0
$\begingroup$

I read about Markov chains in quite a lot of different resources. However, I can't seem to find a consistent definition of what the requirements are for a Markov chain to have a stationary distribution. (Somewhere, I found that it has to be ergodic and aperiodic, but then again I read that for Markov chain to be ergodic, it must be aperiodic.) And is there any additional requirement so that the stationary distribution is unique? (I found that an irreducible chain has a unique stationary distribution, but how does this connect to the ergodicity?)

I have to admit I'm a bit confused by the various concepts of a chain being ergodic, irreducible, aperiodic, recursive, or fulfilling detailed balance.

Can anyone help me understand all these concepts and definitions better or point me to a good source on this topic?

$\endgroup$
1
$\begingroup$

Aperiodic, Ergodic, and Stationary are all related. However, a MC does not need to be Aperiodic or Ergodic to be Stationary. Stationary simply means what it seems to mean-- the probabilities aren't changing over time. Ergodic chains are both Aperiodic and Positive Reccurent. But there are non-ergodic stationary MCs.

$\endgroup$

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.