for any two independent, multivariate random variables X
and Z. Shannon demonstrated extraordinary insight in
proposing this inequality. He used it to establish a lower
bound to the capacity of the band-limited channel with
arbitrarily distributed additive noise. Shannon provided a
variational argument to show that the entropy power of the
sum of two independent random vectors of given entropy
powers has zero variation with respect to the probability
distributions of the added vectors if the two random vectors
are Gaussian with proportional covariance matrices.