In AI book by Norvig and Russell ergodic Markov Chains are defined as follows:
- Every state is reachable from every other.
- There are no strictly periodic cycles in it.
Can someone explain what is strictly periodic cycle or give any reference?
In AI book by Norvig and Russell ergodic Markov Chains are defined as follows:
Can someone explain what is strictly periodic cycle or give any reference?
After some reading on wikipedia I believe what they mean is that for any state $i$ and time $0$, the times $t$ where you can reach $i$ again is not limited to a multiple of any integer greater than $1$. For instance, a random walk is periodic with period $2$ (you cannot come back to where you started after an odd number of moves; it has to be even).
In contrast, say we have a normal 1-D random walk (infinite or finite), but with the single change that if the walker is on square $100$, he can also choose to stand still in addition to moving left or right. Then it's not periodic, since there's a positive probability that he'll be back at the origin at the odd time $201$, even though up until that point it has seemed periodic.
This is more an account of the wikipedia definition than it addresses your concern regarding the word strictly, but I hope it helps you some part along the way.