Take the 2-minute tour ×
Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

Suppose we have a continuous-time stochastic process $X(t)$, which consists of a sequence of delta functions that, at each time $t$, have a probability $p(t)$ of taking a non-zero value. $p(t)$ lies in $[0,1]$ for all $t$ and just a scaled survival function of an exponential distribution, so $p(t)$ is decreasing in time. We are interested in the random variable $Y = \int_0^\infty X(t) f(t) dt$ where $f(t)$ is a (deterministic) decreasing function of time. Is it possible to take the expectation of this integral; that is, $E\left[Y\right] = E\left[\int_0^\infty X(t) f(t) dt\right]$?

share|improve this question
It is not immediately clear from your description that the integral would exist in any reasonable sense, much less that one would be able to take an expectation. Why should, for example, $\{t: X(t,\omega) > 0 \}$ be a Lebesgue measurable set for any fixed $\omega$? Perhaps I have missed some subtlety in your description. –  cardinal Jan 12 '13 at 22:13
Thank you for your help, @cardinal, and I don't think you have missed any subtlety in my description -- this problem goes a bit beyond my undergraduate training in statistics, so let me explain the problem more fully. I have a poisson process of rate $\lambda$ that lasts until $t=U$, where $U$ is a draw from an exponential distribution. Each event in the process is a purchase of (deterministic) value m. The random variable I am interested in is the discounted value of all purchases. That is, when a purchase at time $t$ is made, the value is $m \times exp(\delta t)$, to account for discounting –  Eli Stein Jan 15 '13 at 15:23
where $\delta$ is a discount factor. I want to take the expected value of this random variable, call it $S$. I am having a hard time finding the expected value of $S$. One way I tried to formulate this problem was defining $S$ as the integral above, where $X(t)$ takes the value $1$ if a purchase is made "epsilon close" to $t$ and $f(t) = exp(\delta t)$. Now, if the expectation operator could be passed through the integral (which, I know, it can't), this problem can be solved as $E[X(t)] = \lambda \times dt \times S(t)$, where $S(t)$ is the survival function for an exponential distribution. As –  Eli Stein Jan 15 '13 at 15:32
the expectation operator can't be passed through the integral, however, I am a bit stuck. Any ideas? Thank you for your help! –  Eli Stein Jan 15 '13 at 15:32
add comment

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.