Optimal quantization for discrete distributions (2008.03255v2)

Abstract: In this paper, we first determine the optimal sets of $n$-means and the $n$th quantization errors for all $1\leq n\leq 6$ for two nonuniform discrete distributions with support the set ${1, 2, 3, 4, 5, 6}$. Then, for a probability distribution $P$ with support ${\frac 1n : n\in \mathbb N}$ associated with a mass function $f$, given by $f(x)=\frac 1 {2^k}$ if $x=\frac 1 k$ for $k\in \mathbb N$, and zero otherwise, we determine the optimal sets of $n$-means and the $n$th quantization errors for all positive integers up to $n=300$. Further, for a probability distribution $P$ with support the set $\mathbb N$ of natural number associated with a mass function $f$, given by $f(x)=\frac 1 {2^k}$ if $x=k$ for $k\in \mathbb N$, and zero otherwise, we determine the optimal sets of $n$-means and the $n$th quantization errors for all positive integers $n$. At last we discuss for a discrete distribution, if the optimal sets are given, how to obtain the probability distributions.