Approximate Discrete Entropy Monotonicity for Log-Concave Sums

Abstract: It is proved that for any

n g e q 1,

if

X_{1}, l d o t s, X_{n}

are i.i.d. integer-valued, log-concave random variables then

h(X_1+ldots+X_n + U_1+ldots+U_n) = H(X_1+ldots+X_n) + o(1),

as

H (X_{1}) o i n f t y

, where

h

stands for the differential entropy,

H

dentoes the (discrete) Shannon entropy and

U_{1}, l d o t s, U_{n}

are independent continuous uniforms on

(0, 1)

. As a corollary, it is shown that a conjecture of Tao (2010) holds true for log-concave random variables on the integers:

H(X_1+ldots+X_{n+1}) geq H(X_1+cdots+X_{n}) + frac{1}{2}log{Bigl(frac{n+1}{n}Bigr)} - o(1)

as

H (X_{1}) o i n f t y

. Explicit bounds for the

o (1)

-terms are provided.

This page was built for publication: Approximate Discrete Entropy Monotonicity for Log-Concave Sums