4
$\begingroup$

Take $n$ non-negative dependent random variables $X_1,...,X_n$ with $Pr(X_i \leq t) = t, t\in[0,1]$ for every $i$ (uniform marginal distributions).

What is an example of a joint pdf for $X_1,...,X_n$ (with the given common marginal distribution), such that $E[\min_i X_i] = 1/2$?

$\endgroup$
  • $\begingroup$ Got something from the answer below? $\endgroup$ – Did Apr 18 '18 at 8:17
2
$\begingroup$

Assume that $(X_1,X_2,\ldots,X_n)$ is a solution and let $M=\min\{X_1,X_2,\ldots,X_n\}$.

Then, for every $i$, $M\leqslant X_i$ almost surely and $\mathbb E(M)=\frac12=\mathbb E(X_i)$ hence $M=X_i$ almost surely. This holds for every $i$ hence $X_1=X_2=\cdots=X_n$ almost surely.

In which case, naturally, $\mathbb E(M)=\frac12$ indeed holds, and $(X_1,X_2,\ldots,X_n)$ has no joint PDF.

On the other hand, for every positive $\epsilon$, there exists $(X_1^{(\epsilon)},X_2^{(\epsilon)},\ldots,X_n^{(\epsilon)})$ with uniform marginals and with a joint PDF, such that $\mathbb E(M^{(\epsilon)})\geqslant\frac12-\epsilon$.

$\endgroup$
  • $\begingroup$ I agree that $M \leq X_i$, but how do you know that $E(M) = \frac{1}{2}$ immediately? That's what we are trying to obtain considering a certain joint probability distribution, because otherwise $E(M)$ can be something else. $\endgroup$ – onaheimi3 Apr 12 '18 at 14:45
  • $\begingroup$ Of course, but we assume that a given joint distribution yields E(M)=1/2, then we try to identify this distribution. $\endgroup$ – Did Apr 12 '18 at 17:13

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.