I have a non-negative continuous random variable $X$ and I am given $P(x>a+b)=P(x>a)P(x>b)$. I need to prove that the probability distribution is exponential. I already know that $F(x)=0$ if $x \leq 0$ where $F(x)=P(X \leq x)$ so I should be able to show that otherwise $F(x)=1- \exp(-ax)$ for some constant a.

  • $\begingroup$ Can i say that the exponential distribution fits and such it must be unique fit? Seems such a rough way of phrasing it. $\endgroup$ – Niel May 25 '14 at 23:22
  • $\begingroup$ Have you considered the "memorylessness" of the exponential distribution? (Try rearranging the given equation to one that describes conditional probability.) $\endgroup$ – White Shirt May 25 '14 at 23:23
  • $\begingroup$ In essence, what you want to know is: if $f:(0,\infty) \to \mathbb R$ satisfies $f(a+b) = f(a) + f(b)$ (which you have assumed is the case if I set $f(a) = \log(1-F(a))$, and if $f$ satisfies some very mild extra hypothesis (e.g. measurable, or monotone), then $f$ is linear. This has been dealt with in many places. But I just cannot think of the correct phrase which will let google find it. Anyway, its is not really a problem in probability. Probably, in the form I present it, it has been answered elsewhere on stackexchange. $\endgroup$ – Stephen Montgomery-Smith May 25 '14 at 23:23
  • $\begingroup$ i think this might help me then, reading it now math.stackexchange.com/questions/152944/… $\endgroup$ – Niel May 25 '14 at 23:28
  • 1
    $\begingroup$ Cauchy functional equation. $\endgroup$ – Robert Israel May 25 '14 at 23:29

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.