In: Statistics and Probability
P(Xi+1 = rice| Xi-1 = rice, Xi = rice) = 0.7
P(Xi+1 = rice| Xi-1 = noodles, Xi = rice) = 0.6
P(Xi+1 = rice| Xi-1 = rice, Xi = noodles) = 0.3
P(Xi+1 = rice| Xi-1 = noodles, Xi = noodles) = 0.55
Q1. Is {Xn} a Markov Chains? Why?
Q2. How to transform the process into a M.C. ?
1.
Since the transition probabilities to the state "rice" or "noodles" depends on last two states, {Xn} is not a Markov Chains.
2.
Let there would be four states - (rice, rice) , (noodles, rice), (rice, noodles) , (noodles, noodles) which denotes the lunch condition on last two days (i, i-1).
The transition probability from state (rice, rice) to (rice, rice) is,
P(Yi+1 = (rice, rice) | Yi = (rice, rice) ) = (P(Xi+1 = rice | Xi-1 = rice, Xi = rice) = 0.7
P(Yi+1 = (rice, noodles) | Yi = (rice, rice) ) = 0
P(Yi+1 = (noodles, rice) | Yi = (rice, rice) ) = (P(Xi+1 = noodles | Xi-1 = rice, Xi = rice) = 1 - (P(Xi+1 = rice | Xi-1 = rice, Xi = rice) = 1 - 0.7 = 0.3
P(Yi+1 = (noodles, noodles) | Yi = (rice, rice) ) = 0
P(Yi+1 = (rice, rice) | Yi = (rice, noodles) ) = P(Xi+1 = rice| Xi-1 = noodles, Xi = rice) = 0.6
P(Yi+1 = (rice, noodles) | Yi = (rice, noodles) ) = 0
P(Yi+1 = (noodles, rice) | Yi = (rice, noodles) ) = P(Xi+1 = noodles| Xi-1 = noodles, Xi = rice) = 1 - P(Xi+1 = rice| Xi-1 = noodles, Xi = rice) = 1 - 0.6 = 0.4
P(Yi+1 = (noodles, noodles) | Yi = (rice, noodles) ) = 0
P(Yi+1 = (rice, rice) | Yi = (noodles, rice) ) = 0
P(Yi+1 = (rice, noodles) | Yi = (noodles, rice) ) = P(Xi+1 = rice| Xi-1 = rice, Xi = noodles) = 0.3
P(Yi+1 = (noodles, rice) | Yi = (noodles, rice) ) = 0
P(Yi+1 = (noodles, noodles) | Yi = (noodles, rice) ) = P(Xi+1 = noodles| Xi-1 = rice, Xi = noodles) = 1 - P(Xi+1 = rice| Xi-1 = rice, Xi = noodles) = 1 - 0.3 = 0.7
P(Yi+1 = (rice, rice) | Yi = (noodles, noodles) ) = 0
P(Yi+1 = (rice, noodles) | Yi = (noodles, noodles) ) = P(Xi+1 = rice| Xi-1 = noodles, Xi = noodles) = 0.55
P(Yi+1 = (noodles, rice) | Yi = (noodles, noodles) ) = 0
P(Yi+1 = (noodles, noodles) | Yi = (noodles, noodles) ) = P(Xi+1 = noodles| Xi-1 = noodles, Xi = noodles) = 1 - P(Xi+1 = rice| Xi-1 = noodles, Xi = noodles) = 1 - 0.55 = 0.45
For all the states, the transition probabilities to the other states only depends on last states, and hence {Yn} is a Markov chain.