One primary requirement for recipients of the Computational Science Graduate Fellowship is the "laboratory practicum" research experience at a Department of Energy (DOE) national laboratory. This is an opportunity for students to actively contribute to an impactful science project under the guidance of some of the most experienced and respected scientists in the entire world. During my time in graduate school, I completed three very rewarding practicums at the DOE national laboratores: two at Argonne National Laboratory in Illinois and one at the Lawrence Berkeley National Laboratory in California. This presentation gives an overview of these three projects, which all share one important theme: that effectively using high performance computing resources and techniques enables powerful scientific research and interesting discoveries. At Argonne, I optimized extreme-scale quantum chemistry calculations using novel load balancing strategies. A year later, I worked with scientists at the Center for Exascale Simluation of Advanced Reactors to effectively use new computer accelerator devices to simulation Monte Carlo simulations of nuclear reactor physics. Finally, at Berkeley I applied a cutting-edge programming model, called UPC++, to tackle critical issues in calculating fundamental quantum chemistry quantities on distributed supercomputers. In short, this talk describes what I learned at the national laboratories when simulating picoscale physics with petascale computers.
