MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

How do I compute the following?

$$ \mathrm{Prob}\left( \sum_{i=1}^N x_i >1\right) = ?$$

where $x_i \sim \mathrm{Log}\mathcal{-N}(\mu_i, \sigma_i^2)$.

AFAIK, we do not know how the sum of log-normal distributions is distributed, so what I have in mind is to pick an appropriate interval for $x$, sample uniformly from there, and calculate sum of $x_i$s. If it's greater than 1, then I count it as 1. Am I doing it right? Is this the "Monte Carlo Method"? Are there any (more appropriate) other methods you can suggest? I wanted to get some experts' words to be sure. Thanks.

share|cite|improve this question
up vote 1 down vote accepted

The following paper: "The Distribution of Products of Beta, Gamma and Gaussian Random Variables" Author(s): M. D. Springer and W. E. Thompson (on JSTOR)

gives a solution for the problem in the case of mean-zero independent normal random variables, using the Mellin transform.

In this case the density is given using Meijer's G-function.

share|cite|improve this answer

Is $N$ large (where "large" is probably "bigger than five", in practice)? If it is, the central limit theorem gives a good approximation to the distribution of $\sum_{i=1}^N x_i,$ assuming some weak conditions on the variances (something like sum of the variances (see Feller, vol 2, page 251-252, or, Theorem 8)

share|cite|improve this answer

Your Answer


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

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