0
$\begingroup$

(Cross-post from math.stackexchange.com Q#166689)

I would like to lower-bound $E[X Y]$ where $X, Y$ are two random variables such that:

  1. $X \in [x_0, 1], Y \in [y_0, 1]$

  2. $E[X] = x, E[Y] = y$

  3. $X \geq Y^k$

Here $x_0, y_0, x > x_0, y > y_0, k$ are known constants. There is the trivial bound $E[X Y] \geq x y_0, \geq x_0 y$. Are any better bounds available?

$\endgroup$
1
  • 1
    $\begingroup$ Under what assumption? If $x=x_0$ and y=y_0,$ then your bound is sharp... $\endgroup$
    – Igor Rivin
    Commented Jul 9, 2012 at 16:20

1 Answer 1

1
$\begingroup$

I'll assume $x_0, y_0 \ge 0$. Presumably $k \ge 0$, since if $k < 0$ the only way to have $X \ge Y^k$ with $0 \le X,Y \le 1$ is $X=Y=1$. Since $X \ge Y^k$, Jensen's inequality says $E[XY] \ge E[Y^{k+1}] \ge y^{k+1}$

$\endgroup$
1
  • $\begingroup$ That's not obviously a better bound than what the OP states... (depends on the values of $x_0$ and $k$... $\endgroup$
    – Igor Rivin
    Commented Jul 9, 2012 at 18:28

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .