A note on the expected values of powers of a matrix

Abstract It is known [Z] that if X is any positive random variable, then E (Xn)>[E (X)]n for any integer n, provided the expectations exist. A matrix generalization of this result for n = ‐1 was given in [1]. We will show that a simple exercise in matrix analysis yields the matrix generalization for any integer n, positive or negative.