New algorithms for generating Poisson variates (Q915333)

The author discusses three algorithms for generating Poisson (\(\lambda\)) random variates. The first one is the KPFRAC algorithm, designed for \(\lambda <0\), the second one is KPLOW, for \(0<\lambda \leq 30\) and the third one is KEMPOIS, for 10\(\leq \lambda \leq 5000\). The conclusions of the work is that these algorithms prove that new features can be incorporated into the search approach and make it more competitive. A mixed routine which uses KPLOW, when \(\lambda <30\) and KEMPOIS, when \(\lambda\geq 30\) is indicated and so the speed of calculus is higher.