Extreme Computing to Power Accurate Atomistic Simulations

The sophistication of modeling and simulation will be enhanced not only by the wealth of data available from MaRIE but by the increased computational capacity made possible by the advent of extreme computing.

Advances in high-performance computing and theory allow longer and larger atomistic simulations than currently possible. Such approach can then inform sub-grid models used in continuum simulations used for engineering design.

The sophistication of modeling and simulation will be enhanced not only by the wealth of data available from MaRIE, but also by the increased computational capacity made possible by the advent of extreme (e.g. exascale) computing. This will allow novel methods for solving complex models, as well as large-scale and time-accelerated atomistic simulations approaching microns in size and milliseconds in time, and electronic structure calculations on unit cells as large as several thousand atoms.

Parameterizing and manipulating the free energy landscape will lead to prediction and control of new materials. Ultimately, the goal is a coupled, adaptive strategy in which methods at certain scales modify, and are in turn modified by, those at other scales. The process, if carried out self-consistently and combined with uncertainty analysis, provides a road map towards materials properties by design. It also drives reduced uncertainty in the predictions of time-dependent performance of engineered systems.

Characterizing and parameterizing exploring the associated nonequilibrium and nonadiabatic behavior are at the heart of MaRIE. For example, through MaRIE we will control the strain rate at which dislocation plasticity or twinning deformation mode is preferred, and design entirely new classes of materials with controlled functionality.

www.lanl.gov/science-engineering/science-facilities/dmmsc/tmc