I am looking for a quick-to-compute approximation of the CDF of $X+Y$, where $X \sim N(0,\sigma_1^2)$ and $Y$ is a truncated gaussian, more specifically, a gaussian with mean $0$, standard deviation of $\sigma_2$, truncated on one side, at $a$ ($a < 0$). I have found a closed-form density function for the distribution of $X+Y$ here, but not a CDF.

Unfortunately, using the density function means that I have to numerically integrate it repeatedly, which is computationally very intensive, if done many times. Is there an approximation to the CDF of $X+Y$, which I could use instead?

I'd appreciate any suggestions or pointers on how to solve this problem.

  • 1
    $\begingroup$ Does the Gaussian CDF $\Phi$ qualify as a closed-form function for you? Then you could as well define a function $\Psi$ as the CDF of $X+Y$ and call that a closed-form solution. The only difference is that more is known about the former and better algorithms are available for evaluating it. $\endgroup$ – Eckhard Mar 28 '13 at 9:58
  • $\begingroup$ But how would you go about defining $\Psi$ given that Y is truncated on one side? $\endgroup$ – Pavel Mar 28 '13 at 11:22
  • $\begingroup$ @Eckhard: I suppose what I am really looking for is not necessarily a closed-form solution, but rather something that's quick to compute. I changed my question accordingly. $\endgroup$ – Pavel Mar 29 '13 at 11:17

Your Answer

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

Browse other questions tagged or ask your own question.