Accurate Reactive Molecular Dynamics Potentials for Complex Energetic Materials
POSTER
Abstract
Many of the chemical mechanisms controlling the safety and performance of energetic materials are not well understood. Molecular dynamics simulations offer insights into mechanisms controlling energetic material chemistry, but are limited by several factors, including the accuracy of the atomic interaction potential. The Chebyshev Interaction Model for Efficient Simulation (ChIMES) many-body reactive potential is based on matching to accurate quantum simulations. ChIMES has been applied to a number of molecular systems at extreme conditions typical of shockwaves, but complex energetic materials with four or more elements remain challenging for this approach. The linear and orthogonal nature of the ChIMES interaction model allows the use of advanced regression techniques to control solution accuracy and transferability. We will describe efforts to extend ChIMES to more complex chemistry and improved accuracy, based on parameter selection through highly scalable Least Absolute Shrinkage and Selection Operator (LASSO) algorithms and other techniques to improve model generation.
Publication: RK Lindsey, CH Pham, N Goldman, S Bastea, and LE Fried, "Machine-Learning a Solution for Reactive Simulations of Energetic Materials", Propellants Explosives Pyrotechnics, 47(8), art. e202200001, (2022).
Presenters
-
Laurence E Fried
Lawrence Livermore Natl Lab, Lawrence Livermore National Laboratory
Authors
-
Laurence E Fried
Lawrence Livermore Natl Lab, Lawrence Livermore National Laboratory
-
Cong Huy Pham
Lawrence Livermore National Lab
-
Nir Goldman
Lawrence Livermore Natl Lab
-
Rebecca K Lindsey
University of Michigan, Ann Arbor, University of Michigan