Determining the independence of random variables
James L. Massey · 2002
A graphical calculus is presented for determining the independence and conditional independence of random variables in a specified probabilistic setting. The calculus is developed first for the case of random variables that form a Markov chain. The calculus is then extended to the "general causal case" where the random variables are obtained from a sequence of random experiments in which each experiment can be carried out in full when the results of specified previous experiments are made available to it.