ippl::random::detail::NewtonRaphson< T, Distribution > Struct Template Reference

Functor for solving equations using the Newton-Raphson method. More...

#include <Utility.h>

Collaboration diagram for ippl::random::detail::NewtonRaphson< T, Distribution >:

Public Member Functions
KOKKOS_FUNCTION	NewtonRaphson ()=default
KOKKOS_FUNCTION	~NewtonRaphson ()=default
KOKKOS_INLINE_FUNCTION	NewtonRaphson (const Distribution &dist_)
KOKKOS_INLINE_FUNCTION void	solve (unsigned int d, T &x, T &u)
	Solve an equation using the Newton-Raphson method.

Public Attributes
Distribution	dist
double	atol = 1e-12
unsigned int	max_iter = 20

Detailed Description

template<typename T, class Distribution>
struct ippl::random::detail::NewtonRaphson< T, Distribution >

Functor for solving equations using the Newton-Raphson method.

In particular, find the root x of the equation dist.obj(x, u)= 0 for a given u using Newton-Raphson method.

Template Parameters

T	Data type for the equation variables.
Distribution	Class of target distribution to sample from.

Parameters

dist	Distribution object providing objective function cdf(x)-u and its derivative.
atol	Absolute tolerance for convergence (default: 1.0e-12).
max_iter	Maximum number of iterations (default: 20).

Definition at line 28 of file Utility.h.

Constructor & Destructor Documentation

NewtonRaphson() [1/2]

template<typename T, class Distribution>

KOKKOS_FUNCTION ippl::random::detail::NewtonRaphson< T, Distribution >::NewtonRaphson ( )

default

~NewtonRaphson()

template<typename T, class Distribution>

KOKKOS_FUNCTION ippl::random::detail::NewtonRaphson< T, Distribution >::~NewtonRaphson ( )

default

NewtonRaphson() [2/2]

template<typename T, class Distribution>

KOKKOS_INLINE_FUNCTION ippl::random::detail::NewtonRaphson< T, Distribution >::NewtonRaphson ( const Distribution & dist_ )

inline

Definition at line 39 of file Utility.h.

References dist.

Member Function Documentation

solve()

template<typename T, class Distribution>

KOKKOS_INLINE_FUNCTION void ippl::random::detail::NewtonRaphson< T, Distribution >::solve ( unsigned int d, T & x, T & u )

inline

Solve an equation using the Newton-Raphson method.

This function iteratively solves an equation of the form "cdf(x) - u = 0" for a given sample u using the Newton-Raphson method.

Parameters

d	Dimension index.
x	Variable to solve for (initial guess and final solution).
u	Given sample from a uniform distribution [0, 1].

Definition at line 52 of file Utility.h.

References atol, dist, and max_iter.

Referenced by ippl::random::InverseTransformSampling< T, Dim, DeviceType, Distribution >::fill_random< GeneratorPool >::operator()().

Here is the caller graph for this function:

Member Data Documentation

atol

template<typename T, class Distribution>

double ippl::random::detail::NewtonRaphson< T, Distribution >::atol = 1e-12

Definition at line 30 of file Utility.h.

Referenced by solve().

dist

template<typename T, class Distribution>

Distribution ippl::random::detail::NewtonRaphson< T, Distribution >::dist

Definition at line 29 of file Utility.h.

Referenced by NewtonRaphson(), and solve().

max_iter

template<typename T, class Distribution>

unsigned int ippl::random::detail::NewtonRaphson< T, Distribution >::max_iter = 20

Definition at line 31 of file Utility.h.

Referenced by solve().

---

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

• Utility.h