Getting confused with how to apply inequality

67 Views Asked by At

A random variable X is defined by the transformation Z = log(X), where the mean of the random variable Z is E(Z) = 0. Is E(X) greater than, less than, or equal to 1?

I know we need to apply Jensens Inequality, And I know that $X= e^Z$. But am getting really confused on what is E(X).

1

There are 1 best solutions below

0
On

Jensen's inequality should be thought of independently of expectations and probabilities.

It's just a statement that the weighted average of a convex function evaluated at a number of points is at least as great as the function evaluated at the weighted average of the points. So for a convex function $f$, this means that$$ w_1 \, f(x_1) + w_2 \, f(x_2) + \ldots + w_n \, f(x_n) \ge f(w_1 x_1 + w_2 x_2 + \ldots + w_n x_n) $$ where $$ w_i \ge 0 \text{ and } \sum_{i=1}^n w_i = 1. $$

If you pick $n=2$, you can see this immediately by noticing that a convex function always lies below the line connecting any two of its points Convex function lies below the line connecting two of its points

Tying back to probabilities and expectations, notice that expectations are weighted averages, and that the function $f(x) = \exp(x)$ is convex. So the expectation of $f(x)$ will be greater than $f( \text{expectation}(x) )$.