules without requiring partitioning processors or sequencing modules. Thus, the adaptive runtime also supports automatic fault tolerance, dynamic load balancing, minimization of contention by topology aware mapping, and communication optimizations, without significant efforts from the programmer.

AMPI (Adaptive MPI) is an implementation of MPI within the migratable-objects model, allowing MPI programs to avail of the asynchronous and dynamic capabilities of the model. In addition, Charm++/AMPI modules can interoperate with plain MPI modules as well.

The talk will be illustrated by examples from multiple CSE applications in active use, including NAMD (biophysics), ChaNGa (astronomy), OpenAtom (quantum chemistry), etc. Time permitting, I will also provide a brief overview of the work on higher-level languages supported by the same runtime system.

An Optimization-Based Strategy for Computational Design of Nanoporous Carbon-Zero Materials
Thursday, August 9, 2:00–3:00 pm, 15-253-CR(80)
Maciej Haranczyk, Computational Research Division

Eigenvalue Computations in the Context of Data-Sparse Approximations of Integral Operators
Friday, August 10, 1:00–2:00 pm, 50F-1647
Paulo Vasconcelos, Fulbright Fellow, University of Porto, Portugal

We consider the numerical solution of a large eigenvalue problem resulting from a finite rank discretization of an integral operator. We are interested in computing a few eigenpairs, with an iterative method, so a matrix representation that allows for fast matrix-vector products is required. Hierarchical matrices are appropriate for this setting, and also provide cheap LU decompositions required in the spectral transformation technique. We illustrate the use of freely available software tools to address the problem, in particular SLEPc for the eigensolvers and HLib for the construction of H-matrices.

We develop analytical expressions for the approximate degenerate kernels and deduce error upper bounds for these approximations. Numerical tests show the benefits of the data-sparse representation compared to standard storage schemes, in terms of computational cost as well as memory requirements.

Link of the Week: Supercomputer Learns How to Recognize Cats

For those of you who have been waiting for supercomputers to do something really useful, Google researchers have the answer: a 16,000-processor neural network that used 10 million YouTube video images to teach itself how to recognize cats. The key point is that the researchers did not tell the neural network to do that—it happened spontaneously, with no human supervision, and it’s considered a major step forward in machine learning. Read more.

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.