Convolutional Kernels¶

Provides the kernel classes for the convolution done when computing the weighted density of the points.

Authors: Simone Balducci, Felice Pantaleo, Marco Rovere, Wahid Redjeb, Aurora Perego, Francesco Giacomini

namespace clue¶

template<std::floating_point TData = float> class ExponentialKernel¶

#include <ConvolutionalKernel.hpp>

The ExponentialKernel class implements an exponential kernel for convolution. It computes the kernel value based on the exponential function, which is defined by its average and amplitude.

Template Parameters:¶

std::floating_point TData = float¶: The data type for the kernel values

Public Types¶

using value_type = std::remove_cv_t<std::remove_reference_t<TData>>¶

Public Functions¶

ExponentialKernel(value_type exp_avg, value_type exp_amplitude)¶

Construct an ExponentialKernel object.

Parameters:¶

value_type exp_avg¶: The average value for the exponential kernel
value_type exp_amplitude¶: The amplitude for the exponential kernel

template<typename TAcc> ALPAKA_FN_ACC auto operator()(const TAcc &acc, value_type dist_ij, int point_id, int j) const¶

Computes the kernel value between two points.

Parameters:¶

const TAcc &acc¶: The accelerator to use for the computation
value_type dist_ij¶: The distance between the two points
int point_id¶: The index of the first point
int j¶: The index of the second point

Returns:¶

The computed kernel value

Private Members¶

value_type m_exp_avg¶

value_type m_exp_amplitude¶

template<std::floating_point TData = float> class FlatKernel¶

#include <ConvolutionalKernel.hpp>

The FlatKernel class implements a flat kernel for convolution. It returns a constant value for the kernel, regardless of the distance between points.

Template Parameters:¶

std::floating_point TData = float¶: The data type for the kernel values

Public Types¶

using value_type = std::remove_cv_t<std::remove_reference_t<TData>>¶

Public Functions¶

FlatKernel(value_type flat)¶

Construct a FlatKernel object.

Parameters:¶

value_type flat¶: The flat value for the kernel

template<typename TAcc> ALPAKA_FN_ACC auto operator()(const TAcc &acc, value_type dist_ij, int point_id, int j) const¶

Computes the kernel value between two points.

Parameters:¶

const TAcc &acc¶: The accelerator to use for the computation
value_type dist_ij¶: The distance between the two points
int point_id¶: The index of the first point
int j¶: The index of the second point

Returns:¶

The computed kernel value

Private Members¶

value_type m_flat¶

template<std::floating_point TData = float> class GaussianKernel¶

#include <ConvolutionalKernel.hpp>

The GaussianKernel class implements a Gaussian kernel for convolution. It computes the kernel value based on the Gaussian function, which is defined by its average, standard deviation, and amplitude.

Template Parameters:¶

std::floating_point TData = float¶: The data type for the kernel values

Public Types¶

using value_type = std::remove_cv_t<std::remove_reference_t<TData>>¶

Public Functions¶

GaussianKernel(value_type gaus_avg, value_type gaus_std, value_type gaus_amplitude)¶

Construct a GaussianKernel object.

Parameters:¶

value_type gaus_avg¶: The average value for the Gaussian kernel
value_type gaus_std¶: The standard deviation for the Gaussian kernel
value_type gaus_amplitude¶: The amplitude for the Gaussian kernel

template<typename TAcc> ALPAKA_FN_ACC auto operator()(const TAcc &acc, value_type dist_ij, int point_id, int j) const¶

Computes the kernel value between two points.

Parameters:¶

const TAcc &acc¶: The accelerator to use for the computation
value_type dist_ij¶: The distance between the two points
int point_id¶: The index of the first point
int j¶: The index of the second point

Returns:¶

The computed kernel value

Private Members¶

value_type m_gaus_avg¶

value_type m_gaus_std¶

value_type m_gaus_amplitude¶

namespace concepts¶