Modifications of the molecular-dynamics method for different statistical ensembles are examined. Particular emphasis is given to the Parrinello-Rahman method wherein the volume and shape of a molecular-dynamics cell are allowed to vary with time. The latter circumstance is of great importance because it enables processes involving marked structural changes in the system to be studied.Computer simulations, as well as real experiments, are an efficient tool of research into the properties and structure of different materials . Among the most popular computer simulation methods are the variational or molecularstatistical method, molecular-dynamics (MD) method, and statistical simulation or Monte-Carlo method.The variational method calls for minimization of the energy of the system as a function of 3N+m variables (N is the number of particles in the system and m is the number of variables describing the MD cell parameters). Determination of a minimum of this composite nonlinear function is rather time consuming and is a nonlinear programming problem from a mathematical standpoint solved by well-known numerical methods. By virtue of its peculiar features (finding the equilibrium stable or metastable state of the system), the variational method is commonly used to obtain information on the structure of the object under study.The Monte-Carlo or stochastical simulation method is applied profitably in investigations of microscopic properties of the system. This method is used to advantage in studying the processes of atomic ordering-disordering, crystal growth, annealing of radiation-induced defects, glide of dislocations over a system of obstacles, etc.The MD method provides a description of the time evolution of the system by means of a system of 6N ordinary differential equations in accordance with Newton's laws of motion. This method has proved to be efficient for calculating the structural properties of point defects, dislocation cores, and grain boundaries and for studying the atomic diffusion processes. In recent years, the MD method has been used in investigations of structural phase transformations in metals and alloys. Adequate simulations of processes involving significant structural changes in the system have been made possible with the development of modifications of the MD method that allow changes in the volume and shape of the MD cell.Below is an overview of modifications of the MD method now in use for different statistical ensembles.
THE MOLECULAR-DYNAMICS METHOD FOR DIFFERENT STATISTICAL ENSEMBLESThe MD method is determination of the mechanical trajectory of a system of N particles obeying the following well-known many-body interaction law: 1 2 ( , ,..., ).
N V r r rThe simulations proceed from a well-defined microscopic description of the system with the use of a Hamiltonian Н or a Lagragian L to derive the following equations of motion: