Computing stationary probability vectors of Markov chains with cyclic spectra


Ivo Marek
Czech Institute of Technology Prague

Some algorithms utilizing iterative aggregation/disaggregation techniques (IAD) are shown to be suitable for computing stationary probability vectors of Markov chains (MC) as well as other MC-characteristics. The effciency of IAD compared with other methods grows with growing index of cyclicity of appropriate transition matrices. The effciency of IAD is superior for massively cyclic MC's.