Model agnostic implementation of various multicanonical methods. More...

#include <stdio.h>
#include <vector>
#include <algorithm>
#include <regex>
#include <cmath>
#include "hila.h"
#include "tools/multicanonical.h"

Include dependency graph for multicanonical.cpp:

Go to the source code of this file.

Classes
struct	canonical_iteration
	An internal struct parametrising the canonical weight iteration method. More...

struct	direct_iteration
	An internal struct parametrising the direct weight iteration method. More...

struct	weight_iteration_parameters
	An internal struct parametrising multicanonical methods. More...

Namespaces
namespace	hila
	Implement hila::swap for gauge fields.

Functions
template<class K >
void	hila::muca::to_file (std::ofstream &output_file, string fmt, K input_value)
	Writes a variable to the file, given the format string.

string	hila::muca::generate_outfile_name ()
	Generates a time stamped and otherwise appropriate file name for the saved weight function files.

void	hila::muca::read_weight_parameters (string parameter_file_name)
	Parses the weight parameter file and fills the g_WParam struct.

void	hila::muca::read_weight_function (string W_function_filename)
	Reads a precomputed weight function from file.

void	hila::muca::write_weight_function (string W_function_filename)
	Reads the precomputed weight function from run_parameters struct and saves it into a file.

double	hila::muca::weight_function (double OP)
	Returns a weight associated to the used order parameter.

double	hila::muca::weight (double OP)
	process 0 interface to "weight function" for the user accessing the weights.

void	hila::muca::set_weight_iter_flag (bool YN)
	Sets the static g_WeightIterationFlag to given boolean.

bool	hila::muca::check_weight_iter_flag ()
	Returns the value of the static g_WeightIterationFlag to user.

bool	hila::muca::accept_reject (const double OP_old, const double OP_new)
	Accepts/rejects a multicanonical update.

void	hila::muca::set_direct_iteration_FC (bool(*fc_pointer)(int_vector &n))
	Sets a user provided function to the check in the "direct iteration" method.

void	hila::muca::set_continuous_iteration (bool YN)
	Enable/disable continuous weight iteration.

void	hila::muca::initialise (const string wfile_name)
	Loads parameters and weights for the multicanonical computation.

Detailed Description

Model agnostic implementation of various multicanonical methods.

Author: Jaakko Hällfors

TBA

Definition in file multicanonical.cpp.

Function Documentation

◆ accept_reject()

bool hila::muca::accept_reject	(	const double	OP_old,
		const double	OP_new
	)

Accepts/rejects a multicanonical update.

Using the values of the old and new order parameters the muca update is accepted with the logarithmic probability log(P) = - (W(OP_new) - W(OP_old))

Parameters

OP_old	current order parameter
OP_new	order parameter of proposed configuration

Returns: Boolean indicating whether the update was accepted (true) or rejected (false).

Definition at line 543 of file multicanonical.cpp.

◆ check_weight_iter_flag()

bool hila::muca::check_weight_iter_flag ( )

Returns the value of the static g_WeightIterationFlag to user.

Returns: State of g_WeighITerationFlag

Definition at line 524 of file multicanonical.cpp.

◆ generate_outfile_name()

std::string hila::muca::generate_outfile_name ( )

Generates a time stamped and otherwise appropriate file name for the saved weight function files.

Returns: generated filename string

Definition at line 209 of file multicanonical.cpp.

◆ initialise()

void hila::muca::initialise ( const string wfile_name )

Loads parameters and weights for the multicanonical computation.

Sets up iteration variables. Can be called multiple times and must be called at least once before attempting to use any of the muca methods.

Parameters

wfile_name path to the weight parameter file

Definition at line 1188 of file multicanonical.cpp.

◆ read_weight_function()

void hila::muca::read_weight_function ( string W_function_filename )

Reads a precomputed weight function from file.

The input file is to have three tab-separated columns:

bin edges, of length n + 1
bin centres, of length n
weight values, of length n

Naturally, column 2 can be determined from the first one but they are considered a separated input anyways.

The header can be whatever* and is always skipped. Regex finds the data by finding first row with the substring "OP_value" and assumes that the following contains the data as specified above.

*Avoid substring "OP_value"

Parameters

W_function_filename

Definition at line 329 of file multicanonical.cpp.

◆ read_weight_parameters()

void hila::muca::read_weight_parameters ( string parameter_file_name )

Parses the weight parameter file and fills the g_WParam struct.

Parameters

parameter_file_name parameter file name

Definition at line 230 of file multicanonical.cpp.

◆ set_continuous_iteration()

void hila::muca::set_continuous_iteration ( bool YN )

Enable/disable continuous weight iteration.

Premits the user to enable/disable continuous weight iteration at each call to accept_reject. Simply modifies a flag parameter that is checked in accept_reject.

Parameters

YN	enable (true) or disable (false) the iteration

Definition at line 1175 of file multicanonical.cpp.

◆ set_direct_iteration_FC()

void hila::muca::set_direct_iteration_FC ( bool(*)(int_vector &n) fc_pointer )

Sets a user provided function to the check in the "direct iteration" method.

The for a given magnitude of update the "direct iteration" method periodically checks whether the MCMC chain has covered enough of the desired order parameter range, before reducing the update magnitude. Some preset methods exist (and should suffice) but when needed, a new condition can be set through this function. The input is a vector of integers indicating the number of visits to each bin, and the output is a boolean telling whether the desired condition has been achieved.

Parameters

fc_pointer A function pointer to a suitable condition function

Definition at line 688 of file multicanonical.cpp.

◆ set_weight_iter_flag()

void hila::muca::set_weight_iter_flag ( bool YN )

Sets the static g_WeightIterationFlag to given boolean.

Parameters

YN	boolean indicating whether the iteration is to continue

Definition at line 514 of file multicanonical.cpp.

◆ to_file()

template<class K >

void hila::muca::to_file	(	std::ofstream &	output_file,
		string	fmt,
		K	input_value
	)

Writes a variable to the file, given the format string.

Parameters

output_file
fmt	format string corresponding to input_value
input_value	numerical value to write to output_file

Definition at line 196 of file multicanonical.cpp.

◆ weight()

double hila::muca::weight ( double OP )

process 0 interface to "weight function" for the user accessing the weights.

Definition at line 498 of file multicanonical.cpp.

◆ weight_function()

double hila::muca::weight_function ( double OP )

Returns a weight associated to the used order parameter.

The function uses supplied pairs of points to linearly interpolate the function on the interval. This interpolant provides the requested weights to be used as the multicanonical weight.

Parameters

OP	value of the order parameter

Returns: The value of the weight.

Definition at line 455 of file multicanonical.cpp.

◆ write_weight_function()

void hila::muca::write_weight_function ( string W_function_filename )

Reads the precomputed weight function from run_parameters struct and saves it into a file.

The printing happens in an format identical to what is expected by the funciton read_weight_function. See its documentation for details. TBA: Add string input that can contain user specified header data.

Parameters

W_function_filename
g_WParam	struct of weight iteration parameters

Definition at line 420 of file multicanonical.cpp.

Classes

Namespaces

Functions

Detailed Description

Function Documentation

◆ accept_reject()

◆ check_weight_iter_flag()

◆ generate_outfile_name()

◆ initialise()

◆ read_weight_function()

◆ read_weight_parameters()

◆ set_continuous_iteration()

◆ set_direct_iteration_FC()

◆ set_weight_iter_flag()

◆ to_file()

◆ weight()

◆ weight_function()

◆ write_weight_function()