Consider a discrete Markov source, ona finite alphabet set. Let the initial distribution be and the transition probability for the step be . When can we say that is stationary?

Clearly, the source has to be time invariant and thefore we need . For to be stationary, we need

etc, where is the distribution. But . Thus guarantees that all ‘s have the same distribution. Now, consider, say, and .

It is clear that for the two joint distributions to be equal, is enough and therefore is sufficient.

Advertisements

September 15, 2009 at 3:30 pm

Why don’t you write more often man!

Became nostalgic after reading this one..