Hi,

I'm submitting a file that implements the standard 3 functions for a Poisson distribution:

pdf-poisson (*)
cdf-poisson
random-poisson

(*) For discrete distributions, this is properly a probability mass function (pmf). Would it be helpful or annoying to call this pmf-poisson?

The random-poisson function implements a pair of methods for generating Poisson-distributed random variates so it's reasonably fast at both small and large mu.

The pdf-poisson function has a very direct implementation.

The cdf-poisson function uses the incomplete gamma function that I submitted in a separate submission. I can re-send it if you like.

Thanks!
Eric

Logged In: YES 
user_id=400048
Originator: NO

Incorporated this code

File Deleted - 259984: 



File Added - 260327: poisson.zip

Logged In: YES 
user_id=816411
Originator: YES

I'm submitting a collection of files as a zip archive. These include:
 * poisson.tcl
    - pdf-poisson
    - cdf-poisson
    - random-poisson
    - Definitions of tests & helper procs
 * possion_test.tcl
    - The code that runs the tests, including a visual comparison of a histogram produced by random-poisson and the pdf
 * simpletest.tcl
    - The code to run the non-visual tests
 * incgamma.tcl
    - Implementation of the incomplete gamma function

This code is more explicit about the tests.

Also, it includes one bug fix: the cdf should have been evaluated at floor(x), rather than at x. This is now corrected.

Eric

File Added: poisson.zip

File Added - 259984: random-poisson.tcl