Macros
#define	init_randmt(seed) (init_randmt_r(&__randmt_global_generator, seed))
	Initialize the global generator with a seed. More...

#define	init_randmt_auto() (init_randmt_auto_r(&__randmt_global_generator))
	Initialize the global generator with the current time. More...

#define	rand_uint32() (rand_uint32_r(&__randmt_global_generator))
	Generate a random 32-bit unsigned integer (nonreentrant) More...

#define	rand_unif() (rand_unif_r(&__randmt_global_generator))
	Generate a uniform random number on (0,1) (nonreentrant) More...

#define	rand_normal() (rand_normal_r(&__randmt_global_generator))
	Generate a standard normal distributed random number (nonreentrant) More...

#define	rand_exp(mu) (rand_exp_r(&__randmt_global_generator, mu))
	Generate an exponentially-distributed number (nonreentrant) More...

#define	rand_gamma(a, b) (rand_gamma_r(&__randmt_global_generator, a, b))
	Generate a Gamma-distributed number (nonreentrant) More...

#define	rand_poisson(mu) (rand_poisson_r(&__randmt_global_generator, mu))
	Generate a Poisson-distributed number (nonreentrant) More...

#define	rand_geometric(p) (rand_geometric_r(&__randmt_global_generator, p))
	Generate a geometrically-distributed number (nonreentrant) More...

Detailed Description

These functions use the global pseudorandom number generator.

Macro Definition Documentation

#define init_randmt ( seed ) (init_randmt_r(&__randmt_global_generator, seed))

Initialize the global generator with a seed.

Parameters

seed	the seed value

See Also: init_randmt_auto(); init_randmt_r()

This routine seeds the global random number generator with an unsigned 32-bit integer value.

A constant seed can be used to reproduce the same pseudorandom numbers.

int i;
init_randmt(42);
printf("Ten numbers:\n");
for(i = 0; i < 10; i++)
    printf("%f\n", rand_unif());
init_randmt(42);
printf("The same ten numbers:\n");
for(i = 0; i < 10; i++)
    printf("%f\n", rand_unif());

Definition at line 314 of file randmt.h.

#define init_randmt_auto ( ) (init_randmt_auto_r(&__randmt_global_generator))

Initialize the global generator with the current time.

See Also: init_randmt(); init_randmt_auto_r()

This function seeds the global generator with the current time. It should be called once at the beginning of the program so that different pseudorandom values are produced on different runs.

This function only needs to be called once. Seeding multiple times does not improve the statistical quality of the generator.

Definition at line 328 of file randmt.h.

#define rand_exp ( mu ) (rand_exp_r(&__randmt_global_generator, mu))

Generate an exponentially-distributed number (nonreentrant)

Parameters

mu	mean parameter of the distribution (positive value)

Returns: Pseudo-random exponentially-distibuted value with mean mu

See Also: rand_exp_r()

Generates a pseudo-random value distributed with probability density

$f(x;\mu) = \begin{cases} \frac{1}{\mu} \mathrm{e}^{-x/\mu}, & x \ge 0, \\ 0, & x < 0, \end{cases}$

generated by inversion. The global generator is used to generate the value.

Distribution properties:

CDF: $1 - \exp(-x/\mu)$
Mean: $\mu$
Variance: $\mu^2$

Definition at line 395 of file randmt.h.

#define rand_gamma	(	a,
		b
	)	(rand_gamma_r(&__randmt_global_generator, a, b))

Generate a Gamma-distributed number (nonreentrant)

Parameters

a	shape parameter (positive value)
b	scale parameter (positive value)

Returns: Pseudo-random Gamma-distibuted value

See Also: rand_gamma_r()

Generates a Gamma-distributed random value with density

$f(x;a,b) = x^{a-1} \frac{\exp(-x/b)}{\Gamma(a)b^k}, \quad x \ge 0,$

where $\Gamma(a)$ is the Gamma function, using the method of Marsaglia and Tsang, 2000. The global generator is used to generate the value.

Distribution properties:

CDF: $\gamma(a,x/b)/\Gamma(a)$ , where $\gamma$ is the lower incomplete gamma function
Mean:
Variance:

Definition at line 416 of file randmt.h.

#define rand_geometric ( p ) (rand_geometric_r(&__randmt_global_generator, p))

Generate a geometrically-distributed number (nonreentrant)

Parameters

p	probability of success

Returns: Pseudo-random geometrically-distibuted value

See Also: rand_geometric_r()

Generates a pseudo-random value distributed with probability mass fuction

$\mathrm{P}(X = k) = (1 - p)^{k-1} p, \quad k = 1, 2, \ldots,$

where $0 < p \le 1$ , generated by inversion. The global generator is used to generate the value.

Distribution properties:

CDF: $1 - (1 - p)^{\lfloor k \rfloor}$
Mean:
Variance:

Definition at line 460 of file randmt.h.

#define rand_normal ( ) (rand_normal_r(&__randmt_global_generator))

Generate a standard normal distributed random number (nonreentrant)

Returns: Pseudo-random double value with standard normal distribution

See Also: rand_normal_r()

A pseudo-random normally distributed number with density

$f(x) = \frac{1}{\sqrt{2\pi}} \mathrm{e}^{-x^2/2},$

generated by the Box-Muller transform. The global generator is used to generate the value.

Distribution properties:

CDF: $\frac{1}{2}\bigl( 1 + \mathrm{erf}(x/\sqrt{2})\bigr)$
Mean: 0
Variance: 1

Definition at line 374 of file randmt.h.

#define rand_poisson ( mu ) (rand_poisson_r(&__randmt_global_generator, mu))

Generate a Poisson-distributed number (nonreentrant)

Parameters

mu	the mean parameter of the distribution (positive value)

Returns: Pseudo-random Poisson-distibuted value

See Also: rand_poisson_r()

A pseudo-random normally distributed number with probability mass function

$\mathrm{P}(X = k) = \frac{\mu^k \mathrm{e}^{-\mu}}{k!}, \quad k = 0, 1,\ldots,$

using a simple direct algorthm for mu < 10 and the "PTRS" algorthm of Hormann for larger mu. The global generator is used to generate the value.

Reference: Wolfgang Hormann, "The transformed rejection method for generating Poisson random variables," Insurance: Mathematics and Economics 12, pp. 39-45, 1993.

Distribution properties: