[Rd] norm_rand() in R-extension vs rnorm in R ---which is better?

[Kyeongmi Cheon]

I need to generate good quality of random numbers from univariate
normal distribution for further transformation. I tried rnorm in R but
it was not good sometimes. Someone said C++ standard library or
Fortran's built-in functions for that are good. I found that there is
double norm_rand() in R-extension. Does this function, norm_rand()
use the same algorithm as rnorm in R or something similar in C/C++? Or
does anyone know which one between norm_rand() in R-extension and
rnorm in R is better? If anyone could help, I would appreciate it.
Kyeongmi

University of Memphis

[Daniel Nordlund]

-----Original Message-----
From: r-devel-bounces at r-project.org [mailto:r-devel-bounces at r-project.org] On Behalf
Of Kyeongmi Cheon
Sent: Saturday, March 29, 2008 9:09 PM
To: r-devel at r-project.org
Subject: [Rd] norm_rand() in R-extension vs rnorm in R ---which is better?
I need to generate good quality of random numbers from univariate
normal distribution for further transformation. I tried rnorm in R but
it was not good sometimes. Someone said C++ standard library or
Fortran's built-in functions for that are good. I found that there is
double norm_rand() in R-extension. Does this function, norm_rand()
use the same algorithm as rnorm in R or something similar in C/C++? Or
does anyone know which one between norm_rand() in R-extension and
rnorm in R is better? If anyone could help, I would appreciate it.
Kyeongmi

What do you mean when you write that rnorm "was not good sometimes"? Can you provide an example which demonstrates that rnorm performs poorly? Enquiring minds would like to know.

Dan

Daniel Nordlund
Bothell, WA USA