If f is convex and x is random variable, then f(E(x)) <= E(f(x))
my opinion : For any positive epsilon, if sum(epsilon) = 1, for any positive x_n,
f(epsilon_1 * x_1 + …. epsilon_n * x_n) <= epsilon_1 * f(x_1) + … + epsilon_n * f(x_n)
If f is convex function then f satisfies jensen’s inequality