Random Deviate Generators.

Random deviate generators (RDGs) produce random numbers with various distributions on the basis of [0,1) uniformly distributed numbers. More...

Classes
class	librandom::BinomialRandomDev
	Class BinomialRNG. More...
class	librandom::ClippedRedrawContinuousRandomDev< BaseRDV >
	Wrapper template turning any continuous RDV into a clipped RDV. More...
class	librandom::ClippedRedrawDiscreteRandomDev< BaseRDV >
	Wrapper template turning any discrete RDV into a clipped RDV. More...
class	librandom::ClippedToBoundaryContinuousRandomDev< BaseRDV >
	Wrapper template turning any continuous RDV into a clipped-to-boundary RDV. More...
class	librandom::ClippedToBoundaryDiscreteRandomDev< BaseRDV >
	Wrapper template turning any discrete RDV into a clipped-to-boundary RDV. More...
class	librandom::ExpRandomDev
	Class ExpRandomDev Create exponential random numbers. More...
class	librandom::GammaRandomDev
	Class GammaRandomDev Create gamma distributed random numbers. More...
class	librandom::GSL_BinomialRandomDev
	Class GSL_BinomialRandomDev. More...
class	librandom::LognormalRandomDev
	Create lognormal random numbers with uniform variance. More...
class	librandom::NormalRandomDev
	Create normal (Gaussian) random numbers with uniform variance. More...
class	librandom::PoissonRandomDev
	Class PoissonRandomDev Create Poisson distributed random numbers. More...
class	librandom::UniformRandomDev
	Class UniformRandomDev Create uniformly distributed random numbers in [low, high). More...
class	librandom::UniformIntRandomDev
	Class UniformIntRandomDev Create uniformly distributed random integers from a given range. More...

Detailed Description

Random deviate generators (RDGs) produce random numbers with various distributions on the basis of [0,1) uniformly distributed numbers.

Discrete and continuous distributions are available. For a complete list of available deviates, please use the following sli command

* SLI ] rdevdict info
* --------------------------------------------------
* Name                     Type                Value
* --------------------------------------------------
* binomial                 rdvfacttype         <rdvfacttype>
* poisson                  rdvfacttype         <rdvfacttype>
* normal                   rdvfacttype         <rdvfacttype>
* exponential              rdvfacttype         <rdvfacttype>
* gamma                    rdvfacttype         <rdvfacttype>
* uniformint               rdvfacttype         <rdvfacttype>
* --------------------------------------------------
* 

Note

RDGs that are to be used in multithreaded code, must be called with the thread-RNG as argument whenever a number is drawn, to ensure thread-consistent number generation.

All RDGs provide double numbers, generators for discrete distributions may provide unsigned longs as well (eg, Poisson). This can be checked with the has_ldev().

Here is a code example for the use of the Poisson generator:

#include "poisson_randomdev.h"
class poisson_generator : 
  public Node, 
  protected Device
{
...
private: 
  librandom::PoissonRandomDev poisson_dev_;  
... 
 };

nest::poisson_generator::poisson_generator()
:...
  poisson_dev_(0.0),
 ...
{
  calibrate(network()->get_resolution());
}
void nest::poisson_generator::calibrate(Time const & dt)
{
 poisson_dev_.set_lambda(dt.get_ms() * rate_*1e-3);
}

void nest::poisson_generator::update(thread thrd, Time const & T)
{
  ...
  librandom::RngPtr rng=Node::network()->get_rng(thrd);
  ...
    long_t n_spikes = poisson_dev_.ldev(rng);
  ...
}

Note that the RNG used by the deviate generator must be obtained via get_rng(thrd) on each call to update, to get the proper generator for the present thread.