On the structure of 1-dependent Markov chains (Q1194480): Difference between revisions

A stochastic process \({\mathcal W}=\{W_ n\}^ \infty_{n=1}\) is a 2-block factor if there are independent, identically distributed random variables \(\{X_ n\}^ \infty_{n=1}\) and a measurable function \(f\) of two real variables such that \(W_ n=f(X_ n,X_{n+1})\) for all \(n\geq 1\). Clearly, a 2-block factor is stationary and 1-dependent in the sense that \((W_ 1,W_ 2,\dots,W_ n)\) and \((W_{n+2},W_{n+3},\dots)\) are independent for all \(n\geq 1\). Assume now that \(\mathcal W\) is stationary and 1-dependent. The authors give two examples, including a Markov chain with five states, to show that \(\mathcal W\) need not be a 2-block factor. They also show that if \(\mathcal W\) is a Markov chain with four or fewer states, or a renewal process, then \(\mathcal W\) will be a 2-block factor.