At the November 7 Lunch ‘n Learn seminar, three Princeton faculty members described their use of the University’s TIGRESS High Performance Computing Center, a collaborative collection of four major HPC resources, storage, and the programmers needed to facilitate computational science and engineering on campus.
Frans Pretorius, Assistant Professor of Physics, gave a brief overview of the computational techniques and resources needed to solve Einstein’s field equations, and described how the TIGRESS facilities are instrumental to his research. He explained that numerical relativity is concerned with solving Einstein’s field equations Gαβ =8∂Tαβ. For the computation work on the University’s supercomputing facilities, the field equations form a system of ten coupled, non-linear, second order partial differential equations each depending upon four or more spacetime coordinates.
Pretorius noted that with the advent of modern gravitational wave detectors, we may soon be able for the first time to observe the universe in gravitational waves. Illustrating the product of his research, here are two short movies that illustrate one of the most promising sources of gravitational waves, the collision of two black holes.
Dan Marlow, Professor and Chair of Physics, discussed the Large Hadron Collider (LHC) particle accelerator that will soon begin operation at the CERN lab in Geneva, Switzerland. He emphasized that the LHC represents the next step in our efforts to understand the microscopic nature of matter.
He described the accelerator and the large detector systems (and the associated computer hardware and software) that go with it. The collision energies at the LHC will be significantly higher than ever before, and particle physicists hope that it will reveal new phenomena that will help shed light on such fundamental questions as the origin of mass, the nature of dark matter, and the possible existence of extra dimensions.
When running, the LHC will capture the most promising collision events in a 24/7 data stream comprising 104 events per second. Each event is a complex problem in pattern recognition that worldwide supercomputers will address. As Marlow emphasized, the sun will never set on the worldwide computing complex.
Roberto Car, Ralph W. Dornte *31 Professor in Chemistry, discussed the Quantum mechanics of liquid water. Modeling liquid water has long been a theoretical challenge. On one hand, the hydrogen bond interactions between molecules need a proper quantum treatment of the electronic structure. On the other hand, the light mass of the hydrogen atoms prompts a quantum mechanical treatment of the nuclei too.
A full quantum mechanical simulation of the equilibrium structure of water requires very expensive computer simulations that are only possible on massively parallel architectures such as Princeton’s Blue Gene/L platform. Car noted that the simulation optimized for the IBM Blue Gene/L takes approximately one month to obtain converged statistical distributions. Car presented ongoing simulation results and showed how they compare to recent experimental neutron scattering data.
Posted by Lorene Lavora