A note on the equivalence of two approaches for specifying a Markov process (Q1611567)

There are two specifications of discrete-time high-order vector Markov processes. The first one uses the transition probability and the initial distribution. The other one is based on a stochastic difference equation. The authors prove that these two approaches are equivalent under some conditions. They reduce first a \(p\)th-order Markov chain to a first order vector chain via the stacking operation. Then they use an invertible normal form of the stochastic difference equation. The conditions ensuring equivalence of the two approaches concern equivalence of certain distribution function. (Two unidimensional distribution functions \(F\) and \(G\) are said to be equivalent if and only if the discontinuity points of \(F\) can be mapped to those of \(G\) in a one-to-one manner with the jump size preserved.).