Regular Transition Matrix A transition matrix $P$ is said to be regular if some power of $P$ is positive. That is, $P^n >0$,for some $n≥1$.

My question is: Do all regular matrices have limiting distributions? And do all matrices with limiting distributions imply that they are regular?


Since the Markov chain is regular, there exists a $k$ such that for all states $i,j$ in the state space $P^k(i,j) > 0$. Naturally, a regular Markov chain is then irreducible.

Now, since $P^k > 0$ for all $i,j$, then $P^{k+1}$ has entries $$\pi_{ij}^{k+1} = \sum_t \pi_{it} \pi_{tj}^{k} > 0 \,, $$ since for at least one $t$, $\pi_{it} > 0$. This means that $P^{k+1}$ > 0 as well, and so is $P^{k+2}, P^{k+3},...$

The gcd of $\{k, k+1, k+2, \dots\}$ is 1, and thus a regular Markov chain is also aperiodic. Thus, regular Markov chains are irreducible and aperiodic which implies, the Markov chain has a unique limiting distribution.

Conversely, all matrices with a limiting distribution do not imply that they are regular. A counter-example is the example here, where the transition matrix is upper triangular, and thus the transition matrix for every step is upper triangular (and hence not regular), but a limiting distribution exists.

  • 1
    $\begingroup$ +1. It would be good to draw the conclusion explicitly: irreducible and aperiodic imply a unique limiting distribution. $\endgroup$ – whuber Feb 7 '18 at 14:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.