Toward High Fidelity Materials Property Prediction from Multiscale Modeling and Simulation

Abstract

The current approach to materials discovery and design remains dominated by experimental testing, frequently based on little more than trial and error. With the advent of ever more powerful computers, rapid, reliable and reproducible computer simulations are beginning to represent a feasible alternative. As high performance computing reaches the exascale, exploiting the resources efficiently presents interesting challenges and opportunities. Multiscale modelling and simulation of materials is an extremely promising candidate for exploiting these resources based on the assumption of a separation of scales in the architectures of nanomaterials. We present examples of hierarchical and concurrent multiscale approaches which benefit from the weak scaling of monolithic applications, thereby efficiently exploiting large scale computational resources. We discuss several multiscale techniques, incorporating the electronic to the continuum scale, which can be applied to the efficient design of a range of nanocomposites. We then discuss our work on the development of a software toolkit designed to provide verification, validation and uncertainty quantification to support actionable prediction from such calculations.

Uncertainty quantification in classical molecular dynamics

“…Extensive studies we and others have performed in recent years [8,44,[46][47][48][49][64][65][66][67][68][69][70][71][72][73][74] confirm that the most effective and reliable computational route to reproducible binding free energies of ligands to proteins using MD simulation can be achieved using ensemble methods. The same conclusion has been drawn from MD simulations in other areas, including studies on materials applications such as graphene-based systems [75], on DNA nanopores using coarse-grained MD [76] and so on [77,78].…”

“…The same conclusion has been drawn from MD simulations in other areas, including studies on materials applications such as graphene-based systems [ 75 ], on DNA nanopores using coarse-grained MD simulations (Ahmad K, Coveney PV. 2019 unpublished work), on rate parameter estimation for binding kinetics [ 76 ] and so on [ 77 , 78 ].…”

Rapid, accurate, precise and reproducible ligand–protein binding free energy prediction

“…Computer simulations can be performed under conditions where it is difficult or impossible to conduct experiments, for instance, at very high pressures and temperatures. But, beyond the provision of qualitative insight, as our understanding increases one would hope to use these methods to quantitatively predict the outcome of experiments prior to, and indeed even instead of, performing them [ 1 – 3 ]. In this way, computational techniques should reduce time and cost in industrial processes like the discovery of drugs and advanced materials, which take more than 10 years and $2.6 billion for the former [ 4 ], and 20 years and perhaps $10 billion for the latter.…”

From digital hype to analogue reality: Universal simulation beyond the quantum and exascale eras

“…Through our work on the latter problem we seek to reliably design 2D nanocomposite materials through multiscale materials modelling. Reassuringly, quantum simulations of material properties performed by different researchers and with different software are both reproducible and able to produce identical results [ 55 ].…”