A Discrete-Event Network Simulator
API
ns3::EmpiricalRandomVariable Class Reference

The Random Number Generator (RNG) that has a specified empirical distribution. More...

#include "random-variable-stream.h"

+ Inheritance diagram for ns3::EmpiricalRandomVariable:
+ Collaboration diagram for ns3::EmpiricalRandomVariable:

Public Member Functions

 EmpiricalRandomVariable ()
 Creates an empirical RNG that has a specified, empirical distribution, and configured for interpolating mode. More...
 
void CDF (double v, double c)
 Specifies a point in the empirical distribution. More...
 
virtual uint32_t GetInteger ()
 Get the next random value drawn from the distribution. More...
 
double GetValue () override
 Get the next random value drawn from the distribution. More...
 
virtual double Interpolate ()
 Returns the next value in the empirical distribution using linear interpolation. More...
 
bool SetInterpolate (bool interpolate)
 Switch the mode between sampling the CDF and interpolating. More...
 
- Public Member Functions inherited from ns3::RandomVariableStream
 RandomVariableStream ()
 Default constructor. More...
 
 RandomVariableStream (const RandomVariableStream &)=delete
 
 ~RandomVariableStream () override
 Destructor. More...
 
int64_t GetStream () const
 Returns the stream number for the RngStream. More...
 
bool IsAntithetic () const
 Check if antithetic values will be generated. More...
 
RandomVariableStreamoperator= (const RandomVariableStream &)=delete
 
void SetAntithetic (bool isAntithetic)
 Specify whether antithetic values should be generated. More...
 
void SetStream (int64_t stream)
 Specifies the stream number for the RngStream. More...
 
- Public Member Functions inherited from ns3::Object
 Object ()
 Constructor. More...
 
 ~Object () override
 Destructor. More...
 
void AggregateObject (Ptr< Object > other)
 Aggregate two Objects together. More...
 
void Dispose ()
 Dispose of this Object. More...
 
AggregateIterator GetAggregateIterator () const
 Get an iterator to the Objects aggregated to this one. More...
 
TypeId GetInstanceTypeId () const override
 Get the most derived TypeId for this Object. More...
 
template<typename T >
Ptr< T > GetObject () const
 Get a pointer to the requested aggregated Object. More...
 
template<>
Ptr< ObjectGetObject () const
 Specialization of () for objects of type ns3::Object. More...
 
template<typename T >
Ptr< T > GetObject (TypeId tid) const
 Get a pointer to the requested aggregated Object by TypeId. More...
 
template<>
Ptr< ObjectGetObject (TypeId tid) const
 Specialization of (TypeId tid) for objects of type ns3::Object. More...
 
void Initialize ()
 Invoke DoInitialize on all Objects aggregated to this one. More...
 
bool IsInitialized () const
 Check if the object has been initialized. More...
 
- Public Member Functions inherited from ns3::SimpleRefCount< Object, ObjectBase, ObjectDeleter >
 SimpleRefCount ()
 Default constructor. More...
 
 SimpleRefCount (const SimpleRefCount &o[[maybe_unused]])
 Copy constructor. More...
 
uint32_t GetReferenceCount () const
 Get the reference count of the object. More...
 
SimpleRefCountoperator= (const SimpleRefCount &o[[maybe_unused]])
 Assignment operator. More...
 
void Ref () const
 Increment the reference count. More...
 
void Unref () const
 Decrement the reference count. More...
 
- Public Member Functions inherited from ns3::ObjectBase
virtual ~ObjectBase ()
 Virtual destructor. More...
 
void GetAttribute (std::string name, AttributeValue &value) const
 Get the value of an attribute, raising fatal errors if unsuccessful. More...
 
bool GetAttributeFailSafe (std::string name, AttributeValue &value) const
 Get the value of an attribute without raising errors. More...
 
void SetAttribute (std::string name, const AttributeValue &value)
 Set a single attribute, raising fatal errors if unsuccessful. More...
 
bool SetAttributeFailSafe (std::string name, const AttributeValue &value)
 Set a single attribute without raising errors. More...
 
bool TraceConnect (std::string name, std::string context, const CallbackBase &cb)
 Connect a TraceSource to a Callback with a context. More...
 
bool TraceConnectWithoutContext (std::string name, const CallbackBase &cb)
 Connect a TraceSource to a Callback without a context. More...
 
bool TraceDisconnect (std::string name, std::string context, const CallbackBase &cb)
 Disconnect from a TraceSource a Callback previously connected with a context. More...
 
bool TraceDisconnectWithoutContext (std::string name, const CallbackBase &cb)
 Disconnect from a TraceSource a Callback previously connected without a context. More...
 

Static Public Member Functions

static TypeId GetTypeId ()
 Register this type. More...
 
- Static Public Member Functions inherited from ns3::RandomVariableStream
static TypeId GetTypeId ()
 Register this type. More...
 
- Static Public Member Functions inherited from ns3::Object
static TypeId GetTypeId ()
 Register this type. More...
 
- Static Public Member Functions inherited from ns3::ObjectBase
static TypeId GetTypeId ()
 Get the type ID. More...
 

Private Member Functions

double DoInterpolate (double r)
 Linear interpolation between two points on the CDF to estimate the value at r. More...
 
double DoSampleCDF (double r)
 Sample the CDF as a histogram (without interpolation). More...
 
bool PreSample (double &value)
 Do the initial rng draw and check against the extrema. More...
 
void Validate ()
 Check that the CDF is valid. More...
 

Private Attributes

std::map< double, double > m_empCdf
 The map of CDF points (x, F(x)). More...
 
bool m_interpolate
 If true GetValue will interpolate, otherwise treat CDF as normal histogram. More...
 
bool m_validated
 true once the CDF has been validated. More...
 

Additional Inherited Members

- Protected Member Functions inherited from ns3::RandomVariableStream
RngStreamPeek () const
 Get the pointer to the underlying RngStream. More...
 
- Protected Member Functions inherited from ns3::Object
 Object (const Object &o)
 Copy an Object. More...
 
virtual void DoDispose ()
 Destructor implementation. More...
 
virtual void DoInitialize ()
 Initialize() implementation. More...
 
virtual void NotifyNewAggregate ()
 Notify all Objects aggregated to this one of a new Object being aggregated. More...
 
- Protected Member Functions inherited from ns3::ObjectBase
void ConstructSelf (const AttributeConstructionList &attributes)
 Complete construction of ObjectBase; invoked by derived classes. More...
 
virtual void NotifyConstructionCompleted ()
 Notifier called once the ObjectBase is fully constructed. More...
 

Detailed Description

The Random Number Generator (RNG) that has a specified empirical distribution.

Defines a random variable that has a specified, empirical distribution. The cumulative probability distribution function (CDF) is specified by a series of calls to the CDF() member function, specifying a value $x$ and the probability $P(x)$ that the distribution is less than the specified value. When random values are requested, a uniform random variable $ u \in [0, 1] $ is used to select a probability, and the return value is chosen as the largest input value with CDF less than the random value. This method is known as inverse transform sampling.

This generator has two modes: sampling and interpolating. In sampling mode this random variable generator treats the CDF as an exact histogram and returns one of the histogram inputs exactly. This is appropriate when the configured CDF represents the exact probability distribution, for a categorical variable, for example.

In interpolating mode this random variable generator linearly interpolates between the CDF values defining the histogram bins. This is appropriate when the configured CDF is an approximation to a continuous underlying probability distribution.

For historical reasons the default is sampling. To switch modes use the Interpolate Attribute, or call SetInterpolate(). You can change modes at any time.

If you find yourself switching frequently it could be simpler to set the mode to sampling, then use the GetValue() function for sampled values, and Interpolate() function for interpolated values.

The CDF need not start with a probability of zero, nor end with a probability of 1.0. If the selected uniform random value $ u \in [0,1] $ is less than the probability of the first CDF point, that point is selected. If $u$ is greater than the probability of the last CDF point the last point is selected. In either case the interpolating mode will not interpolate (since there is no value beyond the first/last to work with), but simply return the extremal CDF value, as in sampling.

Example

Here is an example of how to use this class:

// Create the RNG with a non-uniform distribution between 0 and 10.
// in sampling mode.
Ptr<EmpiricalRandomVariable> x = CreateObject<EmpiricalRandomVariable> ();
x->SetInterpolate (false);
x->CDF ( 0.0, 0.0);
x->CDF ( 5.0, 0.25);
x->CDF (10.0, 1.0);
double value = x->GetValue ();
value
Definition: second.py:48

The expected values and probabilities returned by GetValue() are

Value Probability
0.0 0
5.0 25%
10.0 75%

The only two values ever returned are 5 and 10, in the ratio 1:3.

If instead you want linear interpolation between the points of the CDF use the Interpolate() function:

double interp = x->Interpolate ();

This will return continuous values on the range [0,1), 25% of the time less than 5, and 75% of the time between 5 and 10.

See empirical-random-variable-example.cc for an example.

Antithetic Values.

If an instance of this RNG is configured to return antithetic values, the actual value returned, $x'$, is generated by using $ 1 - u $ instead. of $u$ on [0, 1].

Definition at line 1971 of file random-variable-stream.h.

Constructor & Destructor Documentation

◆ EmpiricalRandomVariable()

ns3::EmpiricalRandomVariable::EmpiricalRandomVariable ( )

Creates an empirical RNG that has a specified, empirical distribution, and configured for interpolating mode.

Definition at line 1554 of file random-variable-stream.cc.

References NS_LOG_FUNCTION.

Member Function Documentation

◆ CDF()

void ns3::EmpiricalRandomVariable::CDF ( double  v,
double  c 
)

Specifies a point in the empirical distribution.

Parameters
[in]vThe function value for this point
[in]cProbability that the function is less than or equal to v In other words this is cumulative distribution function at v.

Definition at line 1680 of file random-variable-stream.cc.

References m_empCdf, NS_LOG_FUNCTION, and NS_LOG_WARN.

◆ DoInterpolate()

double ns3::EmpiricalRandomVariable::DoInterpolate ( double  r)
private

Linear interpolation between two points on the CDF to estimate the value at r.

Parameters
[in]rThe argument value to interpolate to.
Returns
The interpolated CDF at r

Definition at line 1653 of file random-variable-stream.cc.

References m_empCdf, NS_LOG_FUNCTION, and prev.

Referenced by GetValue(), and Interpolate().

+ Here is the caller graph for this function:

◆ DoSampleCDF()

double ns3::EmpiricalRandomVariable::DoSampleCDF ( double  r)
private

Sample the CDF as a histogram (without interpolation).

Parameters
[in]rThe CDF value at which to sample the CDF.
Returns
The bin value corresponding to r.

Definition at line 1626 of file random-variable-stream.cc.

References m_empCdf, and NS_LOG_FUNCTION.

Referenced by GetValue().

+ Here is the caller graph for this function:

◆ GetInteger()

uint32_t ns3::RandomVariableStream::GetInteger
virtual

Get the next random value drawn from the distribution.

Returns
A random value.
Note
This does not interpolate the CDF, but treats it as a stepwise continuous function.

Reimplemented from ns3::RandomVariableStream.

Definition at line 155 of file random-variable-stream.cc.

◆ GetTypeId()

TypeId ns3::EmpiricalRandomVariable::GetTypeId ( void  )
static

Register this type.

Returns
The object TypeId.

Definition at line 1538 of file random-variable-stream.cc.

References m_interpolate, ns3::MakeBooleanAccessor(), ns3::MakeBooleanChecker(), and ns3::TypeId::SetParent().

+ Here is the call graph for this function:

◆ GetValue()

double ns3::EmpiricalRandomVariable::GetValue ( )
overridevirtual

Get the next random value drawn from the distribution.

Returns
A random value.
Note
This does not interpolate the CDF, but treats it as a stepwise continuous function.

Implements ns3::RandomVariableStream.

Definition at line 1603 of file random-variable-stream.cc.

References DoInterpolate(), DoSampleCDF(), m_interpolate, NS_LOG_FUNCTION, and PreSample().

+ Here is the call graph for this function:

◆ Interpolate()

double ns3::EmpiricalRandomVariable::Interpolate ( )
virtual

Returns the next value in the empirical distribution using linear interpolation.

Returns
The floating point next value in the empirical distribution using linear interpolation.

Definition at line 1637 of file random-variable-stream.cc.

References DoInterpolate(), NS_LOG_FUNCTION, and PreSample().

+ Here is the call graph for this function:

◆ PreSample()

bool ns3::EmpiricalRandomVariable::PreSample ( double &  value)
private

Do the initial rng draw and check against the extrema.

If the extrema apply, value will have the extremal value and the return will be true.

If the extrema do not apply value will have the URNG value and the return will be false.

Parameters
[out]valueThe extremal value, or the URNG.
Returns
true if value is the extremal result, or false if value is the URNG value.

Definition at line 1570 of file random-variable-stream.cc.

References ns3::RandomVariableStream::IsAntithetic(), m_empCdf, m_validated, NS_LOG_FUNCTION, ns3::RandomVariableStream::Peek(), ns3::RngStream::RandU01(), and Validate().

Referenced by GetValue(), and Interpolate().

+ Here is the call graph for this function:
+ Here is the caller graph for this function:

◆ SetInterpolate()

bool ns3::EmpiricalRandomVariable::SetInterpolate ( bool  interpolate)

Switch the mode between sampling the CDF and interpolating.

The default mode is sampling.

Parameters
[in]interpolateIf true set to interpolation, otherwise sampling.
Returns
The previous interpolate flag value.

Definition at line 1561 of file random-variable-stream.cc.

References m_interpolate, NS_LOG_FUNCTION, and prev.

◆ Validate()

void ns3::EmpiricalRandomVariable::Validate ( )
private

Check that the CDF is valid.

A valid CDF has

  • Strictly increasing arguments, and
  • Strictly increasing CDF.

It is a fatal error to fail validation.

Definition at line 1697 of file random-variable-stream.cc.

References m_empCdf, m_validated, NS_FATAL_ERROR, and NS_LOG_FUNCTION.

Referenced by PreSample().

+ Here is the caller graph for this function:

Member Data Documentation

◆ m_empCdf

std::map<double, double> ns3::EmpiricalRandomVariable::m_empCdf
private

The map of CDF points (x, F(x)).

The CDF points are stored in the std::map in reverse order, as follows: Key: CDF F(x) [0, 1] | Value: domain value (x) [-inf, inf].

Definition at line 2069 of file random-variable-stream.h.

Referenced by CDF(), DoInterpolate(), DoSampleCDF(), PreSample(), and Validate().

◆ m_interpolate

bool ns3::EmpiricalRandomVariable::m_interpolate
private

If true GetValue will interpolate, otherwise treat CDF as normal histogram.

Definition at line 2074 of file random-variable-stream.h.

Referenced by GetTypeId(), GetValue(), and SetInterpolate().

◆ m_validated

bool ns3::EmpiricalRandomVariable::m_validated
private

true once the CDF has been validated.

Definition at line 2063 of file random-variable-stream.h.

Referenced by PreSample(), and Validate().


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