[STATISTICS-38] Optimise use of the Gamma function in the PoissonDistribution

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Implemented
• Affects Version/s: 1.0
• Fix Version/s: 1.0
• Component/s: distribution
• Assignee: Unassigned
• Reporter: Alex Herbert
• Created: 07/Oct/21 18:24
• Updated: 06/Dec/22 11:30
• Resolved: 22/Jan/22 13:21

Description

The Poisson distribution currently allows user configured tolerances for computing the Gamma functions for the CDF. These may not be necessary and it should be investigated. The version of the Gamma function used by the Poisson distribution may not be the same as
those that required these tolerances in Commons Math 2 (use of the tolerances go as far back as the commit history in the git repo).

Repeated calling of the RegularizedGamma.P and Q functions from numbers would benefit from allowing pre-computation of the the log( x ) component in the RegularizedGamma function. In this case x is the Poisson mean. See also NUMBERS-167 which discusses pre-computation of the LogGamma(a) component of the input arguments a and x.

Issue Links

• relates to: NUMBERS-174 Update Gamma functions using the Boost implementation