InTheLoop | 09.06.2011
September 6, 2011
Small Particles, Big Impact
Using NERSC systems, scientists have found that the small effects of particles suspended in our atmosphere add up over time and can lead to big errors in climate prediction models. Known as aerosols, particles such as ozone, dust, and sea salt both scatter and absorb sunlight in different proportions depending on their type and elevation. Taken together, the effects of these small particles can tip the Earth’s energy balance toward heating or cooling. Read more.
Grab Your Binoculars to Glimpse a Cosmic Explosion!
Skywatchers — grab your binoculars and telescopes, and head for some clear dark skies to see the new supernova discovered near the Big Dipper by Peter Nugent, co-leader of Berkeley Lab’s Computational Cosmology Center, and collaborators with the Palomar Transient Factory.
At a mere 21 million light-years away from Earth, the supernova is appearing so bright that Earthlings may be able to see it with a good pair of binoculars over the next few weeks. Or swing by the Chabot Space & Science Center on Saturday, Sept. 10, to meet the astronomers that discovered the supernova, and see it for yourself through the museum’s giant, historic telescopes (weather permitting). The telescope viewing is free. Read more and see a video.
Faces at the Lab: Computational Scientist and Drummer Marcus Day
Many a youngster dreams of being in a rock band, but most give up that fantasy when real life hits: career, marriage, children, mortgages, etc. Not Marcus (Marc) Day of the Lab’s Computational Research Division. He manages all this and still finds time to perform and tour with the highly touted Johnny Hi-Fi band as a drummer. He’s been playing the drums since fourth grade on a range of projects, but none quite like Johnny Hi-Fi, which just released their seventh CD.
“In addition to making what I think is great, great music, we have had some awesome opportunities to travel and play in front of all kinds of diverse audiences,” he says. “Additionally, we’ve given several presentations called ‘Passion and Profession,’ typically directed at university kids with a message about balancing work and artistic passion in one’s life.”
Last Sunday Johnny Hi-Fi played on the main stage at the Oakland Pride celebration, and they will also perform at the Lab’s Open House on October 15. Other gigs this fall include Berkeley, Walnut Creek, and Napa. See a video.
SilverStripe CMS Training September 15 and 20
Quinn Interactive is leading two hands-on training sessions for using the new content management system (CMS) behind Computing Sciences’ web sites, called SilverStripe. This training will focus on the CRD site, but NERSC and ESnet folks who missed earlier trainings are welcome to attend on a space-available basis.
Although the CMS interface is somewhat intuitive, this training will help you understand and take best advantage of its features, including customizations created for the NERSC, ESnet, CRD shared code-base, such as the publications database.
We have about 20 seats available for 1:30–3:30 pm Sept. 15, and 10 seats for 10:00 am—noon Sept. 20. Both sessions are identical, so please choose the one that best fits your schedule. Please bring your own laptops.
This Week’s Computing Sciences Seminars
LAPACK Seminar: Finding Zeros of Single-Variable Real-Valued Functions
Wednesday, Sept. 7, 12:10–1:00 pm, Soda Hall, UC Berkeley
Gautam Wilkins, UC San Diego
Finding zeros of single-variable, real-valued functions is a common and basic task in scientific computing. The task of finding a zero of a function is in general twofold. First, we find an interval, [a,b], across which the function f reverses sign (also called a straddle). Second, we must reduce the size of the interval [a,b] until it is sufficiently small to bracket a zero of f.
Cloud Seminar: GraphLab: A Framework for Asynchronous Parallel Machine Learning
Wednesday, Sept. 7, 3:00–4:00 pm, 465H Soda Hall, UC Berkeley
Joseph Gonzalez, Carnegie Mellon University
While high-level data-parallel frameworks like MapReduce (Hadoop) dramatically simplify the design of large-scale data processing systems, they do not naturally express the asynchronous iterative graph computation found in many advanced machine learning techniques. Recent frameworks like Pregel, Giraph, and Pegasus have begun to simplify the design and implementation of large-scale graph algorithms by implementing the classic bulk synchronous computational model. However, none of these frameworks target the more expressive asynchronous computational model needed for a wide variety of popular graph structured machine learning algorithms. To fill this critical void, we developed the GraphLab framework which naturally expresses asynchronous graph computation while ensuring data consistency and achieving a high degree of parallel performance.
In this talk I will demonstrate how the MapReduce and Bulk Synchronous models of computation can lead to highly inefficient parallel learning systems. I will then introduce the GraphLab framework and explain how it addresses these critical limitations while retaining the advantages of a high-level abstraction. I will show how the GraphLab abstraction can be used to build efficient provably correct versions of several popular sequential machine learning algorithms. Finally, I will present scaling results in both the multi-core and cloud settings.
TRUST Security Seminar: The Cloud-y Future of Security Technologies
Thursday, Sept. 8, 1:00–2:00 pm, Soda Hall, Wozniak Lounge, UC Berkeley
Adam J. O’Donnell, Sourcefire
With a healthy dose of technologies and techniques borrowed from big-data companies such as Google and Facebook, engineers in the security industry have led a sea change in how security products are designed and implemented. Rather than rely upon ad hoc threat detection networks and loose partnerships with other security firms to detect new threats, security firms are moving to cloud-based product models, where threat information for malicious network traffic, malware, and various forms of attacks is centrally aggregated and processed from endpoints. Large teams of analysts-focused manual identification of threats are now being displaced by data mining and machine learning, pushing the effective time to live of an attack from weeks to hours. In this talk I will provide an overview of how it's been done.
Link of the Week: Is the Internet Destroying the Middle Class?
Jaron Lanier, the computer scientist who either coined or popularized the term “virtual reality” and received a Lifetime Career Award from the IEEE for his work in that field, is very concerned about the economic and social effects of the Internet. In a conversation titled The Local-Global Flip, or, “The Lanier Effect,” he explains how computers and networks create efficiencies that paradoxically make everyone poorer by destroying jobs:
Everyone’s into Internet things, and yet we have this huge global economic trouble. If you had talked to anyone involved in it twenty years ago, everyone would have said that the ability for people to inexpensively have access to a tremendous global computation and networking facility ought to create wealth. This ought to create wellbeing; this ought to create this incredible expansion in just people living decently, and in personal liberty. And indeed, some of that’s happened. Yet if you look at the big picture, it obviously isn’t happening enough, if it’s happening at all.
The situation reminds me a little bit of something that is deeply connected, which is the way that computer networks transformed finance. You have more and more complex financial instruments, derivatives and so forth, and high frequency trading, all these extraordinary constructions that would be inconceivable without computation and networking technology.
At the start, the idea was, “Well, this is all in the service of the greater good because we’ll manage risk so much better, and we’ll increase the intelligence with which we collectively make decisions.” Yet if you look at what happened, risk was increased instead of decreased.
In parallel it seems as though the middle classes have been having trouble all around the world, not just in the U.S., but in all developed societies at the same time that the Internet has been rising. I’m concerned that it’s not a matter of the Internet doing some good, but not enough good to undo unrelated coincidental troubles. I’m afraid the Internet, as we’ve conceived of it thus far, has been part of the problem. I’m also interested in the idea that if we conceive of it differently, it could be a solution.
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 7,000-plus 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 Department of Energy 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.