npstat::GaussHermiteQuadrature Class Reference

#include <GaussHermiteQuadrature.hh>

Public Types
enum	{ interval_quadrature = 0 }

Public Member Functions
	GaussHermiteQuadrature (unsigned npoints)
unsigned	npoints () const
void	getAbscissae (long double *abscissae, unsigned len) const
void	getAllAbscissae (long double *abscissae, unsigned len) const
void	getWeights (long double *weights, unsigned len) const
void	getAllWeights (long double *weights, unsigned len) const
template<typename FcnResult , typename FcnArg >
long double	integrate (const Functor1< FcnResult, FcnArg > &fcn) const
template<typename FcnResult , typename FcnArg >
long double	integrateProb (long double mean, long double sigma, const Functor1< FcnResult, FcnArg > &fcn) const

Static Public Member Functions
static bool	isAllowed (unsigned npoints)
static std::vector< unsigned >	allowedNPonts ()
static unsigned	minimalExactRule (unsigned polyDegree)

Detailed Description

Gauss-Hermite quadrature. Internally, operations are performed in long double precision.

Constructor & Destructor Documentation

GaussHermiteQuadrature()

npstat::GaussHermiteQuadrature::GaussHermiteQuadrature ( unsigned  npoints )

explicit

At the moment, the following numbers of points are supported: 16, 32, 64, 100, 128, 256, 512.

If an unsupported number of points is given in the constructor, std::invalid_argument exception will be thrown.

Note that, for the quadrature with 512 points, the arguments become as large as 31.4 and the weights become as small as 4.9e-430. Use of such weights with doubles is rather meaningless. Long double integrands are needed.

Member Function Documentation

allowedNPonts()

static std::vector<unsigned> npstat::GaussHermiteQuadrature::allowedNPonts ( )

static

The complete set of allowed rules, in the increasing order

getAbscissae()

void npstat::GaussHermiteQuadrature::getAbscissae	(	long double *	abscissae,
		unsigned	len
	)		const

The abscissae are returned for positive points only, so the buffer length should be at least npoints/2.

getAllAbscissae()

void npstat::GaussHermiteQuadrature::getAllAbscissae	(	long double *	abscissae,
		unsigned	len
	)		const

Return abscissae for all points

getAllWeights()

void npstat::GaussHermiteQuadrature::getAllWeights	(	long double *	weights,
		unsigned	len
	)		const

Return weights for all points

getWeights()

void npstat::GaussHermiteQuadrature::getWeights	(	long double *	weights,
		unsigned	len
	)		const

The weights are returned for positive points only, so the buffer length should be at least npoints/2.

integrate()

template<typename FcnResult , typename FcnArg >

long double npstat::GaussHermiteQuadrature::integrate

(

const Functor1< FcnResult, FcnArg > &

fcn

)

const

Perform the quadrature

integrateProb()

template<typename FcnResult , typename FcnArg >

long double npstat::GaussHermiteQuadrature::integrateProb	(	long double	mean,
		long double	sigma,
		const Functor1< FcnResult, FcnArg > &	fcn
	)		const

Perform the quadrature with Gaussian density weight

isAllowed()

static bool npstat::GaussHermiteQuadrature::isAllowed ( unsigned  npoints )

static

Check if the rule with the given number of points is supported

minimalExactRule()

static unsigned npstat::GaussHermiteQuadrature::minimalExactRule ( unsigned  polyDegree )

static

Minimum number of points, among the supported rules, which integrates a polynomial with the given degree exactly (with the appropriate weight). Returns 0 if the degree is out of range.

npoints()

unsigned npstat::GaussHermiteQuadrature::npoints ( ) const

inline

Return the number of quadrature points

---

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

• nm/GaussHermiteQuadrature.hh