InTheLoop | 11.11.2013
Berkeley Lab Expertise to be Highlighted in DOE Booth at SC13
When the SC13 conference opens Nov. 17 in Denver, LBNL and the other DOE national laboratories will make their return to the conference exhibition. For the first time, the facilities and capabilities of 15 national labs will be highlighted in a booth program featuring presentations by HPC experts, electronic posters, demonstrations, roundtable discussions and a 3D display showing simulations and modeling.
Kathy Yelick and Greg Bell are among the 14 featured speakers giving talks in the booth. ESnet will be demonstrating the MyESnet portal and hosting a discussion of the Science DMZ. Additionally, the lab will have eight electronic posters displayed on interactive screens. For more information on the DOE booth, go to http://scdoe.info/.
CRD Workshop Brings Together Scientists, Potential Machine Learning Solutions
As part of the Computational Research Division’s strategic emphasis on increasing the role of computation in all aspects of scientific discovery, CRD staff organized a one-day workshop on Machine Learning for Science. The meeting was held Nov. 4 at Berkeley Lab. In many areas of science, the generation of data is now often outstripping the abilities of researchers to manage, analyze and understand the data. Machine learning, the development and use of advanced techniques to automatically classify data, detect patterns or extract results, is arguably the most widely used methodology to deal with data of this size and complexity.
The workshop, which drew more than 70 participants, specifically looked at increasing the role of machine learning in the areas of climate research, cosmology, materials, microtomography and metagenomics. About one-third of the participants were machine learning experts from universities, including UC Berkeley, UC Davis, Stanford and Rensselaer Polytechnic Institute. The others were mainly domain scientists and computer scientists From Berkeley Lab, the Stanford Linear Accelerator Center and the Department of Energy’s Joint Genome Institute (JGI). Read more.
Reminder: MATLAB’s Cleve Moler to Speak on Nov. 13
Computing Sciences is launching a Distinguished Lecturer Series, with the first talk to be given by Cleve Moler at 2 p.m. Wednesday, Nov. 13, in the Bldg. 50 Auditorium. Moler is the creator of MATLAB and a cofounder of MathWorks, where he is currently chairman and chief mathematician of the company. In his talk, Moler will show how MATLAB has evolved over more than 30 years from a simple matrix calculator to a powerful technical computing environment. He will demonstrate several examples of MATLAB applications, then conclude with a discussion of current developments, including Parallel MATLAB for multicore and multicomputer systems.
This Week’s Computing Sciences Seminars
Parallel Implementation of Multireference Coupled Cluster Methods and Calculations on Large Systems
Tuesday, November 12, 2 – 3 p.m., Bldg. 50B, Room 4205
Jiri Brabec, J. Heyrovsky Institute of Physical Chemistry, Academy of Sciences of the Czech Republic
Abstract: Multireference coupled cluster (MRCC) methods are amongst the most accurate wavefunction based approaches. In addition to improving on traditional coupled cluster methods, it can also systematically treat both static and dynamic correlation. This is especially important in describing chemical reactions, reaction pathways, complex potential energy surfaces, to name a few. However, their use is limited by their high computational cost, which can be alleviated with scalable MRCC codes.
The proposed two-level parallel algorithm utilizes processor groups to calculate the equations for the MRCC amplitudes. In the basic formulation each processor group constructs the equations related to a specic subset of references. The scalability of Brillouin-Wigner and Mukherjee MRCC code was tested on tens of thousands CPU cores. The applicability is demonstrated also on core-level excited states. The Universal State-Selective (USS) correction is introduced and it is shown that it may improve the accuracy of MRCC approaches.
Bayesian Learning of Stochastic Dynamical Model Formulation
Wednesday, November 13, 3:30 - 4:30 p.m., 939 Evans Hall - UC Berkeley Campus
Pierre Lermusiaux, Massachusetts Institute of Technology
Abstract: In this presentation, we first highlight recent results by our MSEAS group, including high-order Finite-Element schemes for biogeochemical ocean dynamics and exact path planning for swarms of ocean vehicles using level-set equations. We then address a holistic challenge in ocean Bayesian estimation: i) predict the probability distribution functions (pdfs) of large nonlinear ocean systems using stochastic partial differential equations, ii) assimilate data using Bayes' law with these pdfs, iii) predict the future data that optimally reduce uncertainties and rank the known and learn the new model formulations themselves. Overall, we allow the joint inference of the state, equations, geometry, boundary conditions and initial conditions of dynamical models. Examples are provided using time-dependent ocean and fluid flows, including cavity, double-gyre and sudden-expansion flows with jets and eddies. The Bayesian model inference is illustrated by the estimation of obstacle shapes and of biogeochemical reaction equations based on very limited observations. This is joint work with our MSEAS group.
Link of the Week: Video Illustrates the Beauty of Mathematics
According to British philosopher, logician, mathematician, historian and social critic Bertrand Russell, “Mathematics, rightly viewed, possesses not only truth, but supreme beauty — a beauty cold and austere, without the gorgeous trappings of painting or music.” A video posted three weeks ago by Yann Pineill and Nicolas Lefaucheux brings this observation to life.