Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of ML-IAPs based on four local environment descriptors -Behler-Parrinello symmetry functions, smooth overlap of atomic positions (SOAP), the Spectral Neighbor Analysis Potential (SNAP) bispectrum components, and moment tensors -using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model, and consequently computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.

The Spectral Neighbor Analysis Potential (SNAP) is a classical interatomic potential that expresses the energy of each atom as a linear function of selected bispectrum components of the neighbor atoms. An extension of the SNAP form is proposed that includes quadratic terms in the bispectrum components. The extension is shown to provide a large increase in accuracy relative to the linear form, while incurring only a modest increase in computational cost. The mathematical structure of the quadratic SNAP form is similar to the embedded atom method (EAM), with the SNAP bispectrum components serving as counterparts to the two-body density functions in EAM. The effectiveness of the new form is demonstrated using an extensive set of training data for tantalum structures. Similar to artificial neural network potentials, the quadratic SNAP form requires substantially more training data in order to prevent overfitting. The quality of this new potential form is measured through a robust cross-validation analysis.

We use molecular dynamics simulations with the reactive potential ReaxFF to investigate the initial reactions and subsequent decomposition in the high-energy-density material α-HMX excited thermally and via electric fields at various frequencies. We focus on the role of insult type and strength on the energy increase for initial decomposition and onset of exothermic chemistry. We find both of these energies increase with the increasing rate of energy input and plateau as the processes become athermal for high loading rates. We also find that the energy increase required for exothermic reactions and, to a lesser extent, that for initial chemical reactions depend on the insult type. Decomposition can be induced with relatively weak insults if the appropriate modes are targeted but increasing anharmonicities during heating lead to fast energy transfer and equilibration between modes that limit the effect of loading type.

We use molecular dynamics simulations to describe the chemical reactions following shockinduced collapse of cylindrical pores in the high-energy density material RDX. For shocks with particle velocities of 2km/s we find that the collapse of a 40nm diameter pore leads to a deflagration wave. Molecular collisions during the collapse lead to ultra-fast, multi-step chemical reactions that occur under non-equilibrium conditions. Exothermic products formed during these first few picoseconds prevent the nanoscale hotspot from quenching. Within 30ps, a local deflagration wave develops; it propagates at 0.25km/s and consists of an ultra-thin reaction zone of only ~5nm, thus involving large temperature and composition gradients. Contrary to the assumptions in current models, a static thermal hotspot matching the dynamical one in size and thermodynamic conditions fails to produce a deflagration wave indicating the importance of nonequilibrium loading in the criticality of nanoscale hot spots. These results provide insight into the initiation of reactive decomposition. TOC GRAPHICS