lsst::afw::math::NeuralNetCovariogram< T > Class Template Reference

a Covariogram that recreates a neural network with one hidden layer and infinite units in that layer More...

#include <GaussianProcess.h>

Inheritance diagram for lsst::afw::math::NeuralNetCovariogram< T >:

Public Member Functions
	~NeuralNetCovariogram () override
	NeuralNetCovariogram ()
void	setSigma0 (double sigma0)
	set the _sigma0 hyper parameter
void	setSigma1 (double sigma1)
	set the _sigma1 hyper parameter
T	operator() (ndarray::Array< const T, 1, 1 > const &, ndarray::Array< const T, 1, 1 > const &) const override
	Actually evaluate the covariogram function relating two points you want to interpolate from.

Detailed Description

template<typename T>
class lsst::afw::math::NeuralNetCovariogram< T >

a Covariogram that recreates a neural network with one hidden layer and infinite units in that layer

Contains two hyper parameters (_sigma0 and _sigma1) that characterize the expected variance of the function being interpolated

see Rasmussen and Williams (2006) http://www.gaussianprocess.org/gpml/ equation 4.29

Definition at line 193 of file GaussianProcess.h.

Constructor & Destructor Documentation

~NeuralNetCovariogram()

template<typename T>

lsst::afw::math::NeuralNetCovariogram< T >::~NeuralNetCovariogram ( )

overridedefault

NeuralNetCovariogram()

template<typename T>

lsst::afw::math::NeuralNetCovariogram< T >::NeuralNetCovariogram ( )

explicit

Definition at line 2025 of file GaussianProcess.cc.

                                              {
    _sigma0 = 1.0;
    _sigma1 = 1.0;
}

Member Function Documentation

operator()()

template<typename T>

T lsst::afw::math::NeuralNetCovariogram< T >::operator() ( ndarray::Array< const T, 1, 1 > const & p1, ndarray::Array< const T, 1, 1 > const & p2 ) const

overridevirtual

Actually evaluate the covariogram function relating two points you want to interpolate from.

Parameters

[in]	p1	the first point
[in]	p2	the second point

Reimplemented from lsst::afw::math::Covariogram< T >.

Definition at line 2031 of file GaussianProcess.cc.

                                                                                   {
    int i, dim;
    double num, denom1, denom2, arg;
 
    dim = p1.template getSize<0>();
 
    num = 2.0 * _sigma0;
    denom1 = 1.0 + 2.0 * _sigma0;
    denom2 = 1.0 + 2.0 * _sigma0;
 
    for (i = 0; i < dim; i++) {
        num += 2.0 * p1[i] * p2[i] * _sigma1;
        denom1 += 2.0 * p1[i] * p1[i] * _sigma1;
        denom2 += 2.0 * p2[i] * p2[i] * _sigma1;
    }
 
    arg = num / ::sqrt(denom1 * denom2);
    return T(2.0 * (::asin(arg)) / 3.141592654);
}

setSigma0()

template<typename T>

void lsst::afw::math::NeuralNetCovariogram< T >::setSigma0

(

double

sigma0

)

set the _sigma0 hyper parameter

Definition at line 2053 of file GaussianProcess.cc.

                                                     {
    _sigma0 = sigma0;
}

setSigma1()

template<typename T>

void lsst::afw::math::NeuralNetCovariogram< T >::setSigma1

(

double

sigma1

)

set the _sigma1 hyper parameter

Definition at line 2058 of file GaussianProcess.cc.

                                                     {
    _sigma1 = sigma1;
}

---

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

• /j/snowflake/release/lsstsw/stack/lsst-scipipe-10.1.0/Linux/afw/g4595892280+3897dae354/include/lsst/afw/math/GaussianProcess.h
• /j/snowflake/release/lsstsw/stack/lsst-scipipe-10.1.0/Linux/afw/g4595892280+3897dae354/src/math/GaussianProcess.cc