A Markov chain deterministically alternates between states A and B. Does a stationary distribution exist? Does its…
Quant ResearchLevel: IntermediateStationarity versus convergence
A Markov chain deterministically alternates between states A and B. Does a stationary distribution exist? Does its distribution converge from A?
Hints
- Solve pi P=pi first.
- Then evolve a point-mass initial condition.
Answer
The stationary law is (1/2,1/2), unchanged by swapping coordinates. From A the law alternates between (1,0) and (0,1), so no convergence. Stationary existence is distinct from convergence; periodicity matters.
Source and licence: Lehman, Leighton & Meyer — Mathematics for Computer Science (2018-06-06) · CC-BY-SA-3.0 · answer: Quantoros editorial