PI: Prof. P.B. Price (Dean of Physical Sciences)
Co-PIs:
Prof. R.W. Falcone (Physics Chair), Dr. D.M. Lowder
The Physics Department will create a computing facility using Intel computers to support computationally intensive research projects and advanced instructional technology in the physical sciences.
The Physics Department at UC Berkeley Physics at Berkeley is synonymous with discovery and achievement. Seven of UC Berkeley's thirteen Nobel Prizes were awarded to physicists. Many notable discoveries have been made by Berkeley physicists - from Sfundamental properties of elementary particles to the physics behind magnetic resonance imaging, from the development of materials harder than a diamond to probes of the cosmic background radiation left from the Big Bang. With its outstanding faculty and proximity to the wealth of scientific resources in the Bay Area, the Physics Department is a national leader in diverse research and in education.
Proposed NOW Research The Physics Department
is participating in the Antarctic Muon and Neutrino Detector Array (AMANDA), an international collaboration which operates a high-energy neutrino observatory at the South Pole. Neutrino detection is a key methodology for study of the dynamics of active galactic nuclei. An initial neutrino detector is operating currently, but the final detector version will have a volume of one cubic-kilometer! Two Intel workstations will reside in the South Pole laboratory. Together with Intel computers housed at the Physics Department, these South Pole workstations will handle the enormous quantity of data expected from the AMANDA project, as well as from additional projects in laser-plasma interactions and nonlinear dynamics. Additionally, several workstations will be used to further the teaching mission of the Department as part of a new Physical Sciences Computing Facility (PSCF), scheduled to open at Berkeley in January 1998. By complementing the Macintosh computers which have been donated for use in this facility, Intel Pentium workstations and NT software will be available at the PSCF for use by students in science, engineering, and mathematics.
Importance and Visibility The AMANDA collaboration consists of 50 physicists from universities in the U.S., Sweden, and Germany. Members from each institution spend six weeks at the South Pole each year. While working there, they discuss data analysis techniques with members of the universities--including Berkeley--that comprise the NSF Center for Astrophysical Research in Antarctica. Science writers, members of the U.S. Congress, NSF officials such as Director Neal Lane, foreign dignitaries, and school teachers come to the South Pole Station to learn about the research being done there. News that Intel Pentium computers are used to handle the huge data stream generated by AMANDA would therefore be propagated throughout the U.S. and many other countries. The Pentium computers utilized in the PSCF will also be highly visible, as they are used for teaching and research purposes by a wide range of students and faculty in the physical and biological sciences and in engineering.
How Intel Machines Would Impact Proposed Research The addition of Intel workstations would have a sig
nificant impact on AMANDA's ability to handle its data. AMANDA generates events roughly 100 times per second, day in and day out. This means that the analysis procedures must be capable of handling roughly one-million events per day, which translates to between 5 and 10 Gigabytes of raw data per day. All of the data must be examined carefully to reject the background, primarily consisting of down-going muons, and pick out the few events per day expected from real up-going muons produced by the neutrinos coming up through the earth. Reconstruction of each muon event requires a log-likelihood fit in a 5-dimensional parameter space riddled with local minima. The magnitude of this problem can be seen from the fact that the reconstruction software now being used takes an average of 3 seconds to reconstruct a single event on a DEC Alpha; this is a factor of 1000 too slow to keep up with the experiment! Although fast event preprocessing algorithms help, the problem requires a cluster of fast workstations, each handling a portion of the events. The acquisition of INTEL workstations and the server would have an enormous and immediate impact on AMANDA's ability to handle its data; this will become especially critical when a full year of data from the 10-string array arrives from the South Pole at the end of this year.
Other Projects Using this Equipment The Physics Department could also utilize the donated Intel equipment for Particle-In-Cell (PIC) Codes to model Laser-Plasma Interactions and for Nonlinear Dynamics in Dissipative Systems.
February 1999