Questions: Stationary Distributions

5 questions to test your understanding

Score: 0 / 5
Question 1 Multiple Choice

For a finite, irreducible, aperiodic Markov chain with stationary distribution π, you run the chain for a very long time starting from state i. As the number of steps n → ∞, the probability of being in state j approaches:

AP_{ij}, the one-step transition probability from i to j
Bπ_j, the stationary probability of state j, regardless of the starting state i
C1/k where k is the total number of states, since the chain visits all states equally
DThe probability depends on which state i you started in — there is no universal limit
Question 2 Multiple Choice

A Markov chain satisfies detailed balance: π_i P_{ij} = π_j P_{ji} for all states i and j. What can you conclude?

AThe chain is irreducible and will converge to π from any starting distribution
Bπ is a stationary distribution for the chain, but detailed balance alone does not guarantee irreducibility or convergence
CThe chain must be reversible and have uniform stationary distribution
DDetailed balance is equivalent to the chain being aperiodic
Question 3 True / False

If you start a Markov chain in its stationary distribution π, the distribution of the chain's state at every future time step remains π.

TTrue
FFalse
Question 4 True / False

A finite irreducible Markov chain can have two distinct stationary distributions if its transition probabilities are chosen carefully.

TTrue
FFalse
Question 5 Short Answer

Explain why the existence of a stationary distribution makes Markov Chain Monte Carlo (MCMC) a practical algorithm for sampling from complex probability distributions.

Think about your answer, then reveal below.