MooseRandom Class Reference

This class encapsulates a useful, consistent, cross-platform random number generator with multiple utilities. More...

#include <MooseRandom.h>

Public Member Functions
void	seed (std::size_t i, unsigned int seed)
	The method seeds one of the independent random number generators. More...
Real	rand (std::size_t i)
	This method returns the next random number (Real format) from the specified generator. More...
Real	randNormal (std::size_t i, Real mean, Real sigma)
	This method returns the next random number (Real format) from the specified generator, drawn from a normal distribution centered around mean, with a width of sigma. More...
Real	randNormal (std::size_t i)
	Return next random number drawn from a standard distribution. More...
uint32_t	randl (std::size_t i)
	This method returns the next random number (long format) from the specified generator. More...
uint32_t	randl (std::size_t i, uint32_t lower, uint32_t upper)
	This method returns the next random number (long format) from the specified generator within the supplied range. More...
void	saveState ()
	This method saves the current state of all generators which can be restored at a later time (i.e. More...
void	restoreState ()
	This method restores the last saved generator state. More...
std::size_t	size ()
	Return the number of states. More...

Static Public Member Functions
static void	seed (unsigned int seed)
	The method seeds the random number generator. More...
static Real	rand ()
	This method returns the next random number (Real format) from the generator. More...
static Real	randNormal (Real mean, Real sigma)
	This method returns the next random number (Real format) from the generator, drawn from a normal distribution centered around mean, with a width of sigma. More...
static Real	randNormal ()
	Return next random number drawn from a standard distribution. More...
static uint32_t	randl ()
	This method returns the next random number (long format) from the generator. More...

Private Attributes
std::unordered_map< std::size_t, std::pair< mt_state, mt_state > >	_states
	We store a pair of states in this map. More...
bool	_saved = false
	Flag to make certain that saveState is called prior to restoreState. More...

Friends
void	dataStore (std::ostream &stream, MooseRandom &v, void *context)
void	dataLoad (std::istream &stream, MooseRandom &v, void *context)

Detailed Description

This class encapsulates a useful, consistent, cross-platform random number generator with multiple utilities.

1. SIMPLE INTERFACE: There are three static functions that are suitable as a drop in replacement for the random number capabilities available in the standard C++ library. This interface DOES NOT SUPPORT restart/recover
2. ADVANCED INTERFACE: When creating an instance of this class, one can maintain an arbitrary number of multiple independent streams of random numbers. Furthermore, the state of these generators can be saved and restored for all streams by using the "saveState" and "restoreState" methods. Finally, this class uses a fast hash map so that indexes for the generators are not required to be contiguous.

Definition at line 37 of file MooseRandom.h.

Member Function Documentation

rand() [1/2]

static Real MooseRandom::rand ( )

inlinestatic

This method returns the next random number (Real format) from the generator.

Returns

the next random number in the range [0,1) with 64-bit precision

Definition at line 50 of file MooseRandom.h.

Referenced by RandomPartitioner::_do_partition(), RandomICBase::generateRandom(), RankTwoTensorTempl< Real >::genRandomSymmTensor(), SymmetricRankTwoTensorTempl< Real >::genRandomSymmTensor(), RankTwoTensorTempl< Real >::genRandomTensor(), Sampler::getRand(), RandomInterface::getRandomReal(), and PropertyReadFile::initVoronoiCenterPoints().

{ return mt_ldrand(); }

rand() [2/2]

Real MooseRandom::rand ( std::size_t  i )

inline

This method returns the next random number (Real format) from the specified generator.

Parameters

i	the index of the generator

Returns

the next random number in the range [0,1) with 64-bit precision

Definition at line 85 of file MooseRandom.h.

  {
    // mooseAssert(_states.find(i) != _states.end(), "No random state initialized for id: " << i);
    return mts_ldrand(&(_states[i].first));
  }

randl() [1/3]

static uint32_t MooseRandom::randl ( )

inlinestatic

This method returns the next random number (long format) from the generator.

Returns

the next random number in the range [0,max(uinit32_t)) with 32-bit number

Definition at line 71 of file MooseRandom.h.

Referenced by Sampler::getRandl(), RandomInterface::getRandomLong(), RandomICBase::RandomICBase(), MooseUtils::resample(), MooseUtils::resampleWithFunctor(), MooseUtils::shuffle(), and RandomData::updateGenerators().

{ return mt_lrand(); }

randl() [2/3]

uint32_t MooseRandom::randl ( std::size_t  i )

inline

This method returns the next random number (long format) from the specified generator.

Parameters

i	the index of the generator

Returns

the next random number in the range [0,max(uinit32_t)) with 32-bit number

Definition at line 116 of file MooseRandom.h.

  {
    mooseAssert(_states.find(i) != _states.end(), "No random state initialized for id: " << i);
    return mts_lrand(&(_states[i].first));
  }

randl() [3/3]

uint32_t MooseRandom::randl ( std::size_t  i, uint32_t  lower, uint32_t  upper  )

inline

This method returns the next random number (long format) from the specified generator within the supplied range.

Parameters

lower	lower bounds of value
upper	upper bounds of value
i	the index of the generator

Returns

the next random number in the range [0,max(uinit32_t)) with 32-bit number

Definition at line 131 of file MooseRandom.h.

  {
    mooseAssert(_states.find(i) != _states.end(), "No random state initialized for id: " << i);
    return rds_iuniform(&(_states[i].first), lower, upper);
  }

randNormal() [1/4]

static Real MooseRandom::randNormal ( Real  mean, Real  sigma  )

inlinestatic

This method returns the next random number (Real format) from the generator, drawn from a normal distribution centered around mean, with a width of sigma.

Parameters

mean	center of the random number distribution
sigma	width of the random number distribution

Returns

the next random number following a normal distribution of width sigma around mean with 64-bit precision

Definition at line 60 of file MooseRandom.h.

{ return rd_normal(mean, sigma); }

randNormal() [2/4]

static Real MooseRandom::randNormal ( )

inlinestatic

Return next random number drawn from a standard distribution.

Definition at line 65 of file MooseRandom.h.

Referenced by randNormal().

{ return randNormal(0.0, 1.0); }

randNormal() [3/4]

Real MooseRandom::randNormal ( std::size_t  i, Real  mean, Real  sigma  )

inline

This method returns the next random number (Real format) from the specified generator, drawn from a normal distribution centered around mean, with a width of sigma.

Parameters

i	the index of the generator
mean	center of the random number distribution
sigma	width of the random number distribution

Returns

the next random number following a normal distribution of width sigma around mean with 64-bit precision

Definition at line 100 of file MooseRandom.h.

  {
    mooseAssert(_states.find(i) != _states.end(), "No random state initialized for id: " << i);
    return rds_normal(&(_states[i].first), mean, sigma);
  }

randNormal() [4/4]

Real MooseRandom::randNormal ( std::size_t  i )

inline

Return next random number drawn from a standard distribution.

Definition at line 109 of file MooseRandom.h.

Referenced by randNormal().

{ return randNormal(i, 0.0, 1.0); }

restoreState()

void MooseRandom::restoreState ( )

inline

This method restores the last saved generator state.

Definition at line 153 of file MooseRandom.h.

Referenced by Sampler::restoreGeneratorState(), and RandomData::updateSeeds().

  {
    mooseAssert(_saved, "saveState() must be called prior to restoreState().");
    std::for_each(_states.begin(),
                  _states.end(),
                  [](std::pair<const std::size_t, std::pair<mt_state, mt_state>> & pair)
                  { pair.second.first = pair.second.second; });
  }

saveState()

void MooseRandom::saveState ( )

inline

This method saves the current state of all generators which can be restored at a later time (i.e.

re-generate the same sequence of random numbers of this generator

Definition at line 141 of file MooseRandom.h.

Referenced by Sampler::saveGeneratorState(), and RandomData::updateSeeds().

  {
    _saved = true;
    std::for_each(_states.begin(),
                  _states.end(),
                  [](std::pair<const std::size_t, std::pair<mt_state, mt_state>> & pair)
                  { pair.second.second = pair.second.first; });
  }

seed() [1/2]

static void MooseRandom::seed ( unsigned int  seed )

inlinestatic

The method seeds the random number generator.

Parameters

seed	the seed number

Definition at line 44 of file MooseRandom.h.

Referenced by Sampler::init(), RankTwoTensorTempl< Real >::initRandom(), SymmetricRankTwoTensorTempl< Real >::initRandom(), PropertyReadFile::initVoronoiCenterPoints(), MooseInit::MooseInit(), RandomICBase::RandomICBase(), RandomPartitioner::RandomPartitioner(), seed(), and RandomData::updateGenerators().

{ mt_seed32new(seed); }

seed() [2/2]

void MooseRandom::seed ( std::size_t  i, unsigned int  seed  )

inline

The method seeds one of the independent random number generators.

Parameters

i	the index of the generator
seed	the seed number

Definition at line 78 of file MooseRandom.h.

{ mts_seed32new(&(_states[i].first), seed); }

size()

std::size_t MooseRandom::size ( )

inline

Return the number of states.

Definition at line 165 of file MooseRandom.h.

Referenced by Sampler::advanceGenerators(), Sampler::getRand(), and Sampler::getRandl().

{ return _states.size(); }

Friends And Related Function Documentation

dataLoad

void dataLoad ( std::istream &  stream, MooseRandom &  v, void *  context  )

friend

Definition at line 191 of file MooseRandom.h.

{
  loadHelper(stream, v._states, context);
}

dataStore

void dataStore ( std::ostream &  stream, MooseRandom &  v, void *  context  )

friend

Definition at line 185 of file MooseRandom.h.

{
  storeHelper(stream, v._states, context);
}

Member Data Documentation

_saved

bool MooseRandom::_saved = false

private

Flag to make certain that saveState is called prior to restoreState.

Definition at line 180 of file MooseRandom.h.

Referenced by restoreState(), and saveState().

_states

std::unordered_map<std::size_t, std::pair<mt_state, mt_state> > MooseRandom::_states

private

We store a pair of states in this map.

The first one is the active state, the second is the backup state. It is used to restore state at a later time to the active state.

Definition at line 173 of file MooseRandom.h.

Referenced by dataLoad(), dataStore(), rand(), randl(), randNormal(), restoreState(), saveState(), seed(), and size().

---

The documentation for this class was generated from the following file:

• include/utils/MooseRandom.h