Poisson limit theorem (source code)

= Poisson limit theorem
{c}
{wiki}

Let $B_n$ have the <binomial distribution> with trial count $m_n\to\infty$ and success <Probability> $p_n\to0$. If $m_np_n\to\lambda\in[0,\infty)$, then
$$
B_n\xrightarrow{d}\operatorname{Poisson}(\lambda).
$$
This is the <Poisson limit theorem>: many <independent> rare trials retain a finite total expected count. For a proof using <characteristic functions>,
$$
\mathbb E e^{itB_n}=[1+p_n(e^{it}-1)]^{m_n}
\longrightarrow\exp[\lambda(e^{it}-1)].
$$
The logarithmic expansion error vanishes because $m_np_n^2=(m_np_n)p_n\to0$. The limit is the <Poisson distribution> <characteristic function>, so the <Lévy continuity theorem> proves convergence. The case $\lambda=0$ is the degenerate count zero; it also follows directly from <Markov's inequality>.<Probability: Theory and Examples (fifth edition) by Rick Durrett>[Section 3.6, Theorem 3.6.1 and its characteristic-function proof]{ref}

For example, $\operatorname{Bin}(100,0.01)$ is approximated by $\operatorname{Poisson}(1)$, and its zero-count <Probability> $0.99^{100}$ is about $0.366$, close to $e^{-1}\approx0.368$. The limiting <mean> is the limit of the product, not either individual parameter. <independent> unequal-probability Bernoulli trials have a related Poisson approximation when their largest success <Probability> vanishes and their total <mean> converges; dependence requires additional control, as in the <Stein-Chen method>.