Example: consider a sequence of independent coin tosses with probability of Heads on a single toss. Let be number of heads minus number of tails after tosses. Put .
is a Markov Chain. State space is , the integers and
This chain has one communicating class (for ) and all states have period 2. According to the strong law of large numbers converges to . If this guarantees that for all large enough , that is, the number of returns to 0 is not infinite. The state 0 is then transient and so all states must be transient.
For the situation is different. It is a fact that
Local Central Limit Theorem (normal approximation to ) (or Stirling's approximation) shows
Start irreducible recurrent chain in state . Let be first such that . Define
Example
Notice stationary initial distribution is
Heuristic: start chain in . Expect to return to every time units. So are in state about once every time units; i.e. limiting fraction of time in state is .
Conclusion: for an irreducible recurrent finite state space Markov chain
Previous conclusion is still right if there is a stationary initial distribution.
Example: HeadsTails after tosses of fair coin. Equations are
Some observations:
You have to go through 1 to get to 0 from 2 so
Symmetry (switching H and T):
The transition probabilities are homogeneous:
Conclusion:
Summary of the situation:
Every state is recurrent.
All the expected hitting times are infinite.
All entries converge to 0.
Jargon: The states in this chain are null recurrent.
Page 229, question 21: runner goes from front or back door, prob 1/2 each. Returns front or back, prob 1/2 each. Has pairs of shoes, wears pair if any at departure door, leaves at return door. No shoes? Barefoot. Long run fraction of time barefoot?
Solution: Let be number of shoes at front door on day . Then is a Markov Chain.
Transition probabilities:
pairs at front door on day . is if goes out back door (prob is 1/2) or out front door and back in front door (prob is 1/4). Otherwise is .
pairs at front door on day . is if out back, in front (prob is 1/4). is if out front, in back. Otherwise is .
0 pairs at front door on day . is 0 if out front door (prob 1/2) or out back door and in back door (prob 1/4) otherwise is .
Transition matrix :
Doubly stochastic: row sums and column sums are 1.
So for all is stationary initial distribution.
Solution to problem: 1 day in no shoes at front door. Half of those go barefoot. Also 1 day in all shoes at front door; go barefoot half of these days. Overall go barefoot of the time.
Insurance company's reserves fluctuate: sometimes up, sometimes down. Ruin is event they hit 0 (company goes bankrupt). General problem. For given model of fluctuation compute probability of ruin either eventually or in next time units.
Simplest model: gambling on Red at Casino. Bet $1 at a time. Win $1 with probability , lose $1 with probability . Start with dollars. Quit playing when down to $0 or up to . Compute
= fortune after plays. .
Transition matrix:
Middle equation is
For we get
Notice that if our formulas for the sum of the geometric series are wrong. But for we get
For or and or 4:
For or
For first step analysis:
In matrix form
Translate to matrix notation:
Solution is