User-supplied Random Number Generation

Description

Function RNGkind allows user-coded uniform and normal random number generators to be supplied. The details are given here.

Details

A user-specified uniform RNG is called from entry points in dynamically-loaded compiled code. The user must supply the entry point user_unif_rand, which takes no arguments and returns a pointer to a double. The example below will show the general pattern. The generator should have at least 25 bits of precision.

Optionally, the user can supply the entry point user_unif_init, which is called with an unsigned int argument when RNGkind (or set.seed) is called, and is intended to be used to initialize the user's RNG code. The argument is intended to be used to set the ‘seeds’; it is the seed argument to set.seed or an essentially random seed if RNGkind is called.

If only these functions are supplied, no information about the generator's state is recorded in .Random.seed. Optionally, functions user_unif_nseed and user_unif_seedloc can be supplied which are called with no arguments and should return pointers to the number of seeds and to an integer (specifically, Int32) array of seeds. Calls to GetRNGstate and PutRNGstate will then copy this array to and from .Random.seed.

A user-specified normal RNG is specified by a single entry point user_norm_rand, which takes no arguments and returns a pointer to a double.

Warning

As with all compiled code, mis-specifying these functions can crash R. Do include the ‘R_ext/Random.h’ header file for type checking.

Examples

## Not run: 
##  Marsaglia's congruential PRNG
#include <R_ext/Random.h>

static Int32 seed;
static double res;
static int nseed = 1;

double * user_unif_rand()
{
    seed = 69069 * seed + 1;
    res = seed * 2.32830643653869e-10;
    return &res;
}

void  user_unif_init(Int32 seed_in) { seed = seed_in; }
int * user_unif_nseed() { return &nseed; }
int * user_unif_seedloc() { return (int *) &seed; }

/*  ratio-of-uniforms for normal  */
#include <math.h>
static double x;

double * user_norm_rand()
{
    double u, v, z;
    do {
        u = unif_rand();
        v = 0.857764 * (2. * unif_rand() - 1);
        x = v/u; z = 0.25 * x * x;
        if (z < 1. - u) break;
        if (z > 0.259/u + 0.35) continue;
    } while (z > -log(u));
    return &x;
}

## Use under Unix:
R CMD SHLIB urand.c
R
> dyn.load("urand.so")
> RNGkind("user")
> runif(10)
> .Random.seed
> RNGkind(, "user")
> rnorm(10)
> RNGkind()
[1] "user-supplied" "user-supplied"

## End(Not run)