Counting sets with exceptions
The paper studies the tail of the hypergeometric distribution as a combinatorial problem and misses to mention the terms ``tail or ``hypergeometric distribution. The main result is that if \(N+E< X\) and and \(K\leq \min\{N,E\}\), then \(1-{NE\over (K+1)X} \leq {1\over {X\choose N}}\sum_{j=0}^K {X-E\choose N-j}{E\choose j}\), where all letters denote natural numbers. The proof cites Gauss' theorem for the evaluation of \({}_2F_1(a,b,c;1)\), but its strength is not needed. Since \(b\) is a negative integer for the purposes of the paper, the hypergeometric series is terminating, and the Vandermonde convolution \(\sum_{i=0}^n {m\choose i} {k\choose n-i} ={m+k\choose n}\) would suffice for the evaluation. Indeed, the identity in Proposition 2 has a simple proof as follows: \[ \begin{aligned} {1\over N+1}\sum_{K=0}^N\sum_{j=0}^K{X-E\choose N-j}{E\choose j} & = \sum_{i=0}^N {N+1-i\over N+1} {X-E\choose N-i}{E\choose i}\\ & = \sum_{i=0}^N{X-E\choose N-i}{E\choose i}-{1\over N+1} \sum_{i=1}^Ni{X-E\choose N-i}{E\choose i}\\ & = \sum_{i=0}^N{X-E\choose N-i}{E\choose i}-{E\over N+1}\sum_{i=1}^N{X-E\choose N-i}{E-1\choose i-1}\\ & ={X\choose N}-{E\over N+1}{X-1\choose N-1}\\ &= {X\choose N}\left(1-{EN\over X(N+1)}\right).\end{aligned} \]
- HyperQuick algorithm for discrete hypergeometric distribution
- Inequalities between hypergeometric tails
- Representations, bounds and approximations for tail probabilites of multivariate non-central hypergeometric and negative hypergeometric distributions
- Exponential bounds for the hypergeometric distribution
- scientific article; zbMATH DE number 4036896
This page was built for publication: Counting sets with exceptions
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4820849)