= Periodic Markov chain
{wiki}
For a state $i$ of a <Markov chain>, its <period of a state in a Markov chain>[period] is the <greatest common divisor> of the positive return lengths,
$$
d(i)=\gcd\{n\geq1:P^n(i,i)>0\}.
$$
A state is <periodic Markov chain>[periodic] when $d(i)>1$. All states in an <communicating class>[irreducible class] have the same <period of a state in a Markov chain>[period]: paths connecting two states transfer return cycles between them. A reducible chain can have classes of different periods, so a single <period of a state in a Markov chain>[period] for the whole chain requires qualification.
The deterministic two-state chain $P=\begin{pmatrix}0&1\\1&0\end{pmatrix}$ has <period of a state in a Markov chain>[period] two: its distributions alternate forever despite the existence of the uniform <Stationary distribution>. More generally, an <irreducible Markov chain>[irreducible] nearest-neighbour walk on a <bipartite graph> alternates its two vertex classes. A positive <holding probability of a Markov chain>[holding probability] at any state of an <communicating class>[irreducible class] supplies a one-step return and makes the entire class <aperiodic Markov chain>[aperiodic]. <aperiodic Markov chain>[Aperiodicity] therefore concerns possible return lengths rather than merely the existence of an invariant law.
Back to article page