Materials Sciences Division / Physics & Life Sciences Directorate
Lawrence Livermore National Laboratory
Crossing Materials Scales with LAMMPS
How to describe and predict macroscopic behaviors from the juggles and wiggles of atomic motion has been a challenge attracting a great many minds over many years. Emerging in the 80s and 90s, the dominant paradigm for bridging the notorious scale gaps in computational sciences has been multiscale modeling. Having preached the multiscale sermon for many years myself, here I will discuss one particular context of materials simulations where straightforward brute force MD is charmingly and disarmingly efficient. All it takes is LAMMPS, an ungodly number of core hours and an unsubstantiated belief that MD simulations on length and time scales that were previously deemed unthinkable, are now possible. In this talk I will describe our recent and ongoing work on direct MD simulations of crystal plasticity intended to bridge the disconnect between angstroms and femtoseconds of atomic motion to microns and microseconds of macroscopic crystal plasticity. As opposed to multiscale, our simulations can be regarded as cross-scale, being sufficiently large to be statistically representative of collective action of dislocations resulting in macroscopic plastic flow and yet fully resolved to every tiniest detail of atomic motion.