The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
 / 
PDL-2.013
River stage three • 93 direct dependents • 101 total dependents
52 ++
/ PDL::GSL::RNG

NAME

PDL::GSL::RNG - PDL interface to RNG and randist routines in GSL

DESCRIPTION

This is an interface to the rng and randist packages present in the GNU Scientific Library.

SYNOPSIS

   use PDL;
   use PDL::GSL::RNG;

   $rng = PDL::GSL::RNG->new('taus');

   $rng->set_seed(time());

   $a=zeroes(5,5,5)

   $rng->get_uniform($a); # inplace

   $b=$rng->get_uniform(3,4,5); # creates new pdl

FUNCTIONS

new()

The new method initializes a new instance of the RNG.

The avaible RNGs are: coveyou cmrg fishman18 fishman20 fishman2x gfsr4 knuthran knuthran2 knuthran2002 lecuyer21 minstd mrg mt19937 mt19937_1999 mt19937_1998 r250 ran0 ran1 ran2 ran3 rand rand48 random128_bsd random128_glibc2 random128_libc5 random256_bsd random256_glibc2 random256_libc5 random32_bsd random32_glibc2 random32_libc5 random64_bsd random64_glibc2 random64_libc5 random8_bsd random8_glibc2 random8_libc5 random_bsd random_glibc2 random_libc5 randu ranf ranlux ranlux389 ranlxd1 ranlxd2 ranlxs0 ranlxs1 ranlxs2 ranmar slatec taus taus2 taus113 transputer tt800 uni uni32 vax waterman14 zuf default The last one (default) uses the enviroment variable GSL_RNG_TYPE. Note that only a few of these rngs are recommended for general use. Please check the GSL documentation for more information.

Usage:

   $blessed_ref = PDL::GSL::RNG->new($RNG_name);

Example:

   $rng = PDL::GSL::RNG->new('taus');

set_seed();

Sets the RNG seed.

Usage:

   $rng->set_seed($integer);

Example:

   $rng->set_seed(666);

min()

Return the minimum value generable by this RNG.

Usage:

   $integer = $rng->min();

Example:

   $min = $rng->min(); $max = $rng->max();

max()

Return the maximum value generable by the RNG.

Usage:

   $integer = $rng->max();

Example:

   $min = $rng->min(); $max = $rng->max();

name()

Returns the name of the RNG.

Usage:

   $string = $rng->name();

Example:

   $name = $rng->name();

get_uniform()

This function creates a piddle with given dimensions or accept an existing piddle and fills it. get_uniform() returns values 0<=x<1,

Usage:

   $piddle = $rng->get_uniform($list_of_integers)
   $rng->get_uniform($piddle);

Example:

   $a = zeroes 5,6; $max=100;
   $o = $rng->get_uniform(10,10); $rng->get_uniform($a);

get_uniform_pos()

This function creates a piddle with given dimensions or accept an existing piddle and fills it. get_uniform_pos() returns values 0<x<1,

Usage:

   $piddle = $rng->get_uniform_pos($list_of_integers)
   $rng->get_uniform_pos($piddle);

Example:

   $a = zeroes 5,6;
   $o = $rng->get_uniform_pos(10,10); $rng->get_uniform_pos($a);

get()

This function creates a piddle with given dimensions or accept an existing piddle and fills it. get() returns integer values beetween a minimum and a maximum specific to evry RNG.

Usage:

   $piddle = $rng->get($list_of_integers)
   $rng->get($piddle);

Example:

   $a = zeroes 5,6;
   $o = $rng->get(10,10); $rng->get($a);

get_int()

This function creates a piddle with given dimensions or accept an existing piddle and fills it. get_int() returns integer values beetween 0 and $max.

Usage:

   $piddle = $rng->get($max, $list_of_integers)
   $rng->get($max, $piddle);

Example:

   $a = zeroes 5,6; $max=100;
   $o = $rng->get(10,10); $rng->get($a);

ran_gaussian()

These functions return random deviates from given distribution.

The general form is

  ran_[distrib](args)

where distrib can be any of the ones shown below.

They accept the parameters of the distribution and a specification of where to put output. This spec can be in form of list of integers that specify the dimensions of the ouput piddle or an existing piddle that will be filled with values inplace.

Usage:

   # gaussian dist
   $piddle = $rng->ran_gaussian($sigma,[list of integers]);
   $rng->ran_gaussian($sigma,$piddle);

   # gaussian tail
   $piddle = $rng->ran_ugaussian_tail($tail,[list of integers]);
   $rng->ran_ugaussian_tail($tail,$piddle);

   # exponential dist
   $piddle = $rng->ran_exponential($mu,[list of integers]);
   $rng->ran_exponential($mu,$piddle);

   # laplacian dist
   $piddle = $rng->ran_laplace($mu,[list of integers]);
   $rng->ran_laplace($mu,$piddle);

   $piddle = $rng->ran_exppow($mu,$a,[list of integers]);
   $rng->ran_exppow($mu,$a,$piddle);

   $piddle = $rng->ran_cauchy($mu,[list of integers]);
   $rng->ran_cauchy($mu,$piddle);

   $piddle = $rng->ran_rayleigh($sigma,[list of integers]);
   $rng->ran_rayleigh($sigma,$piddle);

   $piddle = $rng->ran_rayleigh_tail($a,$sigma,[list of integers]);
   $rng->ran_rayleigh_tail($a,$sigma,$piddle);

   $piddle = $rng->ran_levy($mu,$a,[list of integers]);
   $rng->ran_levy($mu,$a,$piddle);

   $piddle = $rng->ran_gamma($a,$b,[list of integers]);
   $rng->ran_gamma($a,$b,$piddle);

   $piddle = $rng->ran_flat($a,$b,[list of integers]);
   $rng->ran_flat($a,$b,$piddle);

   $piddle = $rng->ran_lognormal($zeta, $sigma,[list of integers]);
   $rng->ran_lognormal($zeta, $sigma,$piddle);

   $piddle = $rng->ran_chisq($nu,[list of integers]);
   $rng->ran_chisq($nu,$piddle);

   $piddle = $rng->ran_fdist($nu1, $nu2,[list of integers]);
   $rng->ran_fdist($nu1, $nu2,$piddle);

   $piddle = $rng->ran_tdist($nu,[list of integers]);
   $rng->ran_tdist($nu,$piddle);

   $piddle = $rng->ran_beta($a,$b,[list of integers]);
   $rng->ran_beta($a,$b,$piddle);

   $piddle = $rng->ran_logistic($m,[list of integers]u)
   $rng->ran_logistic($m,$piddleu)

   $piddle = $rng->ran_pareto($a,$b,[list of integers]);
   $rng->ran_pareto($a,$b,$piddle);

   $piddle = $rng->ran_weibull($mu,$a,[list of integers]);
   $rng->ran_weibull($mu,$a,$piddle);

   $piddle = $rng->ran_gumbel1($a,$b,[list of integers]);
   $rng->ran_gumbel1($a,$b,$piddle);

   $piddle = $rng->ran_gumbel2($a,$b,[list of integers]);
   $rng->ran_gumbel2($a,$b,$piddle);

   $piddle = $rng->ran_poisson($mu,[list of integers]);
   $rng->ran_poisson($mu,$piddle);

   $piddle = $rng->ran_bernoulli($p,[list of integers]);
   $rng->ran_bernoulli($p,$piddle);

   $piddle = $rng->ran_binomial($p,$n,[list of integers]);
   $rng->ran_binomial($p,$n,$piddle);

   $piddle = $rng->ran_negative_binomial($p,$n,[list of integers]);
   $rng->ran_negative_binomial($p,$n,$piddle);

   $piddle = $rng->ran_pascal($p,$n,[list of integers]);
   $rng->ran_pascal($p,$n,$piddle);

   $piddle = $rng->ran_geometric($p,[list of integers]);
   $rng->ran_geometric($p,$piddle);

   $piddle = $rng->ran_hypergeometric($n1, $n2, $t,[list of integers]);
   $rng->ran_hypergeometric($n1, $n2, $t,$piddle);

   $piddle = $rng->ran_logarithmic($p,[list of integers]);
   $rng->ran_logarithmic($p,$piddle);

Example:

  $o = $rng->ran_gaussian($sigma,10,10);
  $rng->ran_gaussian($sigma,$a);

ran_gaussian_var()

This method is similar to ran_[distrib]() except that it takes the parameters of the distribution as a piddle and returns a piddle of equal dimensions. Of course, you can use the same set of distributions as in the previous method (see also the ran_gaussian entry above).

Usage:

   $piddle = $rng->ran_[distribution]_var($distr_parameters_list,$piddle_dim_list);
   $rng->ran_[distribution]_var($distr_parameters_list,$piddle);

Example:

   $sigma_pdl = rvals zeroes 11,11;
   $o = $rng->ran_gaussian_var($sigma_pdl);

ran_additive_gaussian()

Add Gaussian noise of given sigma to a piddle.

Usage:

   $rng->ran_additive_gaussian($sigma,$piddle);

Example:

   $rng->ran_additive_gaussian(1,$image);

ran_additive_poisson()

Add Poisson noise of given sigma to a piddle.

Usage:

   $rng->ran_additive_poisson($mu,$piddle);

Example:

   $rng->ran_additive_poisson(1,$image);

ran_feed_poisson()

This method simulates shot noise, taking the values of piddle as values for mu to be fed in the poissonian RNG.

Usage:

   $rng->ran_feed_poisson($piddle);

Example:

   $rng->ran_feed_poisson($image);

ran_bivariate_gaussian()

Generates $n bivariate gaussian random deviates.

Usage:

   $piddle = $rng->ran_bivariate_gaussian($sigma_x,$sigma_y,$rho,$n);

Example:

   $o = $rng->ran_bivariate_gaussian(1,2,0.5,1000);

ran_dir()

Returns $n random vectors in $ndim dimensions.

Usage:

   $piddle = $rng->ran_dir($ndim,$n);

Example:

   $o = $rng->ran_dir($ndim,$n);

ran_discrete_preproc()

This method returns a handle that must be used when calling ran_discrete(). You specify the probability of the integer number that are returned by ran_discrete().

Usage:

   $discrete_dist_handle = $rng->ran_discrete_preproc($double_piddle_prob);

Example:

   $prob = pdl [0.1,0.3,0.6];
   $ddh = $rng->ran_discrete_preproc($prob);
   $o = $rng->ran_discrete($discrete_dist_handle,100);

ran_discrete()

Is used to get the desired samples once a proper handle has been enstablished (see ran_discrete_preproc()).

Usage:

   $piddle = $rng->ran_discrete($discrete_dist_handle,$num);

Example:

   $prob = pdl [0.1,0.3,0.6];
   $ddh = $rng->ran_discrete_preproc($prob);
   $o = $rng->ran_discrete($discrete_dist_handle,100);

ran_shuffle()

Shuffles values in piddle

Usage:

   $rng->ran_shuffle($piddle);

ran_shuffle_vec()

Shuffles values in piddle

Usage:

   $rng->ran_shuffle_vec(@vec);

ran_choose()

Chooses values from $inpiddle to $outpiddle.

Usage:

   $rng->ran_choose($inpiddle,$outpiddle);

ran_choose_vec()

Chooses $n values from @vec.

Usage:

   @choosen = $rng->ran_choose_vec($n,@vec);

ran_ver()

Returns a piddle with $n values generated by the Verhulst map from $x0 and paramater $r.

Usage:

   $rng->ran_ver($x0, $r, $n);

ran_caos()

Returns values from Verhuls map with $r=4.0 and randomly choosen $x0. The values are scaled by $m.

Usage:

   $rng->ran_caos($m,$n);

BUGS

Feedback is welcome. Log bugs in the PDL bug database (the database is always linked from http://pdl.perl.org/).

SEE ALSO

PDL

The GSL documentation is online at http://www.gnu.org/software/gsl/manual/html_node/

AUTHOR

This file copyright (C) 1999 Christian Pellegrin <chri@infis.univ.trieste.it> Docs mangled by C. Soeller. All rights reserved. There is no warranty. You are allowed to redistribute this software / documentation under certain conditions. For details, see the file COPYING in the PDL distribution. If this file is separated from the PDL distribution, the copyright notice should be included in the file.

The GSL RNG and randist modules were written by James Theiler.