11.2.7 Solved Problems
Consider the Markov chain with three states, $S=\{1, 2, 3 \}$, that has the following transition matrix \begin{equation} \nonumber P = \begin{bmatrix} \frac{1}{2} & \frac{1}{4} & \frac{1}{4} \\[5pt] \frac{1}{3} & 0 & \frac{2}{3} \\[5pt] \frac{1}{2} & \frac{1}{2} & 0 \end{bmatrix}. \end{equation}
- Draw the state transition diagram for this chain.
- If we know $P(X_1=1)=P(X_1=2)=\frac{1}{4}$, find $P(X_1=3,X_2=2,X_3=1)$.
- Solution
-
- The state transition diagram is shown in Figure 11.6
- First, we obtain \begin{align*} P(X_1=3)&=1-P(X_1=1)-P(X_1=2) \\ &=1-\frac{1}{4}-\frac{1}{4}\\ &=\frac{1}{2}. \end{align*} We can now write \begin{align*} P(X_1=3,X_2=2,X_3=1)&=P(X_1=3) \cdot p_{32} \cdot p_{21} \\ &=\frac{1}{2} \cdot \frac{1}{2} \cdot \frac{1}{3}\\ &=\frac{1}{12}. \end{align*}
- The state transition diagram is shown in Figure 11.6
-
Problem
Consider the Markov chain in Figure 11.17. There are two recurrent classes, $R_1=\{1,2\}$, and $R_2=\{5,6,7\}$. Assuming $X_0=3$, find the probability that the chain gets absorbed in $R_1$.
- Solution
-
Here, we can replace each recurrent class with one absorbing state. The resulting state diagram is shown in Figure 11.18
-
Here, we can replace each recurrent class with one absorbing state. The resulting state diagram is shown in Figure 11.18
Problem
Consider the Markov chain of Example 2. Again assume $X_0=3$. We would like to find the expected time (number of steps) until the chain gets absorbed in $R_1$ or $R_2$. More specifically, let $T$ be the absorption time, i.e., the first time the chain visits a state in $R_1$ or $R_2$. We would like to find $E[T|X_0=3]$.
- Solution
- Here we follow our standard procedure for finding mean hitting times. Consider Figure 11.18. Let $T$ be the first time the chain visits $R_1$ or $R_2$. For all $i \in S$, define \begin{align*} t_i&=E[T |X_0=i]. \end{align*} By the above definition, we have $t_{R_1}=t_{R_2}=0$. To find $t_3$ and $t_4$, we can use the following equations \begin{align*} t_i &=1+\sum_{k} t_k p_{ik}, \quad \textrm{ for }i=3, 4. \end{align*} Specifically, we obtain \begin{align*} t_3 &=1+\frac{1}{2} t_{R_1}+ \frac{1}{2} t_4\\ &=1+ \frac{1}{2} t_4, \end{align*} \begin{align*} t_4 &=1+\frac{1}{4} t_{R_1}+ \frac{1}{4} t_3+\frac{1}{2} t_{R_2}\\ &=1+ \frac{1}{4} t_3. \end{align*} Solving the above equations, we obtain \begin{align*} t_3 =\frac{12}{7}, \quad t_4=\frac{10}{7}. \end{align*} Therefore, if $X_0=3$, it will take on average $\frac{12}{7}$ steps until the chain gets absorbed in $R_1$ or $R_2$.
Problem
Consider the Markov chain shown in Figure 11.19. Assume $X_0=1$, and let $R$ be the first time that the chain returns to state $1$, i.e., \begin{align*} R = \min \{n \geq 1: X_n=1 \}. \end{align*} Find $E[R|X_0=1]$.
- Solution
- In this question, we are asked to find the mean return time to state $1$. Let $r_1$ be the mean return time to state $1$, i.e., $r_1=E[R|X_0=1]$. Then \begin{align*} r_1 &=1+\sum_{k} t_k p_{1k}, \end{align*} where $t_k$ is the expected time until the chain hits state $1$ given $X_0=k$. Specifically, \begin{align*} t_1&=0,\\ t_k&=1+\sum_{j} t_j p_{kj}, \quad \textrm{ for } k\neq 1. \end{align*} So, let's first find $t_k$'s. We obtain \begin{align*} t_2 &=1+\frac{1}{3} t_1+ \frac{2}{3} t_3\\ &=1+ \frac{2}{3} t_3, \end{align*} \begin{align*} t_3 &=1+\frac{1}{2} t_{3}+ \frac{1}{2} t_1\\ &=1+\frac{1}{2} t_{3}. \end{align*} Solving the above equations, we obtain \begin{align*} t_3 =2, \quad t_2=\frac{7}{3}. \end{align*} Now, we can write \begin{align*} r_1 &=1+\frac{1}{4} t_{1}+ \frac{1}{2} t_2+\frac{1}{4} t_{3}\\ &=1+\frac{1}{4} \cdot 0+ \frac{1}{2} \cdot \frac{7}{3}+\frac{1}{4} \cdot 2\\ &=\frac{8}{3}. \end{align*}
Problem
Consider the Markov chain shown in Figure 11.20.
- Is this chain irreducible?
- Is this chain aperiodic?
- Find the stationary distribution for this chain.
- Is the stationary distribution a limiting distribution for the chain?
- Solution
-
- The chain is irreducible since we can go from any state to any other states in a finite number of steps.
- The chain is aperiodic since there is a self-transition, i.e., $p_{11}>0$.
- To find the stationary distribution, we need to solve \begin{align*} & \pi_1 =\frac{1}{2} \pi_1+\frac{1}{3} \pi_2+\frac{1}{2} \pi_3, \\ & \pi_2 =\frac{1}{4} \pi_1+\frac{1}{2} \pi_3,\\ & \pi_3 =\frac{1}{4} \pi_1+\frac{2}{3} \pi_2, \\ & \pi_1+\pi_2+\pi_3=1. \end{align*} We find \begin{align*} \pi_1 \approx 0.457, \; \pi_2 \approx 0.257, \; \pi_3 \approx 0.286 \end{align*}
- The above stationary distribution is a limiting distribution for the chain because the chain is irreducible and aperiodic.
-
Problem
Consider the Markov chain shown in Figure 11.21. Assume that $\frac{1}{2} \lt p \lt 1$. Does this chain have a limiting distribution? For all $i,j \in \{0,1,2, \cdots \}$, find \begin{align*} \lim_{n \rightarrow \infty} P(X_n=j |X_0=i). \end{align*}
- Solution
- This chain is irreducible since all states communicate with each other. It is also aperiodic since it includes a self-transition, $P_{00}>0$. Let's write the equations for a stationary distribution. For state $0$, we can write \begin{align*} \pi_0 &=(1-p)\pi_0+(1-p) \pi_1, \end{align*} which results in \begin{align*} \pi_1 &=\frac{p}{1-p}\pi_0. \end{align*} For state $1$, we can write \begin{align*} \pi_1 &= p \pi_0+(1-p) \pi_2\\ &=(1-p) \pi_1+(1-p) \pi_2, \end{align*} which results in \begin{align*} \pi_2 &=\frac{p}{1-p}\pi_1. \end{align*} Similarly, for any $j \in \{1,2,\cdots \}$, we obtain \begin{align*} \pi_{j} &=\alpha \pi_{j-1}, \end{align*} where $\alpha=\frac{p}{1-p}$. Note that since $\frac{1}{2} \lt p \lt 1$, we conclude that $\alpha>1$. We obtain \begin{align*} \pi_{j} &=\alpha^j \pi_{0}, \quad \textrm{ for $j= 1,2,\cdots $ }. \end{align*} Finally, we must have \begin{align*} 1 &=\sum_{j=0}^{\infty} \pi_j\\ &= \sum_{j=0}^{\infty} \alpha^j \pi_0, &(\textrm{where } \alpha> 1)\\ &=\infty \pi_0 . \end{align*} Therefore, the above equation cannot be satisfied if $\pi_0>0$. If $\pi_0=0$, then all $\pi_j$'s must be zero, so they cannot sum to $1$. We conclude that there is no stationary distribution. This means that either all states are transient, or all states are null recurrent. In either case, we have \begin{align*} \lim_{n \rightarrow \infty} P(X_n=j |X_0=i)=0, \textrm{ for all }i,j. \end{align*} We will see how to figure out if the states are transient or null recurrent in the End of Chapter Problems (see Problem 15 in Section 11.5).