InTheLoop | 01.28.2013
The Weekly Newsletter of Berkeley Lab Computing Sciences
January 28, 2013
ESnet, Globus Online, and IU Demonstrate More Reliable Data Transfers
A collaboration between ESnet and Globus Online, together with Indiana University (IU), is already producing proof-of-concept demonstrations of moving large scientific datasets. As a lead-up to TIP2013, an international networking conference held from Jan. 13–18 in Hawaii, experts from ESnet, Globus Online, and IU demonstrated the integration of three critical data transfer technologies.
In a demo conducted in early January, the team was able to move one terabyte of data using a dynamically configured 5 gigabit per second (Gbps) OSCARS circuit between Argonne National Lab and Lawrence Berkeley National Lab in about 22 minutes. As a proof of concept, the demo achieved impressive transfer rates of just over 6 gigabits-per-second — highlighting how the integration of these technologies can greatly improve scientific productivity. The demo was conducted using ESnet-operated test GridFTP servers which are available for community testing. Read more.
Horst Simon Chats with Science, Michael Wehner Quoted in Chronicle
On January 24, Science magazine held a live chat on the future of supercomputing and included Berkeley Lab’s Deputy Director Horst Simon.
CRD climate scientist Michael Wehner is quoted in a front-page story on climate change in today’s (January 28) San Francisco Chronicle.
CRD’s Daniel Burke Elevated to IEEE Senior Member
Dan Burke, who joined the Computational Research Division last fall as a project manager for the new Computer Architecture Lab, has been elevated to the grade of Senior Member of the IEEE this year. Senior Member is the highest professional grade of the IEEE for which a member may apply, and only about 8 percent of IEEE’s 416,000 members have achieved this level.
At the lab, Burke is exploring low-energy approaches for the Department of Energy’s Exascale Computing Initiative. Although he only joined the Lab last October, since 2006 Burke has collaborated with John Shalf and other CRD researchers on a number of projects, including RAMP and Green Flash, which evolved into the Exascale Computing Initiative.
To be eligible for Senior Member status, IEEE members must be engineers, scientists, educators, technical executives, or originators in IEEE-designated fields with at least 10 years experience; have experience reflecting professional maturity; and show significant performance over a period of at least five of their years in professional practice.
Lab Pedestrian/Traffic Safety Blog Prompts Lively Discussions
In a little over a week, a new Berkeley Lab safety blog on pedestrian, bicycle, and traffic safety has already generated over a dozen ideas and suggestions. You can add your comments here (LDAP login required), or email your concerns to firstname.lastname@example.org.
This Week’s Computing Sciences Seminars
Structure Prediction of Elemental and Binary Systems at Ambient and High Pressure on the Ab-Initio Level
Monday, January 28, 10:00–11:00 am, 50F-1647
Aniket Kulkarni, Max-Planck-Institute for Solid State Research, Stuttgart, Germany
In recent years, new theoretical methodologies and techniques have become available to explore the energy landscape of chemical systems. Furthermore, experimental solid state chemistry has opened new opportunities by advances in controlling synthesis routes, for example by low-temperature atom beam de-position or physical vapor deposition and advancements in measurement techniques such as in-situ measurements and various kinds of microscopy techniques at low temperature. Still, it is quite difficult to understand structural changes at standard and extreme conditions. Often, it is impossible to understand structural changes without assistance from theoretical simulations. Sometimes it is advantageous to predict structures before the actual synthesis, because it can give more insight for deciding upon the synthesis route, and by knowing the structure, one can simulate its physical properties. Due to the knowledge of the system and its interesting properties beforehand, one can synthesize the chemical compound and validate the predictions. Crystal structure prediction methods are mainly based on the fact that stable and metastable modifications of chemical systems correspond to locally ergodic regions of the energy landscape. At low temperature, individual minima can be locally ergodic, if they are separated by sufficiently high barriers. Usually, there may exist many such locally ergodic regions encompassing one or several local minima. We are also interested in thermodynamic and dynamical stability of newly predicted modifications. In present research work deals with theoretical structure prediction of an elemental system, lithium metal, and binary systems such as calcium carbide and (per) nitride compounds using different ab-initio methods.
Julia: A High-Performance Dynamic Programming Language for Technical Computing
Tuesday, January 29, 4:00–5:00 pm, 891 Evans Hall, UC Berkeley
Alan Edelman, Massachusetts Institute of Technology
Julia is a high-level, high-performance dynamic programming language for technical computing. It provides a sophisticated compiler, distributed parallel execution, numerical accuracy, and an extensive mathematical function library. The library, largely written in Julia itself, also integrates mature, best-of-breed C and Fortran libraries for linear algebra, random number generation, signal processing, and string processing. Julia programs are typically organized around multiple dispatch; by defining functions and overloading them for different combinations of argument types, which can also be user-defined. Julia has a rapidly growing community comprising of academics, professionals and students, with developers contributing a number of external packages through Julia’s built-in package manager. Julia is open source and available under the MIT license.
This is joint work with Viral B. Shah, Jeff Bezanson, and Stefan Karpinski.
A Computational Model of Cardiovascular Hemodynamics
Wednesday, January 30, 10:00–11:00 am, 50F-1647
Amanda Peters Randles, Harvard University
Accurate and reliable modeling of cardiovascular hemodynamics has the potential to improve understanding of the localization and progression of heart diseases, which are currently the most common cause of death in Western countries. However, building a detailed, realistic model of human blood flow is a formidable mathematical and computational challenge. The simulation must combine the motion of the fluid, the intricate geometry of the blood vessels, continual changes in flow and pressure driven by the heartbeat, and the behavior of suspended bodies such as red blood cells. Such simulations can provide insight into factors like endothelial shear stress that act as triggers for the complex biomechanical events that can lead to atherosclerotic pathologies. Currently, it is not possible to measure endothelial shear stress in vivo, making these simulations a crucial component to understanding and potentially predicting the progression of cardiovascular disease. In this talk, I will present and examine our approach for modeling the cardiovascular hemodynamics while accounting for the additional force from the expansion and contraction of the heart. I will discuss the performance of our application on the IBM Blue Gene/P and Blue Gene/Q architectures and the methods leveraged to enable efficient modeling of the fluid movement in real-patient geometries. I will also discuss future work to enable the simulation of larger physical systems for longer time periods using three methods that rely on the interplay of coarse and fine models of the system: adaptive mesh refinement (AMR), adaptive mesh and algorithm refinement (AMAR), and parallelization in time.
CITRIS Research Exchange: Big Data and Democracy Initiative —
The Future of Cooperation
Wednesday, Jan. 30, 12:00–1:00 pm, 310 Sutardja Dai Hall (Banatao Auditorium), UC Berkeley
Tim O’Reilly, CEO, O’Reilly Media
Some of the biggest changes sweeping the technology world today are new forms of network and computer-enabled cooperation. It was easy enough to see this pattern in open source software development or Wikipedia, a bit more challenging to see how it powered Web 2.0 giants like Google and Amazon, but it gets really interesting when you are able to see how new kinds of man-machine symbiosis and networked cooperation are at the heart of projects like the Google self-driving car, the reinvention of retail by Apple and Square, transportation services from RelayRides to Uber, and even in new models for networked government.
Tim O’Reilly is the founder of O’Reilly Media and a supporter of the free software and open source movements.
Registration (free) is required by today (Monday) at 3:00 pm. A boxed lunch will be provided. Live broadcast at mms://media.citris.berkeley.edu/webcast. Ask questions live on Twitter: #CITRISRE. The talk may be viewed post-event at http://www.youtube.com/citrisuc.
Standing Waves and Stability of a Viscoelastic Cylinder
Wednesday, January 30, 4:00–5:00 pm, 939 Evans Hall, UC Berkeley
Trevor Potter, University of California, Berkeley
Oden and Lin  discovered that an elastic cylinder, when rotated at certain speeds, displays standing wave solutions that bifurcate from the axisymmetric solution. They noted a hierarchy of N-bump standing wave solutions, where the rotation speed decreases to a critical value as N goes to infinity. In the first part of my talk, I observe a more complex hierarchy of standing wave solutions using numerical eigenvalue software. If a viscoelastic model is added, the N-bump standing waves disappear and become decaying modes. In the second part of my talk, I observe that a commonly used viscoelastic model can lead to nonphysical blow-up of the cylinder at high speeds. The blow-up is shown not to occur for a fully nonlinear viscoelastic model based on the 2nd Law of Thermodynamics.
EECS Colloquium: Revisiting the Design of Database Systems, Inside and Out
Wednesday, January 30, 4:00–5:00 pm, 306 Soda Hall (HP Auditorium), UC Berkeley
Sam Madden, MIT
Until recently, all commercial relational databases were based on a software architecture and programming interface designed at Berkeley and IBM 1970s and 80s. Recently that has changed.
In this talk, I’ll discuss two new important developments in the world of data management on which my research has focused. In the first, I’ll describe the design of a new type of “column-oriented” database for so called “analytical workloads” where queries scan and perform aggregation (“reduce”) operations over a large fraction of the data. The basic idea, known as “vertical partitioning,” has been used before, but, by building a radically different database architecture around this concept, we showed that it is possible to get order-of-magnitude performance gains over the systems designed in the 70s and 80s. These ideas have since been adopted by most commercial database engines.
One outcome of the early database designs was that they led to a separation between “application logic” and “database logic,” with declarative database commands embedded a (typically) imperative host language. In the second part of the talk, I’ll describe a new line of work looking at how program analysis techniques can be used to make the boundary between application and database logic much more fluid. Specifically, I’ll describe techniques that make it possible to automatically transform the boundary between application logic written in an imperative host language to declarative SQL logic, and programmatically adjust the placement of application logic on a database server or an application server.
Simulating and Analyzing the First Stars in the Universe
Thursday, January 31, 10:00–11:00 am, 50F-1647
Matthew Turk, Columbia University
The first stars in the Universe formed out of primordial, metal-free gas that collapsed into the wells of dark matter halos. The masses of these stars as they enter the main sequence will sensitively govern their end product, and thus the pattern of enrichment in their surroundings. In this talk, I will present on efforts to understand the chemical, magnetic and hydrodynamic properties of the clouds out of which these stars form, including how these properties may influence the initial mass function of the first stars. I will also present the simulation and analysis frameworks that have been developed to study and understand their formation. I will discuss the challenges of analysis of next generation simulations, and the community-driven development of the analysis framework yt to address these challenges. I will conclude by briefly describing recent applications of yt to non-astrophysical domains such as seismology.
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.