T_WL_Data_Structure_Internal< Conformation, Bin_properties, DS_snapshot_ > Class Template Reference

bin management: main data structure used by Wang_landau algorithm More...

#include <WL_data_structure_internal.hpp>

Public Member Functions
virtual Snapshot_ptr	update_increment ()
	update_increment is called after new points have been generated and accepted(or not) in the WL algo. it updates the increment by which the (log) density is increased when a new point is generated in a given bin.
void	update_densities (const Bin_index &index)
	update_densities updates the density vector when a new point is sampled by the WL algorithm.
std::pair< bool, Snapshot_ptr >	energy_out_of_range (const DS_conformation &conf)
	energy_out_of_range is called when a point is sampled in a not already existing bin. It should handle the creation of a new bin if supported. This should be modified to initialize Custom_per_bin_stats at propers values if needed.
void	update_histogram (const DS_conformation &newpt, const DS_conformation &oldpt, const bool &accepted, const bool &boundary_check)
	update_histogram is called when a new point is sampled. It should be used to store some statistics on how the algorithm is working, which might be later used in update_parameters()
double &	get_density (const Bin_index &index)
	get_density returns the density of a point
Bin_index	get_bin_index_from_energy (const double &e)
	get_bin_index_from_energy finds the bin containing the energy e
Bin_index	get_bin_index_from_energy (const double &e, const double &e1, const Bin_index &i1)
	get_bin_index_from_energy should uses that the point previously sampled should be in a bin close to the new newly sampled point
bool	is_energy_in_range (const DS_conformation &conf)
	is_energy_in_range checks if a Bin_index is valid.

Protected Member Functions
Bin_index	add_bin (Bin_energy_range bin)
	add_bin is a helper function which add a bin to the bin vector/ properties vector

Detailed Description

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>
class SBL::GT::T_WL_Data_Structure_Internal< Conformation, Bin_properties, DS_snapshot_ >

bin management: main data structure used by Wang_landau algorithm

• manages bins (creation, index of bin for point, etc)
• handles the flat histogram or 1/t policy
• performs the DoS update for a new point
• records all relevant stats

Member Function Documentation

add_bin()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

Bin_index add_bin ( Bin_energy_range bin )

inlineprotected

add_bin is a helper function which add a bin to the bin vector/ properties vector

Parameters

bin

Returns

energy_out_of_range()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

std::pair< bool, Snapshot_ptr > energy_out_of_range ( const DS_conformation & conf )

inline

energy_out_of_range is called when a point is sampled in a not already existing bin. It should handle the creation of a new bin if supported. This should be modified to initialize Custom_per_bin_stats at propers values if needed.

Parameters

conf

Returns

false is returned if this case is not supported. WL algorithm should then quit gracefully.

get_bin_index_from_energy() [1/2]

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

Bin_index get_bin_index_from_energy ( const double & e )

inline

get_bin_index_from_energy finds the bin containing the energy e

Parameters

e

Returns

return a Bin_index describing the bin contatining energy e. If no bin contains e, an invalid bin is returned (here -1)

get_bin_index_from_energy() [2/2]

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

Bin_index get_bin_index_from_energy ( const double & e, const double & e1, const Bin_index & i1 )

inline

get_bin_index_from_energy should uses that the point previously sampled should be in a bin close to the new newly sampled point

Parameters

e	new energy
e1	previous energy
i1	previous bin

Returns

get_density()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

double & get_density ( const Bin_index & index )

inline

get_density returns the density of a point

Parameters

index

Returns

is_energy_in_range()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

bool is_energy_in_range ( const DS_conformation & conf )

inline

is_energy_in_range checks if a Bin_index is valid.

Parameters

conf

Returns

update_densities()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

void update_densities ( const Bin_index & index )

inline

update_densities updates the density vector when a new point is sampled by the WL algorithm.

Parameters

index

update_histogram()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

void update_histogram ( const DS_conformation & newpt, const DS_conformation & oldpt, const bool & accepted, const bool & boundary_check )

inline

update_histogram is called when a new point is sampled. It should be used to store some statistics on how the algorithm is working, which might be later used in update_parameters()

Parameters

oldpt	previous point in the algorithm
newpt	point sampled by the Point_generator
accepted	is the new point accepted after the metropolis test
boundary_check	is the result of the boundary check

update_increment()

template<class Conformation, class Bin_properties = Bin_properties_Internal, template< class >class DS_snapshot_ = T_DS_snapshot_default>

virtual Snapshot_ptr update_increment ( )

inlinevirtual

update_increment is called after new points have been generated and accepted(or not) in the WL algo. it updates the increment by which the (log) density is increased when a new point is generated in a given bin.