Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Suppose $\mu$ is an invariant measure for a Markov chain with state space $S$ with $\mu(i)p_{ij}=\mu(j)p_{ji}$ $\forall i,j \in S$. Describe a Markov chain with this property. Also, show that $\mu$ is an invariant measure.

In constructing the Markov chain, I know I want it to have transition matrix $P$ that satisfies $\mu'=\mu'P$ (prime denoting transpose). I can't think of a named markov chain off the top of my head with this property. Perhaps Gambler's Ruin or deterministically monotone?

To prove it's invariant measure, I need $\sum\limits_{i \in S} \mu(i) \neq 1$. Don't I need the chain before I can do this?

share|improve this question

1 Answer 1

up vote 1 down vote accepted
  • an example is the transition matrix$\bigl(\begin{smallmatrix} 0&1\\ 1&0 \end{smallmatrix} \bigr)$ with $\mu_1 = \mu_2 = \frac 12$.
  • If $\mu_i p(i,j) = \mu_j p(j,i) $ then summing over $i$ puts $$ \mu_i= \sum_j \mu_i p(i,j) = \sum_j \mu_j p(j,i) = (\mu p)_i $$that is: $\mu$ is invariant.
share|improve this answer
    
How do I know that $(\mu p)_i \neq 1$? –  user113525 Mar 13 at 2:01
    
why do you need this condition? –  mookid Mar 13 at 2:02
    
The definition for invariant measure I was given in lecture said that this is what made it so. –  user113525 Mar 13 at 2:04
    
I have never seen such a condition... for me all there is is: $\mu_i\ge 0$ and $\mu P = \mu$ –  mookid Mar 13 at 2:06
    
I am reading in my textbook now that my lecturer was wrong: $\mu_j=E_i \sum\limits_{0\leq n \leq \tau_i(1)-1} 1_{[X_n=j]}$ qualifies as invariant. What you are saying is stationary distribution (I think). –  user113525 Mar 13 at 2:07

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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