Entropy power inequality

In information theory, the entropy power inequality (EPI) is a result that relates to so-called "entropy power" of random variables. It shows that the entropy power of suitably well-behaved random variables is a superadditive function. The entropy power inequality was proved in 1948 by Claude Shannon in his seminal paper "A Mathematical Theory of Communication". Shannon also provided a sufficient condition for equality to hold; Stam (1959) showed that the condition is in fact necessary.

Statement of the inequality

For a random vector $X:\Omega \to {ℝ}^n$ with probability density function $f:{ℝ}^n\to ℝ$, the differential entropy of $X$, denoted $h(X)$, is defined to be

$h(X)=-\underset{{ℝ}^n}{\int }f(x)\log f(x)dx$

and the entropy power of $X$, denoted $N(X)$, is defined to be

$N(X)=\frac{1}{2\pi e}{e}^{\frac{2}{n}h(X)}.$

In particular, $N(X)=|K{|}^{1/n}$ when $X$ is normally distributed with covariance matrix $K$.

Let $X$ and $Y$ be independent random variables with probability density functions in the $L^p$ space $L^p({ℝ}^n)$ for some $p>1$. Then

$N(X+Y)\ge N(X)+N(Y).$

Moreover, equality holds if and only if $X$ and $Y$ are multivariate normal random variables with proportional covariance matrices.

Alternative form of the inequality

The entropy power inequality can be rewritten in an equivalent form that does not explicitly depend on the definition of entropy power (see Costa and Cover reference below).

Let $X$ and $Y$ be independent random variables, as above. Then, let $X^{\prime }$ and $Y^{\prime }$ be independent random variables with Gaussian distributions and proportional covariance matrices such that

$h(X^{\prime })=h(X)$ and $h(Y^{\prime })=h(Y)$

Then,

$h(X+Y)\ge h(X^{\prime }+Y^{\prime })$

See also

• Information entropy
• Information theory
• Limiting density of discrete points
• Self-information
• Kullback–Leibler divergence
• Entropy estimation