CRD’s Xiaoye ‘Sherry’ Li Named 2016 SIAM Fellow

Berkeley Lab’s Xiaoye “Sherry” Li has been named a 2016 SIAM Fellow. SIAM stands for the Society for Industrial and Applied Mathematics. Li, who currently heads the Scalable Solvers Group in Berkeley Lab’s Computational Research Division, is being recognized for advances in the development of fast and scalable sparse matrix algorithms and fostering their use in large-scale scientific and engineering applications. Her areas of impact include computational mathematics, linear algebra and matrix theory.

SIAM Fellows are designated each year to recognize members of the community for their distinguished contributions to the disciplines of applied mathematics, computational science and related fields. The Fellows Selection Committee selects Fellows based on nominations by SIAM members.

Forbes Article Quotes NERSC’s Skinner on HPC’s Role in Energy Innovation

A recent Forbes article focused on one a finding of the recently released Department of Energy Quadrennial Energy Review (PDF): High-performance computing is set to accelerate industrial energy innovations. The article quotes NERSC’s Strategic Partnerships Lead David Skinner, Kristin Persson, who heads the Materials Project, and Deputy Laboratory Director Horst Simon extensively.

ESnet Shares $150K Grant to Expand Workshops

A new $150,000 award from the National Science Foundation will allow networking experts from Indiana University and partners at the Department of Energy’s Energy Sciences Network (ESnet) and Internet2 to expand the successful Operating Innovative Networks workshop series. Since 2013, the workshops have helped network engineers learn to manage ever-increasing scientific data flows, hastening scientific discovery in the process.

NERSC Data Day in the Queue

Please consider responding to the NERSC Data Day survey fielded in preparation for the center’s first ever Data Day event to be held this summer. Featuring technical presentations, tutorials, a hackathon and more, the goal of the event is to foster in-depth discussions with data users at all levels of NERSC systems experience. The survey closes April 18.

HPC users are also invited to participate in an HPC community demographics survey from peer supercomputing center EPCC (Edinburgh Parallel Computer Centre, UK) and the Women in HPC Network. Direct all questions to the survey authors Athina Frantzana and Toni Collis.

CRD’s de Jong Talks About Modifying NWChem for Next-Gen Processors

Bert de Jong, lead for CRD’s Computational Chemistry, Materials and Climate Group, recently spoke to Scientific Computing about the work Lawrence Berkeley National Laboratory’s Intel Parallel Computing Center has done to optimize the popular NWChem code to take advantage of upcoming hardware advances, like the next-generation Intel Xeon Phi processors. He hopes that this work will shorten the time to scientific discovery, as well as allow the scientific community to pursue new frontiers in the fields of chemistry and materials modeling.

NWChem is a widely used open source software computational chemistry package that includes both quantum chemical and molecular dynamics functionality. The NWChem project started in the mid-1990s. The code was designed from the beginning to take advantage of parallel computer systems. The software aims to provide its users with computational chemistry tools that are scalable both in their ability to treat large scientific computational chemistry problems efficiently, and in their use of available parallel computing resources from high-performance parallel supercomputers to conventional workstation clusters.

David Patterson to Give Farewell Lecture & Symposium

On Friday, May 6 at 4 p.m., CRD’s David Patterson, who holds a joint appointment with Berkeley Lab and UC Berkeley, will deliver a farewell lecture at the International House on the campus of UC Berkeley to mark his retirement from teaching. The tongue-in-cheek title of his talk is “How to Be a Bad Professor.” It will be followed by a reception.

On Saturday, May 7, there will also be a one-day symposium with talks by colleagues and former students addressing the future of topics associated with Patterson’s 40 years at UC Berkeley, such as microprocessors, storage, cloud computing, data science and machine learning, also held at the International House.

The organizers request attendees register for both events. Those who sign up for the symposium by April 7 will receive a commemorative book with perspectives by Stanford President John Hennessy, National Medal of Science Winner Richard Tapia, former UCSC Chancellor Karl Pister, and other luminaries.

This Week’s CS Seminars

Wednesday, April 6

Applied Math Seminar
A Space-Time Discontinuous-Galerkin Spectral Element Method for Compressible Flows
3:30 – 4:30 p.m., 939 Evans Hall, UC Berkeley
Laslo Diosady, NASA Ames Research Center

High-Reynolds number separated flows involve large-scale unsteady motion which require scale-resolving computational fluid dynamics (CFD) simulations in order to provide accurate predictions. Higher-order methods have been shown to be more efficient for simulations requiring high spatial and temporal resolution, allowing for solutions with fewer degrees of freedom and lower computational cost than traditional second-order CFD methods. A higher-order space-time discontinuous-Galerkin(DG) method is presented for the scale-resolving simulation of high-Reynolds number compressible flows. The space-time DG method is implemented efficiently using a tensor-product formulation and solved with a Jacobian-free Newton-Krylov scheme and tensor-product based preconditioners. The effect of the discrete formulation on the nonlinear stability of the scheme is assessed through numerical simulations. With increasing Reynolds number, it is shown that an entropy-stable formulation is required in order to maintain stability at high-order. Using an entropy variable formulation consistent with established entropy stability theory ensures nonlinear stability at high and infinite Reynolds number. Numerical results are presented demonstrating the success of the method on a variety of subsonic compressible flows.

Friday, April 8

BIDS Data Science Lecture Series

Facilitating a Citizen Science Network to Monitor Mammals through Camera Trapping
1:10 – 2:20 p.m., 190 Doe Library, UC Berkeley
William McShea, Smithsonian Conservation Biology Institute

The Smithsonian Institution has created a data platform that allows for the management of volunteers to provide wildlife photographs and metadata that share a metadata structure, can be verified by experts, are curated by the Smithsonian, and can subsequently be shared across projects at a public website (emammal.org). The public website was launched in January, and the site has more than 25 active projects and 5 million images. Dr. Bill McShea will explain the dataflow and the utility of the site for science, conservation, education, and outreach.