$$X\ \sim Exp(U) ~ and\ U\ \sim U(0,1) $$

The question asked for the value of $ P(X\geqslant 1)$

I saw the solution and it went like this:

$$P(X\geqslant 1) = E[P(X\geqslant 1)|U] = E[e^{-u}] = \int_{0}^{1}e^{-x}dx $$

Could someone explain the solution to me?

More specifically how did the solution derive each step?


1 Answer 1


You are given that the conditional distribution of $X$ given $U=u$ is $\text{Exp}(u)$ (presumably $u$ is the rate parameter), where $U$ itself is uniformly distributed on $(0,1)$.

The solution is an application of the law of total expectation, or equivalently, the law of total probability.

Suppose $I(X\ge 1)$ is an indicator variable that equals $1$ if $X\ge 1$ and equals $0$ otherwise. And let the density of $U$ be $f_U$. Then,

\begin{align} P(X\ge 1)&=E\left[I(X\ge 1)\right] \\&=E\left[E\left[I(X\ge 1\mid U\right]\right] \\&=E\left[P(X\ge 1\mid U)\right] \\&=E\left[e^{-U}\right] \\&=\int e^{-x}f_U(x)\,dx \\&=\int_0^1 e^{-x}\,dx \end{align}

In the penultimate step, law of the unconscious statistician was used to find the expectation.

  • $\begingroup$ Yes I think I understand this better now. Would it be equally accurate if I wrote it this way? P(X≥1) =∫P(X≥1|U)fu(u) du = E[P(X≥1∣U)] $\endgroup$
    – Grey Han
    Apr 24 at 2:34
  • $\begingroup$ You can directly say $P(X\ge 1)=\int P(X\ge 1\mid U=x)f_U(x)\,dx$ by law of total probability. $\endgroup$ Apr 24 at 5:45

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.