In probability theory, the matrix geometric method is a method for the analysis of quasi-birth–death processes, continuous-time Markov chain whose transition rate matrices with a repetitive block structure.[1] The method was developed "largely by Marcel F. Neuts and his students starting around 1975."[2]