The expectation of event occurrence (two-state model in CTMC)

98 Views Asked by At

Below problem is excerpted from Stochastic Processes (2e, Ross). The solution for 5.12(b) can be found here.

5.12 Suppose that the “state” of the system can be modeled as a two-state continuous-time Markov chain with transition rates $\nu_0=\lambda$, $\nu_1=\mu$ . When the state of the system is $i$, “events” occur in accordance with a Poisson process with rate $\alpha_i$, $i=0, 1$. Let $N(t)$ denote the number of events in $(0, t)$.

(b) If the initial state is state $0$, find $E\big(N(t)\big)$.

I am not sure about why $E(N(t))=\alpha_0 E(T_0(t))+\alpha_1 E(T_1(t))$, where $T_i(t)$ stands for the time spent in state $i$ before time $t$. It is quite obvious for Poisson process. But I fail to understand it in this case. (I know little about renewal theory, so is it possible to circumvent it?) Also, I think it is a Doubly stochastic Poisson Process (Cox process) instead of the typical conditional Poisson process. Is it true?

1

There are 1 best solutions below

0
On

Sketch of the solution: Let $M(t) = E(N_t|X_0=0)$. Conditioning on the state at $t$, we have $M^{'}(t)=\alpha_0P_{00}(t)+\alpha_1 P_{01}(t).$

The transition probability $P_{0i}$ can be solved via the Kolmogorov forward equation.