Practical Guide to Monte Carlo

Abstract

I show how to construct Monte Carlo algorithms (programs), prove that they are correct and document them. Complicated algorithms are build using a handful of elementary methods. This construction process is transparently illustrated using graphical representation in which complicated graphs consist of only several elementary building blocks. In particular I discuss the equivalent algorithms, that is different MC algorithms, with different arrangements of the elementary building blocks, which generate the same final probability distribution. I also show how to transform a given MC algorithm into another equivalent one and discuss advantages of the various ``architectures''.