Princeton Plasma Physics LaboratoryCoordinator: William Tang
Review abstracts for current and past practicum experiences at Princeton Lab >>
The Princeton Plasma Physics Laboratory (PPPL) at Princeton University (PU) is the DOE national laboratory for magnetic fusion energy and basic plasma physics research. Associated computational research and development opportunities span a wide range of stimulating areas encompassing, as examples — kinetic plasma turbulence and confinement, nonlinear magnetohydrodynamics, energetic particle dynamics, the design of advanced and alternative fusion devices (including optimized stellarator configurations, field-reversed devices, low temperature plasma simulations for industrial applications, plasma materials interactions, electromagnetic wave heating, and the prominent whole device modeling area that includes the PPPL transport analysis (TRANSP) and the national Exascale Computing Project (ECP). These collectively provide a rich environment for developing and applying advanced numerical techniques on modern high-performance computing (HPC) platforms.
Since problems in computational plasma physics/fusion energy sciences present truly large, multi-scale challenges, they require engaging innovative software development with the potential of fully utilizing supercomputers featuring massively parallel heterogeneous architectures. In this regard, the recent arrival of the PPPL/PU 1.4 PF GPU/CPU cluster, Traverse, introduces an exciting local HPC resource. Moreover, PPPL has an impressive record of highly productive collaborative access to powerful HPC resources at major DOE computational resource centers at Oak Ridge, Argonne, and Lawrence Berkeley national laboratories.
In addition to building the scientific foundations needed to accelerate R&D progress by the familiar hypothesis-driven/first principles approaches just noted to predict complex dynamical behavior, an increasingly prominent methodology deployed in many scientific and industrial domains involves engagement of big-data-driven statistical methods featuring machine learning. Associated recent R&D activities at PPPL have accordingly attracted the interests and subsequent successful practicum experiences of DOE CSGF fellows that have proven key to advancing the artificial intelligence (AI) enabled deep learning software with big data application studies that delivered much-needed guidance for predicting very dangerous large-scale disruption events in magnetically-confined plasmas.
More generally, DOE CSGF fellows have had and can be expected to continue to have major impacts on PPPL research through their practicum experiences.
Related PPPL resources: