Simple set cardinality estimation through random sampling
Marco Bressan, Enoch Peserico, Luca Pretto · arXiv (Cornell University) · 2015
We present a simple algorithm that estimates the cardinality $n$ of a set $V$ when allowed to sample elements of $V$ uniformly and independently at random. Our algorithm with probability $(1-δ)$ returns a $(1\pmε)-$approximation of $n$ drawing $O\big(\sqrt{n} \cdot ε^{-1}\sqrt{\log(δ^{-1})}\big)$ samples (for $ε^{-1}\sqrt{\log(δ^{-1})} = O(\sqrt{n})$).