Base class defining the interface common to all iterative solvers. More...

#include <IterativeSolver.h>

Inheritance diagram for molpro::linalg::itsolv::IterativeSolver< R, Q, P >:

[legend]

Detailed Description

template<class R, class Q, class P>
class molpro::linalg::itsolv::IterativeSolver< R, Q, P >

Base class defining the interface common to all iterative solvers.

As well as through the interface, some behaviour (profiling, and tuning of BLAS operations) is influenced by the contents of a global molpro::Options object, which defaults to molpro::Options("ITERATIVE-SOLVER", "").

Template Parameters

R	container for "working-set" vectors. These are typically implemented in memory, and are created by the client program. R vectors are never created inside IterativeSolver.
Q	container for other vectors. These are typically implemented on backing store and/or distributed across processors. IterativeSolver constructs a number of instances of Q containers to store history.
P	a class that specifies the definition of a single P-space vector, which is a strictly sparse vector in the underlying space.

Public Types
using	value_type = typename R::value_type
	The underlying type of elements of vectors. More...

using	scalar_type = typename array::ArrayHandler< R, Q >::value_type

using	value_type_abs = typename array::ArrayHandler< R, R >::value_type_abs

using	VectorP = std::vector< value_type >

using	fapply_on_p_type = std::function< void(const std::vector< VectorP > &, const CVecRef< P > &, const VecRef< R > &)>

Public Member Functions
virtual	~IterativeSolver ()=default

	IterativeSolver ()=default

	IterativeSolver (const IterativeSolver< R, Q, P > &)=delete

IterativeSolver< R, Q, P > &	operator= (const IterativeSolver< R, Q, P > &)=delete

	IterativeSolver (IterativeSolver< R, Q, P > &&) noexcept=default

IterativeSolver< R, Q, P > &	operator= (IterativeSolver< R, Q, P > &&) noexcept=default

virtual bool	solve (const VecRef< R > &parameters, const VecRef< R > &actions, const Problem< R > &problem, bool generate_initial_guess=false)=0
	Simplified one-call solver. More...

virtual bool	solve (R &parameters, R &actions, const Problem< R > &problem, bool generate_initial_guess=false)=0

virtual bool	solve (std::vector< R > &parameters, std::vector< R > &actions, const Problem< R > &problem, bool generate_initial_guess=false)=0

virtual int	add_vector (const VecRef< R > &parameters, const VecRef< R > &actions)=0
	Take, typically, a current solution and residual, and add it to the solution space. More...

virtual int	add_vector (std::vector< R > &parameters, std::vector< R > &action)=0

virtual int	add_vector (R &parameters, R &action, value_type value=0)=0

virtual size_t	add_p (const CVecRef< P > &pparams, const array::Span< value_type > &pp_action_matrix, const VecRef< R > &parameters, const VecRef< R > &action, fapply_on_p_type apply_p)=0
	Add P-space vectors to the expansion set for linear methods. More...

virtual void	clearP ()=0

virtual void	solution (const std::vector< int > &roots, const VecRef< R > &parameters, const VecRef< R > &residual)=0
	Construct solution and residual for a given set of roots. More...

virtual void	solution_params (const std::vector< int > &roots, const VecRef< R > &parameters)=0
	Constructs parameters of selected roots. More...

virtual size_t	end_iteration (const VecRef< R > &parameters, const VecRef< R > &residual)=0
	Behaviour depends on the solver. More...

virtual bool	end_iteration_needed ()=0
	signal whether end_iteration should be called More...

virtual std::vector< size_t >	suggest_p (const CVecRef< R > &solution, const CVecRef< R > &residual, size_t max_number, double threshold)=0
	Get the solver's suggestion of which degrees of freedom would be best to add to the P-space. More...

virtual void	solution (const std::vector< int > &roots, std::vector< R > &parameters, std::vector< R > &residual)=0

virtual void	solution (R &parameters, R &residual)=0

virtual void	solution_params (const std::vector< int > &roots, std::vector< R > &parameters)=0

virtual void	solution_params (R &parameters)=0

virtual size_t	end_iteration (std::vector< R > &parameters, std::vector< R > &action)=0

virtual size_t	end_iteration (R &parameters, R &action)=0

virtual const std::vector< int > &	working_set () const =0
	Working set of roots that are not yet converged. More...

virtual std::vector< scalar_type >	working_set_eigenvalues () const

virtual size_t	n_roots () const =0

virtual void	set_n_roots (size_t nroots)=0

virtual const std::vector< scalar_type > &	errors () const =0

virtual const Statistics &	statistics () const =0

virtual void	report (std::ostream &cout, bool endl=true) const =0
	Writes a report to cout output stream. More...

virtual void	report () const =0
	Writes a report to std::cout. More...

virtual void	set_convergence_threshold (double thresh)=0
	Sets the convergence threshold. More...

virtual double	convergence_threshold () const =0
	Reports the convergence threshold. More...

virtual void	set_convergence_threshold_value (double thresh)=0
	Sets the value convergence threshold. More...

virtual double	convergence_threshold_value () const =0
	Reports the value convergence threshold. More...

virtual void	set_verbosity (Verbosity v)=0

virtual void	set_verbosity (int v)=0

virtual Verbosity	get_verbosity () const =0

virtual void	set_max_iter (int n)=0

virtual int	get_max_iter () const =0

virtual void	set_max_p (int n)=0

virtual int	get_max_p () const =0

virtual void	set_p_threshold (double thresh)=0

virtual double	get_p_threshold () const =0

virtual const subspace::Dimensions &	dimensions () const =0

virtual void	set_options (const Options &options)=0

virtual std::shared_ptr< Options >	get_options () const =0

virtual scalar_type	value () const =0
	Report the function value for the current optimum solution. More...

virtual bool	nonlinear () const =0
	Report whether the class is a non-linear solver. More...

virtual void	set_profiler (molpro::profiler::Profiler &profiler)=0
	Attach a profiler in order to collect performance data. More...

virtual const std::shared_ptr< molpro::profiler::Profiler > &	profiler () const =0

virtual bool	test_problem (const Problem< R > &problem, R &v0, R &v1, int verbosity=0, double threshold=1e-5) const =0
	Test a supplied problem class. More...

Member Typedef Documentation

◆ fapply_on_p_type

template<class R , class Q , class P >

using molpro::linalg::itsolv::IterativeSolver< R, Q, P >::fapply_on_p_type = std::function<void(const std::vector<VectorP>&, const CVecRef<P>&, const VecRef<R>&)>

Function type for applying matrix to the P space vectors and accumulating result in a residual

◆ scalar_type

template<class R , class Q , class P >

using molpro::linalg::itsolv::IterativeSolver< R, Q, P >::scalar_type = typename array::ArrayHandler<R, Q>::value_type

The type of scalar products of vectors

◆ value_type

template<class R , class Q , class P >

using molpro::linalg::itsolv::IterativeSolver< R, Q, P >::value_type = typename R::value_type

The underlying type of elements of vectors.

◆ value_type_abs

template<class R , class Q , class P >

using molpro::linalg::itsolv::IterativeSolver< R, Q, P >::value_type_abs = typename array::ArrayHandler<R, R>::value_type_abs

◆ VectorP

template<class R , class Q , class P >

using molpro::linalg::itsolv::IterativeSolver< R, Q, P >::VectorP = std::vector<value_type>

type for vectors projected on to P space, each element is a coefficient for the corresponding P space parameter

Constructor & Destructor Documentation

◆ ~IterativeSolver()

template<class R , class Q , class P >

virtual molpro::linalg::itsolv::IterativeSolver< R, Q, P >::~IterativeSolver ( )

virtualdefault

◆ IterativeSolver() [1/3]

template<class R , class Q , class P >

molpro::linalg::itsolv::IterativeSolver< R, Q, P >::IterativeSolver ( )

default

◆ IterativeSolver() [2/3]

template<class R , class Q , class P >

molpro::linalg::itsolv::IterativeSolver< R, Q, P >::IterativeSolver ( const IterativeSolver< R, Q, P > & )

delete

◆ IterativeSolver() [3/3]

template<class R , class Q , class P >

molpro::linalg::itsolv::IterativeSolver< R, Q, P >::IterativeSolver ( IterativeSolver< R, Q, P > && )

defaultnoexcept

Member Function Documentation

◆ add_p()

template<class R , class Q , class P >

virtual size_t molpro::linalg::itsolv::IterativeSolver< R, Q, P >::add_p	(	const CVecRef< P > &	pparams,
		const array::Span< value_type > &	pp_action_matrix,
		const VecRef< R > &	parameters,
		const VecRef< R > &	action,
		fapply_on_p_type	apply_p
	)

pure virtual

Add P-space vectors to the expansion set for linear methods.

Note: the apply_p function is stored and used by the solver internally.

Parameters

Pparams	the vectors to add. Each Pvector specifies a sparse vector in the underlying space. The size of the P space must be at least the number of roots to be sought.
pp_action_matrix	Matrix projected onto the existing+new, new P space. It should be provided as a 1-dimensional array, with the existing+new index running fastest.
parameters	Used as scratch working space
action	On exit, the residual of the interpolated solution. The contribution from the new, and any existing, P parameters is missing, and should be added in subsequently.
apply_p	A function that evaluates the action of the matrix on vectors in the P space

Returns: The number of vectors contained in parameters, action, parametersP

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ add_vector() [1/3]

template<class R , class Q , class P >

virtual int molpro::linalg::itsolv::IterativeSolver< R, Q, P >::add_vector	(	const VecRef< R > &	parameters,
		const VecRef< R > &	actions
	)

pure virtual

Take, typically, a current solution and residual, and add it to the solution space.

Parameters

parameters	On input, the current solution or expansion vector. On exit, undefined.
actions	On input, the residual for parameters (non-linear), or action of matrix on parameters (linear). On exit, a vector set that should be preconditioned before returning to end_iteration().
value	The value of the objective function for parameters. Used only in Optimize classes.

Returns: The size of the new working set. In non-linear optimisation, the special value -1 can also be returned, indicating that preconditioning should not be carried out on action.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ add_vector() [2/3]

template<class R , class Q , class P >

virtual int molpro::linalg::itsolv::IterativeSolver< R, Q, P >::add_vector	(	R &	parameters,
		R &	action,
		value_type	value = `0`
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ add_vector() [3/3]

template<class R , class Q , class P >

virtual int molpro::linalg::itsolv::IterativeSolver< R, Q, P >::add_vector	(	std::vector< R > &	parameters,
		std::vector< R > &	action
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ clearP()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::clearP ( )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ convergence_threshold()

template<class R , class Q , class P >

virtual double molpro::linalg::itsolv::IterativeSolver< R, Q, P >::convergence_threshold ( ) const

pure virtual

Reports the convergence threshold.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ convergence_threshold_value()

template<class R , class Q , class P >

virtual double molpro::linalg::itsolv::IterativeSolver< R, Q, P >::convergence_threshold_value ( ) const

pure virtual

Reports the value convergence threshold.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ dimensions()

template<class R , class Q , class P >

virtual const subspace::Dimensions & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::dimensions ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ end_iteration() [1/3]

template<class R , class Q , class P >

virtual size_t molpro::linalg::itsolv::IterativeSolver< R, Q, P >::end_iteration	(	const VecRef< R > &	parameters,
		const VecRef< R > &	residual
	)

pure virtual

Behaviour depends on the solver.

Implemented in molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ end_iteration() [2/3]

template<class R , class Q , class P >

virtual size_t molpro::linalg::itsolv::IterativeSolver< R, Q, P >::end_iteration	(	R &	parameters,
		R &	action
	)

pure virtual

Implemented in molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ end_iteration() [3/3]

template<class R , class Q , class P >

virtual size_t molpro::linalg::itsolv::IterativeSolver< R, Q, P >::end_iteration	(	std::vector< R > &	parameters,
		std::vector< R > &	action
	)

pure virtual

Implemented in molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ end_iteration_needed()

template<class R , class Q , class P >

virtual bool molpro::linalg::itsolv::IterativeSolver< R, Q, P >::end_iteration_needed ( )

pure virtual

signal whether end_iteration should be called

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ errors()

template<class R , class Q , class P >

virtual const std::vector< scalar_type > & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::errors ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ get_max_iter()

template<class R , class Q , class P >

virtual int molpro::linalg::itsolv::IterativeSolver< R, Q, P >::get_max_iter ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ get_max_p()

template<class R , class Q , class P >

virtual int molpro::linalg::itsolv::IterativeSolver< R, Q, P >::get_max_p ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ get_options()

template<class R , class Q , class P >

virtual std::shared_ptr< Options > molpro::linalg::itsolv::IterativeSolver< R, Q, P >::get_options ( ) const

pure virtual

Return all options. This is no different than using getters, but can be used with forward declaration.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ get_p_threshold()

template<class R , class Q , class P >

virtual double molpro::linalg::itsolv::IterativeSolver< R, Q, P >::get_p_threshold ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ get_verbosity()

template<class R , class Q , class P >

virtual Verbosity molpro::linalg::itsolv::IterativeSolver< R, Q, P >::get_verbosity ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ n_roots()

template<class R , class Q , class P >

virtual size_t molpro::linalg::itsolv::IterativeSolver< R, Q, P >::n_roots ( ) const

pure virtual

Total number of roots we are solving for, including the ones that are already converged

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ nonlinear()

template<class R , class Q , class P >

virtual bool molpro::linalg::itsolv::IterativeSolver< R, Q, P >::nonlinear ( ) const

pure virtual

Report whether the class is a non-linear solver.

Returns

Implemented in molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ operator=() [1/2]

template<class R , class Q , class P >

IterativeSolver< R, Q, P > & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::operator= ( const IterativeSolver< R, Q, P > & )

delete

◆ operator=() [2/2]

template<class R , class Q , class P >

IterativeSolver< R, Q, P > & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::operator= ( IterativeSolver< R, Q, P > && )

defaultnoexcept

◆ profiler()

template<class R , class Q , class P >

virtual const std::shared_ptr< molpro::profiler::Profiler > & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::profiler ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ report() [1/2]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::report ( ) const

pure virtual

Writes a report to std::cout.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ report() [2/2]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::report	(	std::ostream &	cout,
		bool	endl = `true`
	)		const

pure virtual

Writes a report to cout output stream.

◆ set_convergence_threshold()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_convergence_threshold ( double thresh )

pure virtual

Sets the convergence threshold.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_convergence_threshold_value()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_convergence_threshold_value ( double thresh )

pure virtual

Sets the value convergence threshold.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_max_iter()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_max_iter ( int n )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_max_p()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_max_p ( int n )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_n_roots()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_n_roots ( size_t nroots )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_options()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_options ( const Options & options )

pure virtual

Set all spcecified options. This is no different than using setters, but can be used with forward declaration.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >, molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >, molpro::linalg::itsolv::OptimizeBFGS< R, Q, P >, and molpro::linalg::itsolv::OptimizeSD< R, Q, P >.

◆ set_p_threshold()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_p_threshold ( double thresh )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_profiler()

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_profiler ( molpro::profiler::Profiler & profiler )

pure virtual

Attach a profiler in order to collect performance data.

Parameters

profiler

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_verbosity() [1/2]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_verbosity ( int v )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ set_verbosity() [2/2]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::set_verbosity ( Verbosity v )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solution() [1/3]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solution	(	const std::vector< int > &	roots,
		const VecRef< R > &	parameters,
		const VecRef< R > &	residual
	)

pure virtual

Construct solution and residual for a given set of roots.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solution() [2/3]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solution	(	const std::vector< int > &	roots,
		std::vector< R > &	parameters,
		std::vector< R > &	residual
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solution() [3/3]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solution	(	R &	parameters,
		R &	residual
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solution_params() [1/3]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solution_params	(	const std::vector< int > &	roots,
		const VecRef< R > &	parameters
	)

pure virtual

Constructs parameters of selected roots.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solution_params() [2/3]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solution_params	(	const std::vector< int > &	roots,
		std::vector< R > &	parameters
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solution_params() [3/3]

template<class R , class Q , class P >

virtual void molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solution_params ( R & parameters )

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solve() [1/3]

template<class R , class Q , class P >

virtual bool molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solve	(	const VecRef< R > &	parameters,
		const VecRef< R > &	actions,
		const Problem< R > &	problem,
		bool	generate_initial_guess = `false`
	)

pure virtual

Simplified one-call solver.

Parameters

parameters	A set of scratch vectors. On entry, these vectors should be filled with starting guesses. Where possible, the number of vectors should be equal to the number of solutions sought, but a smaller array is permitted.
actions	A set of scratch vectors. It should have the same size as parameters.
problem	A Problem object defining the problem to be solved
generate_initial_guess	Whether to start with a guess based on diagonal elements (true) or on the contents of parameters (false)

Returns: true if the solution was found

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::LinearEquationsDavidson< R, Q, P >.

◆ solve() [2/3]

template<class R , class Q , class P >

virtual bool molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solve	(	R &	parameters,
		R &	actions,
		const Problem< R > &	problem,
		bool	generate_initial_guess = `false`
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ solve() [3/3]

template<class R , class Q , class P >

virtual bool molpro::linalg::itsolv::IterativeSolver< R, Q, P >::solve	(	std::vector< R > &	parameters,
		std::vector< R > &	actions,
		const Problem< R > &	problem,
		bool	generate_initial_guess = `false`
	)

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ statistics()

template<class R , class Q , class P >

virtual const Statistics & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::statistics ( ) const

pure virtual

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ suggest_p()

template<class R , class Q , class P >

virtual std::vector< size_t > molpro::linalg::itsolv::IterativeSolver< R, Q, P >::suggest_p	(	const CVecRef< R > &	solution,
		const CVecRef< R > &	residual,
		size_t	max_number,
		double	threshold
	)

pure virtual

Get the solver's suggestion of which degrees of freedom would be best to add to the P-space.

Parameters

solution	Current solution
residual	Current residual
max_number	Suggest no more than this number
threshold	Suggest only axes for which the current residual and update indicate an energy improvement in the next iteration of this amount or more.

Returns

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ test_problem()

template<class R , class Q , class P >

virtual bool molpro::linalg::itsolv::IterativeSolver< R, Q, P >::test_problem	(	const Problem< R > &	problem,
		R &	v0,
		R &	v1,
		int	verbosity = `0`,
		double	threshold = `1e-5`
	)		const

pure virtual

Test a supplied problem class.

Parameters

problem
v0
v1
v2

Returns: false if errors were found, otherwise true

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ value()

template<class R , class Q , class P >

virtual scalar_type molpro::linalg::itsolv::IterativeSolver< R, Q, P >::value ( ) const

pure virtual

Report the function value for the current optimum solution.

Returns

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ working_set()

template<class R , class Q , class P >

virtual const std::vector< int > & molpro::linalg::itsolv::IterativeSolver< R, Q, P >::working_set ( ) const

pure virtual

Working set of roots that are not yet converged.

Implemented in molpro::linalg::itsolv::IterativeSolverTemplate< LinearEigensystem, R, R, std::map< size_t, typename R::value_type > >, molpro::linalg::itsolv::IterativeSolverTemplate< LinearEquations, R, R, std::map< size_t, typename R::value_type > >, and molpro::linalg::itsolv::IterativeSolverTemplate< Optimize, R, R, std::map< size_t, typename R::value_type > >.

◆ working_set_eigenvalues()

template<class R , class Q , class P >

virtual std::vector< scalar_type > molpro::linalg::itsolv::IterativeSolver< R, Q, P >::working_set_eigenvalues ( ) const

inlinevirtual

The calculated eigenvalues for roots in the working set (eigenvalue problems) or zero (otherwise)

Reimplemented in molpro::linalg::itsolv::LinearEigensystemDavidson< R, Q, P >, and molpro::linalg::itsolv::LinearEigensystemRSPT< R, Q, P >.

Detailed Description

Public Types

Public Member Functions

Member Typedef Documentation

◆ fapply_on_p_type

◆ scalar_type

◆ value_type

◆ value_type_abs

◆ VectorP

Constructor & Destructor Documentation

◆ ~IterativeSolver()

◆ IterativeSolver() [1/3]

◆ IterativeSolver() [2/3]

◆ IterativeSolver() [3/3]

Member Function Documentation

◆ add_p()

◆ add_vector() [1/3]

◆ add_vector() [2/3]

◆ add_vector() [3/3]

◆ clearP()

◆ convergence_threshold()

◆ convergence_threshold_value()

◆ dimensions()

◆ end_iteration() [1/3]

◆ end_iteration() [2/3]

◆ end_iteration() [3/3]

◆ end_iteration_needed()

◆ errors()

◆ get_max_iter()

◆ get_max_p()

◆ get_options()

◆ get_p_threshold()

◆ get_verbosity()

◆ n_roots()

◆ nonlinear()

◆ operator=() [1/2]

◆ operator=() [2/2]

◆ profiler()

◆ report() [1/2]

◆ report() [2/2]

◆ set_convergence_threshold()

◆ set_convergence_threshold_value()

◆ set_max_iter()

◆ set_max_p()

◆ set_n_roots()

◆ set_options()

◆ set_p_threshold()

◆ set_profiler()

◆ set_verbosity() [1/2]

◆ set_verbosity() [2/2]

◆ solution() [1/3]

◆ solution() [2/3]

◆ solution() [3/3]

◆ solution_params() [1/3]

◆ solution_params() [2/3]

◆ solution_params() [3/3]

◆ solve() [1/3]

◆ solve() [2/3]

◆ solve() [3/3]

◆ statistics()

◆ suggest_p()

◆ test_problem()

◆ value()

◆ working_set()

◆ working_set_eigenvalues()