Let $X_i$'s ($i=1,..,n$) be i.i.d. random variables with mean $\mu$ and variance $\sigma^2$.

Is there a method that can be used to compute $\mathbb{E}[1/(X_1+...+X_n)]$?

  • 2
    $\begingroup$ No. The information is severely insufficient and the answer is "expectation is undefined" more often than not. $\endgroup$ – fedja Feb 13 '13 at 20:39
  • 1
    $\begingroup$ Since you do not specify the distribution of the $X_i$, I assume it is not known. Then the answer is no. $\endgroup$ – André Nicolas Feb 13 '13 at 20:39

Assuming the expectation does exist, and further assuming $X$ to be positive random variables: $$ \mathbb{E}\left(\frac{1}{X_1+\cdots+X_n}\right) = \mathbb{E}\left( \int_0^\infty \exp\left(-t (X_1+\cdots+X_n) \right) \mathrm{d}t \right) $$ Interchanging the integral over $t$ with expectation: $$ \mathbb{E}\left( \int_0^\infty \exp\left(-t (X_1+\cdots+X_n) \right) \mathrm{d}t \right) = \int_0^\infty \mathbb{E}\left(\exp\left(-t (X_1+\cdots+X_n) \right) \right) \mathrm{d}t $$ Using iid property: $$ \int_0^\infty \mathbb{E}\left(\exp\left(-t (X_1+\cdots+X_n) \right) \right) \mathrm{d}t = \int_0^\infty \mathbb{E}\left(\exp\left(-t X \right) \right)^n \mathrm{d}t $$ So should you know the Laplace generating function $\mathcal{L}_X(t) = \mathbb{E}\left(\mathrm{e}^{-t X} \right)$ we have: $$ \mathbb{E}\left(\frac{1}{X_1+\cdots+X_n}\right) = \int_0^\infty \mathcal{L}_X(t)^n \mathrm{d} t $$


As others have commented, you need more information. However, you do have an estimate (using convexity): assuming the random variables are positive, $$E\left[ \frac{1}{X_1 + \ldots + X_n} \right] \ge \frac{1}{E[X_1 + \ldots + X_n]} = \frac{1}{n\mu}$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.