class TUnuranMultiContDist: public TUnuranBaseDist


TUnuranMultiContDist class describing multi dimensional continuous distributions.
It is used by TUnuran to generate a set of random numbers according to this distribution via
TUnuran::Sample(double *).
The class can be constructed from a multi-dimensional function (TF1 pointer, which can be actually also a
TF2 or a TF3).
It provides a method to set the domain of the distribution ( SetDomain ) which will correspond to the range
of the generated random numbers. By default the domain is [(-inf,-inf,...)(+inf,+inf,...)], indipendently of the
range set in the TF1 class used to construct the distribution.

The derivatives of the pdf which are used by some UNURAN methods are estimated numerically in the
Derivative() method. Some extra information (like distribution mode) can be set using SetMode.
Some methods require instead of the pdf the log of the pdf.
This can also be controlled by setting a flag when constructing this class.

Function Members (Methods)

public:

	TUnuranMultiContDist(const TUnuranMultiContDist&)
	TUnuranMultiContDist(TF1* func = 0, unsigned int dim = 0, bool isLogPdf = false)
virtual	~TUnuranMultiContDist()
static TClass*	Class()
virtual TUnuranMultiContDist*	Clone() const
double	Derivative(const double* x, int icoord) const
const double*	GetLowerDomain() const
const double*	GetMode() const
const double*	GetUpperDomain() const
void	Gradient(const double* x, double* grad) const
virtual TClass*	IsA() const
bool	IsLogPdf() const
unsigned int	NDim() const
TUnuranMultiContDist&	operator=(const TUnuranMultiContDist& rhs)
double	Pdf(const double* x) const
void	SetDomain(const double* xmin, const double* xmax)
void	SetMode(double* x)
virtual void	ShowMembers(TMemberInspector& insp, char* parent)
virtual void	Streamer(TBuffer& b)
void	StreamerNVirtual(TBuffer& b)

Data Members

private:

unsigned int	fDim	number of function dimension
bool	fIsLogPdf	flag to control if function pointer represent log of pdf
vector<double>	fMode	vector representing the x coordinates of the maximum of the pdf
TF1*	fPdf	pointer to the pdf
vector<double>	fXmax	vector with upper x values of the domain
vector<double>	fXmin	vector with lower x values of the domain

Class Charts

Function documentation

TUnuranMultiContDist(TF1* func = 0, unsigned int dim = 0, bool isLogPdf = false)

Constructor from a TF1 object representing the Probability density funciton.
The derivatives of the Pdf are estimated, when required by the UNURAN algorithm,
using numerical derivation.
If a value of dim 0 is passed , the dimension of the function is taken from TF1::GetNdim().
This works only for 2D and 3D (for TF2 and TF3 objects).

virtual ~TUnuranMultiContDist()

Destructor (no operations)

{}

TUnuranMultiContDist * Clone()

Clone (required by base class)

{ return new TUnuranMultiContDist(*this); }

void SetDomain(const double* xmin, const double* xmax)

t the domain of the distribution giving an array of minimum and maximum valu
By default otherwise the domain is undefined, i.e. is [-inf,+inf]
To remove the domain do a SetDomain(0,0).
There is no possibility to have a domain defined in only one coordinate. Use instead inf or DOUBLE_MAX to
specify un infinite domain in that coordinate

const double * GetLowerDomain()

get the distribution lower domain values. Return a null pointer if domain is not defined

const double * GetUpperDomain()

get the distribution upper domain values. Return a null pointer if domain is not defined

const double * GetMode( )

get the mode (vector of coordinate positions of the maxima of the distribution)
If a mode has not defined return a NULL pointer

bool IsLogPdf()

lag to control if given function represent the log of a pd

{ return fIsLogPdf; }

double Derivative(const double* x, int icoord) const

evaluate the partial derivative for the given coordinate. Used by UnuRan