molpro::linalg::itsolv::Interpolator< value_type > Class Template Reference

The interpolant, in the precision of the problem it is used for. More...

#include <Interpolate.h>

Detailed Description

template<typename value_type = double>
class molpro::linalg::itsolv::Interpolator< value_type >

The interpolant, in the precision of the problem it is used for.

The parameter is a real type, and names a precision rather than a scalar type: minimising a function presupposes that its values are ordered, so a complex instantiation is meaningless and is rejected below. What varies is how accurately the line search is carried out. Optimize passes value_type_abs, which for a std::complex<double> container is double, and for a long double one is long double – the case this template exists for, since narrowing the function values and gradients to double would cap the line search at double precision however the solver was instantiated.

See also

Interpolate for the double-precision specialisation, which is the historical name and the one used throughout the double-precision API.

Classes
struct	point

Public Types
using	value_t = value_type

Public Member Functions
	Interpolator (point p0, point p1, std::string interpolant="cubic", int verbosity=0)
	Construct the interpolant. More...
point	operator() (value_type x) const
	Evaluate the interpolant and its derivative at a given point. More...
point	minimize (value_type xa, value_type xb, size_t bracket_grid=100, size_t max_bracket_grid=100000, bool analytic=true) const
	Find the minimum of the interpolant within a range. More...
point	minimize_cubic () const
const std::vector< value_type > &	parameters () const

Static Public Member Functions
static std::vector< std::string >	interpolants ()

Member Typedef Documentation

value_t

template<typename value_type = double>

using molpro::linalg::itsolv::Interpolator< value_type >::value_t = value_type

Constructor & Destructor Documentation

Interpolator()

template<typename value_type >

molpro::linalg::itsolv::Interpolator< value_type >::Interpolator ( point  p0, point  p1, std::string  interpolant = "cubic", int  verbosity = 0  )

explicit

Construct the interpolant.

Parameters

p0	Defining point
p1	Defining point
interpolant	The interpolation method. An exception is thrown if it is not one of the implemented values.
verbosity	Values greater than zero show information on constructing and using the interpolant.

Member Function Documentation

interpolants()

template<typename value_type >

std::vector< std::string > molpro::linalg::itsolv::Interpolator< value_type >::interpolants

static

minimize()

template<typename value_type >

Interpolator< value_type >::point molpro::linalg::itsolv::Interpolator< value_type >::minimize	(	value_type	xa,
		value_type	xb,
		size_t	bracket_grid = 100,
		size_t	max_bracket_grid = 100000,
		bool	analytic = true
	)		const

Find the minimum of the interpolant within a range.

Parameters

xa	first bound of range
xb	second bound of range
bracket_grid	number of intervals in |xa-xb|, to be considered in initial bracketing of the minimum. Large values result in many function evaluations, but if set too small in cases of multiple minima, the global minimum may not be found.

Returns

The minimum point. The result may be one of Interpolate(xa), Interpolate(xb) with non-zero first derivative, if no other minimum was found in the interval

minimize_cubic()

template<typename value_type >

Interpolator< value_type >::point molpro::linalg::itsolv::Interpolator< value_type >::minimize_cubic

operator()()

template<typename value_type >

Interpolator< value_type >::point molpro::linalg::itsolv::Interpolator< value_type >::operator()

(

value_type

x

)

const

Evaluate the interpolant and its derivative at a given point.

Parameters

x

Returns

parameters()

template<typename value_type = double>

const std::vector< value_type > & molpro::linalg::itsolv::Interpolator< value_type >::parameters ( ) const

inline