Some additional recollections, and the absence thereof, about the early history of computer simulations in statistical mechanics

摘要

This lecture is an extension and correction of a previous lecture given by the author ten years ago at ''Corso 97'' in Varenna. Here again he emphasizes that his early work was exclusively with applications of the Metropolis Monte Carlo method. His only connection with the early work on the molecular dynamics method was in collaboration with Alder and Wainwright in their joint effort to reconcile the results of the Monte Carlo and molecular dynamics methods for hard spheres. Here he amplifies a point suggested by a question asked by Professor Ciccotti: Namely, when was it discovered that the Metropolis method consists in the generation of a realization of a Markov chain, for which there was a large body of mathematical theory that made the justification of the method quite a simple matter.