Jensen's Inequality

An inequality stating that for a convex function φ and an integrable random variable X (or measurable function) the convex function evaluated at the expectation (or integral average) is at most the expectation of the convex function: φ(E[X]) ≤ E[φ(X]).