InTheLoop | 04.13.2015
CS, Lab Divisions Host Students for Job Shadow Day
In what has become an annual pilgrimage, 17 Albany High School students interested in science, engineering and other career fields spent a few hours talking with Berkeley Lab mentors, including eight in Computing Sciences. The visits are part of the school’s annual Job Shadow Day in which students in their junior year are paired with mentors around the Bay Area. The lab has hosted 67 students over the past seven years.
Lab staff who volunteered as mentors were Terry Ligocki, Maciek Haranczyk, Keith Beattie, Dan Martin and Craig Tull of CRD; David Skinner and Annette Greiner of NERSC; Marcia Ocon Leimer of Computing Sciences Human Resources; Roy Kaltschmidt and Jon Weiner of Public Affairs, Andrew Lambert and Daniel Cheng of Engineering; Dave Cota of Facilities; Anne Ruminski of Materials Sciences Division; and Kathleen Bjornstad of the Earth Sciences Division. »Read more.
HPC Broader Community Engagement Workshop: Sharpening Research Presentation Skills
Last week, participants from underrepresented groups in high performance computing (HPC) attended a two-day workshop at Berkeley Lab to improve their research presentation skills in preparation for SC15.
The workshop's aim was to involve more people from underrepresented groups in the main technical track of the annual supercomputing conference by providing training and assistance for submitting to the technical program poster session. It covered training and discussion on organizing research results, taking advantage of SC resources and tips on successful poster submission and presentation. In working sessions, participants prepared and received feedback on draft submissions with the aim of having a draft poster submission that they can finalize and submit to SC15. While the workshop doesn't guarantee success, it aims to strengthen participants' chances of winning a place in the highly competitive SC technical program.
The workshop was organized by Tony Drummond of the Computational Research Division (CRD) and Mary Ann Leung from the Sustainable Horizons Institute. Berkeley Lab Deputy Director Horst Simon delivered a keynote presentation on the state and application of HPC. CRD's David Brown, Jonathan Carter, Silvia Criveli, Osni Marques, Khaled Ibrahim and Dan Martin helped judge posters during an exercise presentation. Also participating in the judging was Krishna Muriki from lab IT-HPC Services and Larisse Voufo from Google. »View the photo gallery.
CRD's Peisert Chairs IEEE Symposium on Security and Privacy
Sean Peisert of the Computational Research Division is serving as general chair of the 36th IEEE Symposium on Security and Privacy, May 18-20, 2015 at the Fairmont Hotel in San Jose, California. Early bird registration closes Friday, April 17.
Since 1980, the IEEE Symposium on Security and Privacy has been the premier forum for presenting developments in computer security and electronic privacy, and for bringing together researchers and practitioners in the field. This year's program committee has selected 55 research papers covering a wide range of topics An estimated 500 attendees from around the world are expected.
As in past years, there will be a poster session, a short talks session, and several workshops that will take place alongside the symposium, including workshops on privacy engineering, genomic security and privacy, language theoretic security, mobile security, and Web 2.0 security and privacy. Also at the symposium is a NITRD panel, featuring panelists from NSF, DHS S&T and other government agencies involved in creating the 2015 Federal Cybersecurity R&D Strategic Plan, as well as several "Birds of a Feather" sessions including discussions on network integrity, the Federal Cybersecurity plan and privacy in affective computing.
Public Affairs Offers Media Training on April 22
Berkeley Lab’s Office of Public Affairs is offering a series of onsite media training sessions. The two-hour session is ideal for researchers and others who may need to speak to the news media or the public. The program includes a brief primer on working with the news media and why it’s important to properly communicate science in these challenging budget times, and will offer an opportunity to practice delivering your messages on-camera. Each session is limited to six participants so early registration is suggested. Former CBS News producer, and current Lab communications manager, Jon Weiner will lead this free class. The next session will be held from 1:30 to 3:30 pm on Wednesday April 22. For more information, or to sign up, email firstname.lastname@example.org.
This Week's CS Seminars
Applying Machine Learning for Online Communication Research
Monday, April 13, 2015, 10:30am - 11:30am, Bldg. 50B, Room 4205
Michael Brooks, University of Washington
Studying online text-based communication can yield valuable insights for improving group work, online social interactions, and collaborative software. Qualitative research methods support deep, nuanced, and contextually sensitive interpretation of these data, but are labor-intensive and difficult to scale up to medium and large datasets. Machine learning techniques could help bridge the gap between focused qualitative analysis and broader, large-scale analysis. In this talk, I discuss the development of ALOE, a machine learning tool that was created for a project studying the social dynamics of affect and emotion among members of a distributed online collaboration. Treating this project as a case study, I will outline decisions we made during the design, implementation, and validation of ALOE, and discuss challenges and opportunities for widespread adoption of machine learning in qualitative research projects studying online text-based communication.
UCB Colloquium: Unraveling the Quantum Ensemble
Monday, April 13, 4:15pm - 5:15pm, Room #1, LeConte Hall - UCB Campus
Irfan Siddiqi, University of California, Berkeley
In the quantum world, an object can simultaneously exist in multiple states – the “dead” and “alive” character of Schrödinger’s proverbial feline being a quintessential example. It is the act of measurement which drives such an exotic superposition to a more familiar classical outcome, “dead” or “alive” for the cat, thus bridging the gap between quantum mechanics and our concept of reality. However, the precise nature of this so-called wavefunction collapse remains a topic of debate at the intersection of physics, mathematics, and philosophy. In particular, quantum mechanics does not in general predict the specific result of a single experiment, but rather defines the probability distribution associated with different measurement outcomes given an ensemble of identical systems or repetitive sequences. Applying continuous weak measurement techniques in conjunction with Bayesian statistics to superconducting quantum circuits, we reconstruct the real-time collapse of the wavefunction describing a two-state system at the level of the individual constituent quantum trajectories that form an ensemble measurement. With this dense dataset, a variety of statistical metrics can be extracted, including the most probable path—analogous to the geodesic in space-time—between two points in Hilbert space. Further, we have applied these weak measurement protocols to stabilize coherent oscillations using quantum feedback, and to generate entanglement between remote objects. Future extensions to many-body quantum systems may promise a route to efficiently probe systems with exponentially increasing complexity, generating natural parallels with questions in high energy physics.
Building Markov State Models for Periodically Driven Non-Equilibrium Systems
Thursday, April 16, 10:00am - 11:00am, Bldg. 50F, Room 1647
Han Wang, CAEP Software Center for High Performance Numerical Simulation, Beijing, China
Biomolecular systems under non-equilibrium conditions like external fields have attracted increased interest recently. In this talk, we firstly introduce a non-equilibrium molecular dynamics (NEMD) simulation method that samples the non-equilibrium system in a statistically reliable and thermodynamically consistent way. Then we demonstrate the way to build the Markov state model (MSM) for the non-equilibrium systems under a periodic external driving. The MSM reproduces the essential long timescale conformational dynamics of the original system, and at the same time it does not require the ergodicity of each single molecular trajectory. Therefore, it does not only describes, but also provides more insightful understanding of the non-equilibrium system than the standard analyses of the molecular trajectories. Our methods are tested and validated by an example non-equilibrium system: The alanine dipeptide under periodic electric field.
Electromagnetic scattering by gratings and random rough surfaces: implementation of high performance algorithms for solving eigenvalues problems
Thursday, April 16, 11:00am - 12:00pm, Bldg. 50F, Room 1647
Cihui Pan, Universite de Versailles Saint- Quentin-en-Yvelines (UVSQ)
In this talk, we consider the physical and numerical aspects of the electromagnetic scattering by grating and rough surfaces which has a wild range of applications in optics, material sciences, communications, oceanography and remote sensing. The curvilinear coordinate method (C-method) is an exact method for this physical problem. The C-method is based on Maxwell’s equations under covariant form written in a non-orthogonal coordinate system and this method leads to an eigenvalue problem of the scattering matrix. All the eigenvalues and eigenvectors of this complex, nonsymmetric and dense scattering matrix are required and this leads our research on the global eigensolvers.
For gratings, we developed a new version of C-method which leads to a differential system with initial conditions instead of the eigenvalue problem. This new version of C-method can be used to study multilayer gratings with homogeneous medium or inhomogeneous medium.
For rough surfaces, we present the parallel QR algorithm that is specifically designed for the C-method, where we take advantage of Rayleigh expansion and define the “early shifts”. The multi-window bulge chasing and aggressive early deflation are also used. The implementation shows a significant speed up. To improve the scalability, we conduct research on the “Spectral projection method as a global eigensolver” where we combine the multiple implicitly restarted Arnoldi method with nest subspaces (MIRAMns) and the contour integral based projection method (SSM) to form a global eigensolver. This global eigensolver has a parallel nature and good scalability. We show this global eigensolver has a multilevel parallel nature and a very good potential scalability. We present some experiments to validate this approach.
Computational Modeling of Heavy-Element Compounds: From Theory to Applications
Thursday, April 16, 4:00pm - 5:00pm, Bldg. 70A, Room 3377
Hanshi Hu, William R. Wiley Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory
Understanding the chemical bonding is a central theme in chemical science. Due to complicated electron correlation effects and strong relativistic effects (including spin-orbit coupling), the bonding and electronic structures of heavy-element compounds are often difficult to interpret. Advanced relativistic quantum chemistry approaches are needed to predict the properties of heavy-element systems. In this talk, I will present my research in ab initio coupled-cluster methodology development and applications of first-principles methods to the chemical bonding, electronic structures, and spectroscopic properties of actinide compounds and other systems.
Recently, we employed relativistic quantum chemistry including both density functional theory (DFT) and ab inito methods to investigate the mono-centered actinide compounds by collaborating with experimental spectroscopy technologies. It is found that quadruple bond is viable for carbon in a triatomic uranium carbide oxide molecule CUO, in addition to the series of triply bonded HC≡UX3, doubly bonded H2C=UXY compounds (X, Y=F, Cl, Br) in our earlier work. On the other hand, instead of forming multiple bonds, the silicon prefers to form multiradical bond with actinides for the analogue reaction of which the products are the bridged molecules Si(μ-X)AnF3 (An=U, Th; X = H, F) that form rare triplet silylenes. Furthermore, we have found that the different oxidation states of uranium lead to many intriguing properties. For the compound of UO6, uranium is best described as UVI coordinated by oxygen atoms in various oxidation states as oxo O2−, oxido O−, peroxide O22−, and superoxido O22−. For lower oxidation-state uranium molecules such as UFx− (x=2−4) and UO2−, Multi-electron transitions are observed as a result of strong electron correlation effects originated from their complicated electronic configuration. Meanwhile, it is found that Th(L)34+ complex possesses C3 symmetry with the ThIV coordinated by nine oxygen atoms from three ligands (TMOGA), which forms a chiral twisted TPP geometry. For AnO22+ (An=UVI, NpVI, PuVI) cations, their actinide metal centers are coordinated by six oxygen atoms from two ligands (TMOGA). These results will shed on lights on the mechanism of extraction for actinide partitioning from nuclear waste.