[Rd] Writing new probability functions

[Vincent Goulet]

Dear developeRs,

I am maintaining a package of Actuarial Science functions (soon to be uploaded 
to CRAN) in which we would like to include the usual d*, p*, q* and r* 
functions for some probability laws not currently found in base R.

Taking into account that most of the distributions are functions or 
transformations of the distributions found in base R, should we look into 
writing our functions in C or in R, both in terms of: (1) speed, and (2) 
accuracy?

If we opt for C, is there a canonical reference detailing the precautions one 
has to take to ensure accuracy of the density/distribution/quantile functions 
over the complete domain of the probability laws?

Comments appreciated.

-- 
  Vincent Goulet, Associate Professor
  École d'actuariat
  Université Laval, Québec 
  Vincent.Goulet at act.ulaval.ca   http://vgoulet.act.ulaval.ca