Perspectives on: Molecular dynamics and computational methods

Theoretical and Computational Chemistry, 1999

We review recent progress in understanding fundamental processes in biology, chemistry and physics on the basis of ab initio and molecular dynamics simulations. The first step of the visual process involving the excitation of bovine rhodopsin after absorption of light is taken as an example from biochemistry to demonstrate what is nowadays possible to simulate numerically. The act of freezing of water has recently been simulated, for the first time successfully, by scientists from chemistry. Martensitic transformation in bulk and nanophase materials, a typical and hitherto not completely solved problem from solid state physics, is used to illustrate the achievements of multimillion atoms simulations.

Handbook of Materials Modeling, 2005

R. Car et al. elements such as hydrogen; classical or ab initio path integral approaches can then be applied, albeit at a higher computational cost. The use of Newton's equations of motion for the nuclear evolution implies that vibrational degrees of freedom are not quantized, and will follow a Boltzmann statistics. This approximation becomes fully justified only for temperatures comparable with the highest vibrational level in the system considered.

2003

We review recent progress in understanding fundamental processes in biology, chemistry and physics on the basis of ab initio and molecular dynamics simulations. The first step of the visual process involving the excitation of bovine rhodopsin after absorption of light is taken as an example from biochemistry to demonstrate what is nowadays possible to simulate numerically. The act of freezing of water has recently been simulated, for the first time successfully, by scientists from chemistry. Martensitic transformation in bulk and nanophase materials, a typical and hitherto not completely solved problem from solid state physics, is used to illustrate the achievements of multimillion atoms simulations. 9.1 Molecular Dynamics as a Multidisciplinary Numerical Tool Molecular dynamics (MD) has proved to be an optimum numerical recipe applicable to problems with many degrees of freedom from quite different fields of science. The knowledge of the energy or potential landscape of interacting...

Arxiv preprint arXiv: …, 2011

Abstract. Born-Oppenheimer dynamics is shown to provide an accurate approximation of time-independent Schrödinger observables for a molecular system with an electron spectral gap, in the limit of large ratio of nuclei and electron masses, without assuming that the ...

Reviews of Modern Physics, 1994

An overview is presented of methods for time-dependent treatments of molecules as systems of electrons and nuclei. The theoretical details of these methods are reviewed and contrasted in the light of a recently developed time-dependent method called electron-nuclear dynamics. Electron-nuclear dynamics (END) is a formulation of the complete dynamics of electrons and nuclei of a molecular system that eliminates the necessity of constructing potential-energy surfaces. Because of its general formulation, it encompasses many aspects found in other formulations and can serve as a didactic device for clarifying many of the principles and approximations relevant in time-dependent treatments of molecular systems. The END equations are derived from the time-dependent variational principle applied to a chosen family of efhciently parametrized approximate state vectors. A detailed analysis of the END equations is given for the case of a single-determinantal state for the electrons and a classical treatment of the nuclei. The approach leads to a simple formulation of the fully nonlinear time-dependent Hartree-Fock theory including nuclear dynamics. The nonlinear END equations with the ab initio Coulomb Hamiltonian have been implemented at this level of theory in a computer program, ENDyne, and have been shown feasible for the study of small molecular systems. Implementation of the Austin Model 1 semiempirical Hamiltonian is discussed as a route to large molecular systems. The linearized END equations at this level of theory are shown to lead to the random-phase approximation for the coupled system of electrons and nuclei. The qualitative features of the general nonlinear solution are analyzed using the results of the linearized equa-918 918 920 920 921 921 922 922 922 923 923 923 923 924 924 925 925 926 928 928 928 929 929 930 931 931 933 935 935 937 937 937 940 941 atomic orbitals nal basis of trav-B. Nonorthogonal representation 1. Derivation in an orthonormal basis a. Metric b. Density matrix c. Energy 2. Derivation in the atomic-orbital basis a. Definition of parameters b. Dynamic orbitals c. Metric d. Density matrix e. Energy 3. Details of semiempirical approaches References 950 950 959

Lecture Notes in Physics, 2008

Physical Review B, 2012

As the processing power available for scientific computing grows, first principles Born-Oppenheimer molecular dynamics simulations are becoming increasingly popular for the study of a wide range of problems in materials science, chemistry and biology. Nevertheless, the computational cost of Born-Oppenheimer molecular dynamics still remains prohibitively large for many potential applications. Here we show how to avoid a major computational bottleneck: the self-consistent-field optimization prior to the force calculations. The optimization-free quantum mechanical molecular dynamics method gives trajectories that are almost indistinguishable from an "exact" microcanonical Born-Oppenheimer molecular dynamics simulation even when low pre-factor linear scaling sparse matrix algebra is used. Our findings show that the computational gap between classical and quantum mechanical molecular dynamics simulations can be significantly reduced.

Journal of Computational Physics, 1999

International Journal of Quantum Chemistry, 1994

The content of an ab-inddo time-dependent theory of quantm molecular dynamics of electrons and atomic nuclei is presented. Employing the time-dependent variational principle and a family of approximate state vectors yields a set of dynamical equations approximating the time-dependent SchrOdinger equation. These equations govern the time evolution of the relevant state vector parameters as molecular orbital coefficients, nuclear positions and momenta. This approach does not impose the Born-Oppenheimer approximation, does not use potential energy a. surfaces and takes into account elecow-nuclear coupling. Basic cnservation laws are fully ' obeyed. The simplest model of the theory employs a single determinantal state for the electrons CL and classical nuclei and is implemented in the computer code ENDyne. Results from this ab-. _ kio theory are reported for ion-atom and ion-molecule collisions.