Scalable Atomistic Simulations of Energetic Materials
Invited
Abstract
We have developed an extension of the divide-and-conquer algorithmic framework called divide-conquer-recombine to make quantum molecular dynamics (QMD) and reactive molecular dynamics (RMD) simulations scalable on emerging exascale supercomputers and beyond. On today’s supercomputing platform, for instance, the framework has achieved over 98% of the perfect speedup on 786,432 IBM Blue Gene/Q processors for 40 trillion electronic degrees-of-freedom QMD in the framework of density functional theory and 68 billion-atom RMD. Production simulations on energetic materials include: (1) dynamic phase transition, crossover in anisotropic mechanochemistry and multistage reaction pathways in energetic crystals; and (2) reaction dynamics and enhanced energetic performance of metallic-nanoparticle/graphene composites.
–
Presenters
-
Aiichiro Nakano
University of Southern California, Physics, University of Southern California, Physics & Astronomy, University of Southern California
Authors
-
Aiichiro Nakano
University of Southern California, Physics, University of Southern California, Physics & Astronomy, University of Southern California
-
Rajiv Kalia
University of Southern California, Physics, University of Southern California, Physics & Astronomy, University of Southern California
-
Priya Vashishta
University of Southern California, Physics, University of Southern California, Collaboratory for Advanced Computing and Simulations, University of Southern California, Physics & Astronomy, University of Southern California