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.
Three distinct forms are derived for the force virial contribution to the pressure and stress tensor of a collection of atoms interacting under periodic boundary conditions. All three forms are written in terms of forces acting on atoms, and so are valid for arbitrary many-body interatomic potentials. All three forms are mathematically equivalent. In the special case of atoms interacting with pair potentials, they reduce to previously published forms. (i) The atom-cell form is similar to the standard expression for the virial for a finite nonperiodic system, but with an explicit correction for interactions with periodic images. (ii) The atom form is particularly suited to implementation in modern molecular dynamics simulation codes using spatial decomposition parallel algorithms. (iii) The group form of the virial allows the contributions to the virial to be assigned to individual atoms.
The time evolution of species concentrations in biochemical reaction networks is often modeled using the stochastic simulation algorithm (SSA) [Gillespie, J. Phys. Chem. 81, 2340 (1977)]. The computational cost of the original SSA scaled linearly with the number of reactions in the network. Gibson and Bruck developed a logarithmic scaling version of the SSA which uses a priority queue or binary tree for more efficient reaction selection [Gibson and Bruck, J. Phys. Chem. A 104, 1876 (2000)]. More generally, this problem is one of dynamic discrete random variate generation which finds many uses in kinetic Monte Carlo and discrete event simulation. We present here a constant-time algorithm, whose cost is independent of the number of reactions, enabled by a slightly more complex underlying data structure. While applicable to kinetic Monte Carlo simulations in general, we describe the algorithm in the context of biochemical simulations and demonstrate its competitive performance on small- and medium-size networks, as well as its superior constant-time performance on very large networks, which are becoming necessary to represent the increasing complexity of biochemical data for pathways that mediate cell function.
Large-scale molecular dynamics simulations and the reactive force field ReaxFF were used to study shock-induced initiation in crystalline pentaerythritol tetranitrate (PETN). In the calculations, a PETN single crystal was impacted against a wall, driving a shockwave back through the crystal in the [100] direction. Two impact speeds (4 and 3 km/s) were used to compare strong and moderate shock behavior. The primary difference between the two shock strengths is the time required to exhibit the same qualitative behaviors with the lower impact speed lagging behind the faster impact speed. For both systems, the shock velocity exhibits an initial deceleration due to onset of endothermic reactions followed by acceleration due to the onset of exothermic reactions. At long times, the shock velocity reaches a steady value. After the initial deceleration period, peaks are observed in the profiles of the density and axial stress with the strongly shocked system having sharp peaks while the weakly shocked system developed broad peaks due to the slower shock velocity acceleration. The dominant initiation reactions in both systems lead to the formation of NO(2) with lesser quantities of NO(3) and formaldehyde also produced.
The thermal conductivity of nanostructures generally decreases with decreasing size because of classical size effects. The axial thermal conductivity of polymer chain lattices, however, can exhibit the opposite trend, because of reduced chain-chain anharmonic scattering. This unique feature gives rise to an interesting onedimensional-to-three-dimensional transition in phonon transport. We study this transition by calculating the thermal conductivity of polyethylene with molecular dynamics simulations. The results are important for designing inexpensive high thermal-conductivity polymers.
We propose computational protocol ͑compressive shear reactive dynamics͒ utilizing the ReaxFF reactive force field to study chemical initiation under combined shear and compressive load. We apply it to predict the anisotropic initiation sensitivity observed experimentally for shocked pentaerythritol tetranitrate single crystals. For crystal directions known to be sensitive we find large stress overshoots and fast temperature increase that result in early bond-breaking processes whereas insensitive directions exhibit small stress overshoot, lower temperature increase, and little bond dissociation. These simulations confirm the model of steric hindrance to shear and capture the thermochemical processes dominating the phenomena of shear-induced chemical initiation. Experiments by Bridgman 1 demonstrated that chemical transformations occur far more readily under combined shear and pressure loads. Coupling between mechanical, thermal, and chemical effects is important for initiation of detonation in explosives under mechanical impact. Understanding how the detonation sensitivity depends on formulation and structural properties is a critical issue in explosive technology. A great deal of effort has gone into the development of experimental techniques to study these processes ͑e.g., time resolved emission spectroscopy 2-4 ͒ but they have not yet provided satisfactory resolution of initial steps of detonation. Little is known about how mechanical and chemical processes couple to initiate detonation.An ideal system to examine for hints on the atomistic origin of sensitivity, is single crystal of pentaerythritol tetranitrate ͑PETN, four CH 2 -O-NO 2 chains connected to a central carbon, see inset of Fig. 1͒. showed that shock compression of the single crystal in different directions leads to dramatic differences in the sensitivity. Thus the pressure threshold for detonation along the ͓100͔ direction is at least ϳ4 times that for ͓110͔ direction. 10 Such single crystal experiments eliminate many of the variables that complicate interpretations ͑grain boundaries, voids, and internal defects͒ providing an unambiguous challenge to any proposed mechanisms. Dick et al. 8,9 attributed the orientational anisotropy of PETN sensitivity to steric hindrance to the shear of molecules in the neighboring slip planes. Recently, Plaksin 11 showed clear-cut experimental evidence that initiation of detonation occurs preferentially in directions with the maximum shear stress.Here we use reactive dynamics ͑RD͒ simulations at constant shear rate on uniaxially compressed PETN single crystal to show that the physical and chemical responses depend dramatically on the compression direction and slip system. In contrast, RD simulations of pure uniaxial shock compression in various directions show no correlation of the reactive behavior with anisotropy of sensitivity in experiment. Key to our simulations is the ReaxFF reactive force field, 12,13 whose parameters are trained to match quantum mechanics descriptions of reaction barriers for all plausibl...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.