A-Z Index | Phone Book | Careers

InTheLoop 11.02.2015

November 2, 2015

‘Sidecars’ Lead the Way to Concurrent Analysis of Large-Scale Simulations

A new software tool developed through a multi-disciplinary collaboration at Berkeley Lab allows researchers doing large-scale simulations at the National Energy Research Scientific Computing Center (NERSC) and other supercomputing facilities to perform data analytics and visualizations of their simulations while the simulations are running.

The new capability will give scientists working in cosmology, astrophysics, subsurface flow, combustion research and a variety of other fields a more efficient way to manage and analyze the increasingly large datasets generated by their simulations. It will also help improve their scientific workflows without burning precious additional computing hours.

The team is initially testing its approach on Berkeley Lab’s Nyx code coupled with a new topological halo finder called Reeber. Reeber was developed by Gunther Weber and Dmitriy Morozov, computational researchers in the lab’s Data Analysis and Visualization Group. Nyx is a large-scale cosmological simulation code based on the BoxLib framework developed by researchers in Berkeley Lab's Center for Computational Sciences and Engineering. Because the sidecar approach is managed by the BoxLib, all other codes based on BoxLib will be able to take advantage of this same capability with relatively little additional effort. »Read more.

Bautista One of 2015’s 100 Most Influential Filipinas

Elizabeth Bautista, head of NERSC’s Operations Technology Group, was selected as one of the 100 Most Influential Filipina Women in the World (Global FWN100) by the Filipina Women’s Network. The award was announced as part of the Filipina Leadership Global Summit held October 29-31, 2015 in San Francisco.

The Global FWN100 Award recognizes 100 women of Philippine ancestry who are changing the face of leadership in the global workplace. Awardees are recognized based on their outstanding work in their respective fields, their achievements and contributions to society and mentoring.

At NERSC, Bautista manages a paid internship program for students, conducts outreach to high schools and colleges and collaborates with her partners in national labs and universities to introduce students to computational science as a career. She also collaborates with diversity-in-technology groups in schools and communities to create a pipeline for an inclusive workforce program. Additionally, she serves as chair of Filipinas in Computing, an Anita Borg Systers Community. »Read more.

Peisert Compiles 2nd Workshop Report on Securing Scientific Computing Integrity

Sean Peisert of CRD’s Integrated Data Frameworks Group has published a report from the DOE workshop on ASCR Cybersecurity for Scientific Computing Integrity—Research Pathways and Ideas. Peisert co-chaired the workshop held, June 2-3 in Gaithersburg, Maryland.
 
The goal of this workshop was to define a long-term 10 to 20 year fundamental basic research and development strategy and roadmap regarding scientific computing integrity facing future high performance computing (HPC) and scientific user facilitates. Specifically, this report describes potential research paths in three central areas: Trustworthy Supercomputing, Extreme-Scale Data, Knowledge, and Analytics for Understanding and Improving Cybersecurity, and Trust within High-end Networking and Data Centers.
 
This report builds on the findings of a previous ASCR Cybersecurity workshop—also co-chaired by Peisert—held January 7–9, 2015, in Rockville, MD, to examine computer security research gaps and approaches for assuring scientific computing integrity specific to the mission of the DOE Office of Science. »Read the report.

Attending SC15? Consider Taking a Turn Staffing the DOE Booth

Planning for the joint DOE booth at SC15 is moving right along and the planning committee is now looking for lab folks to help staff the booth during the exhibit hours. Volunteers are needed to steer visitors to electronic posters, job lists, demos and discussions, as well as to talk about their work in general. Fourteen labs are represented at the booth, so visitors really need some guidance to see the best material.

If you could spare a couple of hours on Tuesday, Wednesday or Thursday, you can help the lab take advantage of a good opportunity to tell our story. »Sign up online (Scroll down to the schedule of events to see available times under "booth duty" and then indicate the days and times you will be available.) Volunteers will be given a brief overview of the booth and materials for guidance. Questions? Contact Jon Bashor, jbashor@lbl.gov.

Bevatron Lot Parking Impacted Tomorrow

Tomorrow, Tuesday, Nov. 3, about 100 spaces will be blocked off in the Bevatron lot for the Bay Area Battery Summit. Those who normally park in the B1 lot are encouraged to use the D lot ("the pit") or alternative transportation.

This Week's CS Seminars

 

NMF and SVD: Two Cases of Parallelization

Monday, November 2, 2015, 1–2pm, Bldg. 50F, Room 1647
Marian Vajtersic, University of Salzburg, Austria and Mathematical Institute, Slovak Academy of Sciences, Bratislava, Slovakia

This talk will be devoted to the parallelization of two matrix factorizations: Nonnegative Matrix Factorization (NMF) and Singular Value Decomposition (SVD). Both of them arise in applications in which large matrices have to be represented in the form of smaller factors which are more suitable for further processing. One such example is when the matrix is large and has only nonnegative entries. The goal of NMF is to represent such matrices in an approximate way as the product of two significantly smaller matrices. NMF has the nice property that the resulting matrix factors are also nonnegative. Our approach to compute NMF is based on the Newton method, where in each iteration both approximate factors are computed in alternating manner. The other method is SVD. Its importance for computational practice is well known, and fast methods for SVD deserve serious attention nowadays. Our focus will be concentrated on a problem with large dense matrices. The method considered is the block Jacobi method, in both of its variants: the two- and one-sided ones. Due to the complexity of the problems, for both of these factorizations parallelism is inevitable, especially when the matrix size is large. We will show our original parallel solutions in more detail, discuss their properties, including experiments, and conclude with some remarks concerning further research.

Applied Math Seminar: The Rapidly Growing Zoo of Polytopal Finite Element Methods 

Wednesday, November 4, 3:30–4:30pm, 939 Evans Hall, UC Berkeley
Andrew Gillette, University of Arizona

Finite element methods are one of the most widely studied and broadly employed means for the numerical approximation of the solution to PDEs in practical application settings. While finite element methods over simplicial and cubical meshes are becoming increasingly standardized across disciplinary boundaries, there is a booming interest in the definition and implementation of finite element methods over unstructured meshes of generic polygons and polyhedra. In this talk, I will provide a brief overview of this "zoo" of methods (including virtual elements, weak Galerkin, and hybrid higher order) and will discuss some of my contributions in the arena of generalized barycentric coordinate methods.

Energy Efficient Control in Smart Buildings and Grids

Tuesday, November 3, 10–11am, Bldg. 50B, Room 1211 (AG Node)
Baris Aksanli, UC San Diego

Building energy consumption in the USA has two key components: commercial (36% of the total) and residential (38% of the total). Although the contribution of these two parts are similar, recent studies mainly focused on the commercial buildings. In this presentation, I first show on how residential energy management can be made "smart". We develop HomeSim, a residential energy simulator that makes it possible to investigate the impact of renewable energy, centralized vs. distributed in-home energy storage, and intelligent appliance rescheduling, lighting and HVAC. We also analyze how batteries in smart homes should be used to maximize their benefits. In the second part of my presentation, I focus on quantifying the effects of smart building controllers on the grid. We take UCSD Microgrid circuit structure as the baseline which hosts distributed smart buildings and energy generation. We build Smart Grid Swarm Simulator (S2Sim) to study the grid dynamics and demonstrate how distributed smart building controllers can threaten the stability of the grid.

Big Data Processing on Cloud Platforms

Thursday, November 5, 10–11am, Bldg. 50B, Room 1237 (NOC)
Mariam Kiran, University of Bradford, UK

Cloud computing has introduced new software paradigms with ‘everything-as-a-service’ (XaaS), allowing providers and consumers to consider multiple platforms to deploy services. As an emerging and leading technology it promises to deliver services which are reliable, secure, sustainable and secure for software, infrastructures and platforms. These advances have led to a change in traditional software engineering practices where software services are being designed in advance to cope with high demand, availability and security concerns. Much if the current work is being led by industrial initiatives in Big Data and Smart Cities, where techniques are being developed based on consumer demands.

In this talk, we focus on specific architecture using the design pattern lambda architecture for fast and cost driven big data analysis. Using network data sets as a case study, the approach can be extended to various sensor driven applications such as the ones used in smart city case studies. Presenting this architecture the talk will focus of demands in terms of costs, testing methods and how optimum architectures can be designed from a software engineering perspective and the challenges faced.

Special Applied Math Seminar: Semiclassical computational methods for quantum dynamics with band-crossing

Thursday, November 5, 2:30–3:30pm, 736 Evans Hall, UC Berkeley
Shi Jin, University of Wisconsin-Madison

Band-crossing is a quantum dynamical behavior that contributes to important physics and chemistry phenomena such as quantum tunneling, Berry co nnection, chemical reaction etc. In this talk, we will discuss several recent works in developing semiclassical methods for band-crossing, including examples from surface hopping, Schrodinger equation with periodic potentials, and high frequency solutions of linear hyperbolic systems with polarized waves. For such systems we will also introduce an "asymptotic-preserving" method that is accurate uniformly for all wave numbers.

BIDS Guest Lecture: Spectacular Algorithms

Thursday, November 5, 4–5:30pm, 470 Stephens Hall, UC Berkeley
Malte Ziewitz, Cornell University

Algorithms have developed into somewhat of a modern myth. The subject of media reports, research projects, and congressional hearings, they are increasingly portrayed as powerful yet inscrutable entities that govern, sort, shape, or otherwise control our lives. But what if their operations are not as straightforward as expected? What if it turns out that there isn’t much to hide? This talk will explore these questions through an ethnography of search engine optimization (SEO) consultants—a growing industry of marketing professionals that help their clients rank in search engine results pages. Moving back and forth between spreadsheets, software tools, client meetings, industry conferences, and online conversations, I shall explore the everyday practices of secrecy, publicity, and experimentation that make the engine work. What is often portrayed as a technical tool for ordering and evaluating information turns out to drive a mode of ordering based on ambiguous analytics, continuous provocation, and a self-perpetuating "need to know."

BIDS Data Science Lecture Series: Steps to Greatly Reducing the Number and Impact of Megafires

Friday, November 6, 1–2:30pm, 190 Doe Library, UC Berkeley
Carl Pennypacker, Berkeley Lab

Over the last decade, early detection of fires and very careful and well-managed responses to fires have become possible, although most improvements are still waiting to be implemented in a single system. Inspired by the Berkeley supernova searches, which I helped develop with Rich Muller and Saul Perlmutter, here, I suggest ways such technology might directly benefit Californians and others around the world. Developments in sensors, inexpensive basing of imagers in geosynchronous orbit, computational capabilities, growing databases of fire fuels and landscapes, telecommunication systems, unmanned aerial vehicles, and fire-simulation programs all are ready to be connected and work in synchrony. Such a system, which has been conceptually designed at UC Berkeley's Space Sciences Lab, Lawrence Berkeley National Lab, and the University of California San Diego's WIFIRE group, is called FUEGO: Fire Urgency Estimator on Geosynchronous Orbit. Our calculations indicate that small fires can be seen from space easily. Once a fire is lit, the growing fire holds the potential to then be managed and fought utilizing new the generation of databases, manned and unmanned sensors on airplanes and drones, telecommunication systems connecting all parties, and computer simulations running in real time to aid fire fighters in deploying their precious resources. Although there are many who are eager to help, the FUEGO system we describe in this talk seems to have gained acceptance by a number of key collaborators, and we are implementing aspects of this system now, beginning with tests of UAVs, manned aircraft, and imagers over prescribed burns, and we envision a data experiment in southern California this fall on real fires, pending funding. One of our funding pland is to team with UC Berkeley's crowd-funding system (see https://crowdfund.berkeley.edu/).