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

Hi all,

I read recently that for any three continuous random variables, X,Y and Z, the conditional densities are related by the following formula:

$p(x|y) = \int g(x| z) h(z | y ) dz $

where $p(x|y)$ is conditional density of X given Y = y and so on.

I have never come across this relation before. It seems similar to Chapman-Kolmogorov equation but the paper I found this relation in claims it is true for any 3 continuous random variables. Can anyone provide a proof of the above relation please?


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

This is not true in general since you would need the additional condition $g(x|z)=p(x|y,z)$ on the conditional densities for that formula to hold true. As a counter-example, consider $y,z,\epsilon$ all independent standard normal and let $x=y+z+\epsilon$. Then $x|y$ is a normal $N(y,2)$ (conditional expectation $y$ and variance 2). However, if you apply the formula you gave, that would yield a normal $N(0,3)$ for $x|y$ which is not correct (to verify this integrate $z$ out from the product of an $N(z,2)$ and an $N(0,1)$ for $x|z$ and $z|y$ respectively).

share|cite|improve this answer
Great counterexample; I checked all the details and you are spot on. Thanks very much for your help. (PS I am new to the forum, so I hope I have managed to successfully accept this as the answer to my question...) – quantnewbie Sep 6 '12 at 13:49
You are welcome! – an12 Sep 6 '12 at 23:44

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.