Algorithm 955: Approximation of the Inverse Poisson Cumulative Distribution Function
Algorithm 955: Approximation of the Inverse Poisson Cumulative Distribution Function
复制标题
DOI：
10.1145/2699466
复制
发表时间：
2016-02-01
影响因子：
2.7
通讯作者：
Giles, Michael B.
中科院分区：
文献类型：
--
作者：
Giles, Michael B.
新的近似逆的不完整的伽玛函数推导，这是用来开发有效的评价的逆泊松累积分布函数。基于标准正态近似的渐近近似对于具有MIMD内核的CPU特别好，而对于GPU和具有向量单元的其他硬件，由于每个向量内的条件分支，基于Temme近似的不完整伽马函数的第二渐近近似更有效。在CPU和GPU上评估软件实现的准确性和效率。
New approximations for the inverse of the incomplete gamma function are derived, which are used to develop efficient evaluations of the inverse Poisson cumulative distribution function. An asymptotic approximation based on the standard Normal approximation is particularly good for CPUs with MIMD cores, while for GPUs and other hardware with vector units, a second asymptotic approximation based on Temme's approximation of the incomplete gamma function is more efficient due to conditional branching within each vector. The accuracy and efficiency of the software implementations is assessed on both CPUs and GPUs.