Abstract:Two computationally efficient implementations of first-principles molecular dynamics are examined using the reactive scattering of two linear ozone molecules to yield three molecules of oxygen. These methods are based on Kohn-Sham density-functional theory in which the charge density is fitted variationally. The linear combination of Gaussian-type orbitals approach is used to construct the orbital and local-potential basis sets. The first method is a completely general approach requiring variational fitting of the exchange-correlation energy density on a grid of points. The second method avoids the grid of points, but is restricted to the Xα functional. Key elements of the dynamics calculation include self-consistent-field convergence at each time step, use of fractional occupation numbers to model the configurational mixing that occurs upon bond formation or dissociation, and accurate analytic evaluation of the appropriate forces acting upon the nuclei. Trajectories computed using modern local-density and gradient-corrected functionals, as well as Xα (α = 0.6667) treated analytically, are compared. It is found that reactive trajectories obtained using analytic Xα and the generalized gradient approximation are very similar when the initial kinetic energy is well above threshold for reaction. Under similar conditions, trajectories resulting from use of modern local-density functionals exhibit somewhat distinctive behavior. These results suggest that the particularly efficient analytic Xα approach is suitable for preliminary investigations of chemical reactions using first-principles molecular dynamics based on density-functional theory.

A Bell-Evans-Polanyi principle for molecular dynamics trajectories and its implications for global optimization

Efficient moves for global geometry optimization methods and their application to binary systems

Minimal Curvature Trajectories: Riemannian Geometry Concepts for Model Reduction in Chemical Kinetics

On fundamental unifying concepts for trajectory-based slow invariant attracting manifold computation in multiscale models of chemical kinetics

Free energy, friction, and mass profiles from short molecular dynamics trajectories

Basis Set Generation for Quantum Dynamics Simulations Using Simple Trajectory-Based Methods

Global Convergence of Langevin Dynamics Based Algorithms for Nonconvex Optimization

Entanglement-Optimal Trajectories of Many-Body Quantum Markov Processes

Effect of exchange-correlation functionals and symmetry constraints of electronic structure on the trajectories of reactive molecular collisions

Atomically Detailed Simulations of Helix Formation with the Stochastic Difference Equation

Virtual Molecular Dynamics

Optimization by linear kinetic equations and mean-field Langevin dynamics

Convergence and equilibrium in molecular dynamics simulations

Quasiclassical Trajectory Simulation as a Protocol to Build Locally Accurate Machine Learning Potentials

Locating Transition States by Variational Reaction Path Optimization with an Energy-Derivative-Free Objective Function

Reinforcement Learning for Molecular Dynamics Optimization: A Stochastic Pontryagin Maximum Principle Approach

Trajectory Convergence from Coordinate-wise Decrease of General Energy Functions

An approach for generating trajectory-based dynamics which conserves the canonical distribution in the phase space formulation of quantum mechanics. I. Theories

Trajectory Statistical Learning of the Potential Mean of Force and Diffusion Coefficient from Molecular Dynamics Simulations

Doob's Lagrangian: A Sample-Efficient Variational Approach to Transition Path Sampling

Exploration of the Global Minimum and Conical Intersection with Bayesian Optimization