The acceptance-rejection algorithm is often used to sample from non-standard distributions. For this algorithm to be efficient, however, the user has to create a hat function that majorizes and closely matches the density of the distribution to be sampled from. There are many methods for automatically creating such hat functions, but these methods require that the user transforms the density so that she knows the exact location of the transformed density's inflection points. In this paper, we propose an acceptancerejection algorithm which obviates this need and can thus be used to sample from a larger class of distributions.

Original language | English |
---|

Publication status | Published - 1 Aug 2011 |
---|

Series | Research Report Series / Department of Statistics and Mathematics |
---|

Number | 110 |
---|

- Research Report Series / Department of Statistics and Mathematics