Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The emerging machine-learning (ML) potentials achieve highly accurate interpolation within a large DFT database but, being purely mathematical constructions, suffer from poor transferability to unknown structures. We propose a new approach that can drastically improve the transferability of ML potentials by informing them of the physical nature of interatomic bonding. This is achieved by combining a rather general physics-based model (analytical bond-order potential) with a neural-network regression. This approach, called the physically informed neural network (PINN) potential, is demonstrated by developing a general-purpose PINN potential for Al. We suggest that the development of physics-based ML potentials is the most effective way forward in the field of atomistic simulations.

Simulations based on solving the Kohn-Sham (KS) equation of density functional theory (DFT) have become a vital component of modern materials and chemical sciences research and development portfolios. Despite its versatility, routine DFT calculations are usually limited to a few hundred atoms due to the computational bottleneck posed by the KS equation. Here we introduce a machine-learning-based scheme to efficiently assimilate the function of the KS equation, and by-pass it to directly, rapidly, and accurately predict the electronic structure of a material or a molecule, given just its atomic configuration. A new rotationally invariant representation is utilized to map the atomic environment around a grid-point to the electron density and local density of states at that grid-point. This mapping is learned using a neural network trained on previously generated reference DFT results at millions of grid-points. The proposed paradigm allows for the high-fidelity emulation of KS DFT, but orders of magnitude faster than the direct solution. Moreover, the machine learning prediction scheme is strictly linear-scaling with system size.

Polymers are a vital part of everyday life. Their chemical universe is so large that it presents unprecedented opportunities as well as significant challenges to identify suitable application-specific candidates. We present a complete end-to-end machine-driven polymer informatics pipeline that can search this space for suitable candidates at unprecedented speed and accuracy. This pipeline includes a polymer chemical fingerprinting capability called polyBERT (inspired by Natural Language Processing concepts), and a multitask learning approach that maps the polyBERT fingerprints to a host of properties. polyBERT is a chemical linguist that treats the chemical structure of polymers as a chemical language. The present approach outstrips the best presently available concepts for polymer property prediction based on handcrafted fingerprint schemes in speed by two orders of magnitude while preserving accuracy, thus making it a strong candidate for deployment in scalable architectures including cloud infrastructures.