Dr. Stuart Slattery
Computational Sciences and Engineering Division
Oak Ridge National Laboratory

Scalable Algorithms for the Material Point Method

ABSTRACT: The Material Point Method (MPM) is a hybrid Lagrangian-Eulerian scheme for solving problems in continuum mechanics. It is particularly amenable to problems with large deformations, complex moving interfaces, and multiple materials. As part of the Exascale Computing Project (ECP) as well as the ORNL LDRD program, we have been developing software tools to allow for Exascale-class MPM simulations as well as smaller-scale simulations which can effectively use multi-core workstations, GPUs, and small clusters. In this talk we will overview MPM, discuss our software development activities, and present results from our ECP work.

BIOGRAPHY: Stuart’s work focuses on the research, development, implementation, and application of massively parallel algorithms for multiphysics and multiscale simulations, radiation transport, hydrodynamics, solid mechanics, and the modeling of various energy systems. This includes the development and application of methods in sparse and dense linear and nonlinear algebra, PDE discretizations, computational meshes, particles, and time integrators. Stuart leads exascale development efforts in multiphysics and multiscale coupling as well as co-design and application efforts for the development of particle methods for hydrodynamics and solid mechanics. He has also contributed to scalable algorithm research for particle methods in radiation transport. These efforts develop scalable, performance portable libraries and applications targeting next-generation supercomputers and impact multiscale and multiphysics simulations in DOE in areas including electric vehicle batteries, additive manufacturing, and nuclear reactor analysis. Stuart earned his Ph.D. in Nuclear Engineering and Engineering Physics at the University of Wisconsin before joining ORNL as a member of the scientific staff in 2013.