A COMPARISON OF THE EFFICIENCY OF MONTE-CARLO (MC) AND MOLECULAR-DYNAMICS (MD) CALCULATIONS AT 1ST ORDER PHASE-TRANSITIONS

As a rule the canonical ensemble is used in the statistical analysis of phase transitions. There the system of interest is coupled to an infinite bath. It will be demonstrated that it is of considerable advantage to use a different ensemble where the size of the bath is comparable to the size of the system of interest. The advantages will be demonstrated for the example of a first order phase transition by making use of the density of states of the modified xy-model which has recently been determined in an MD-calculation for several lattice sizes. In contrast to the canonical ensemble the probability distribution does not suffer from the double hump structure and therefore allows an accurate determination of the properties at the transition. In the end we shall argue that the corresponding ensemble can also be realized in an MC-approach.