Detailed Description

Implements a metadynamics update scheme.

This class implements an integration scheme for metadynamics.

For a review of metadynamics, a free-energy measurement technique, see Barducci et al., Metadynamics, Wiley Interdiscipl. Rev.: Comput. Mol. Sci. 5, pp. 826-843 (2011).

Some of the features of this class are loosely inspired by the PLUMED plugin for Metadynamics, http://www.plumed-code.org/.

The IntegratorMetaDynamics class takes a number of CollectiveVariables, and uses their values during the course of the simulation to update the bias potential, a sum of Gaussians, which disfavors revisiting visited states. The derivative of the bias potential (with respect to the collective coordinate) in turn multiplies the force on the particles as defined by the collective variable.

The bias potential is updated every stride steps.

Well-tempered metadynamics is supported by default, a value of T_shift defines the temperature shift entering the scaling factor for the bias potential (see above ref. for details). Very large values of T_shift correspond to standard metadynamics, T_shift->0 corresponds to standard molecular dynamics.

Two modes of operation are supported: in the first mode, the Gaussians are resummed every time step. With increasing simulation time, this process will take longer, and slow down the simulation. In the second mode, a grid mode, a current grid of values of the collective variables is maintained and updated whenever a new Gaussian is deposited. The value of the bias potential and its derivatives of the potential are calculated by using numerical interpolation. In this mode, the time per step is O(1), so this mode is preferrable for long simulations.

It is possible to output the grid after the simulation, and to restart from the grid file. It is also possible to restart from the grid file and turn off the deposition of new Gaussians, e.g. to equilibrate the system in the bias potential landscape and measure the histogram of the collective variable, to correct for errors.

List of all members.

Public Member Functions
	IntegratorMetaDynamics (boost::shared_ptr< SystemDefinition > sysdef, Scalar deltaT, Scalar W, Scalar T_shift, unsigned int stride, bool add_hills=true, const std::string &filename="", bool overwrite=false, bool use_grid=false)
virtual void	update (unsigned int timestep)
virtual void	prepRun (unsigned int timestep)
virtual void	printStats ()
void	registerCollectiveVariable (boost::shared_ptr< CollectiveVariable > cv, Scalar sigma, Scalar cv_min=Scalar(0.0), Scalar cv_max=Scalar(0.0), int num_points=0)
void	removeAllVariables ()
std::vector< std::string >	getProvidedLogQuantities ()
Scalar	getLogValue (const std::string &quantity, unsigned int timestep)
void	setGrid (bool use_grid)
bool	isInitialized ()
void	dumpGrid (const std::string &filename)
void	restartFromGridFile (const std::string &filename)
void	setAddHills (bool add_hills)

Private Member Functions
void	updateBiasPotential (unsigned int timestep)
	Internal helper function to update the bias potential.
void	openOutputFile ()
	Helper function to open output file for logging.
void	writeFileHeader ()
	Helper function to write file header.
void	setupGrid ()
	Helper function to initialize the grid.
Scalar	interpolateBiasPotential (const std::vector< Scalar > &val)
	Helper function to get value of bias potential by multilinear interpolation.
void	readGrid (const std::string &filename)
	Helper function to read in data from grid file.

Private Attributes
Scalar	m_W
	Height of Gaussians.
Scalar	m_T_shift
	Temperature shift.
unsigned int	m_stride
	Number of timesteps between Gaussian depositions.
std::vector < CollectiveVariableItem >	m_variables
	The list of collective variables.
std::vector< std::vector < Scalar > >	m_cv_values
	History of CV values.
unsigned int	m_num_update_steps
	Number of update steps performed in this run thus far.
std::vector< std::string >	m_log_names
	Names of logging quantities.
std::string	m_suffix
	Suffix for unbiased variables.
Scalar	m_curr_bias_potential
	The sum of Gaussians.
std::vector< Scalar >	m_bias_potential
	List of values of the bias potential.
bool	m_is_initialized
	True if history-dependent potential has been initialized.
const std::string	m_filename
	Name of output file.
bool	m_overwrite
	True if the file should be overwritten.
bool	m_is_appending
	True if we are appending to the file.
std::ofstream	m_file
	Output log file.
std::string	m_delimiter
	Delimiting string.
bool	m_use_grid
	True if we are using a grid.
std::vector< Scalar >	m_grid
	d-dimensional grid to store values of bias potential
IndexGrid	m_grid_index
	Indexer for the d-dimensional grid.
bool	m_add_hills
	True if hills should be added during the simulation.
std::string	m_restart_filename
	Name of file to read restart information from.

Constructor & Destructor Documentation

IntegratorMetaDynamics	(	boost::shared_ptr< SystemDefinition >	sysdef,
		Scalar	deltaT,
		Scalar	W,
		Scalar	T_shift,
		unsigned int	stride,
		bool	add_hills = `true`,
		const std::string &	filename = `""`,
		bool	overwrite = `false`,
		bool	use_grid = `false`
	)

Constructor

Parameters:

sysdef	System definition
deltaT	Time step
W	Weight of Gaussians (multiplicative factor)
T_shift	Temperature shift for well-tempered metadynamics
stride	Number of time steps between deposition of Gaussians
add_hills	True if new Gaussians should be added during the simulation
filename	Name of file to output hill information to
overwrite	True f the file should be overwritten when it exists already
use_grid	True if grid should be used

Member Function Documentation

void dumpGrid ( const std::string & filename )

Output the grid to a file

Parameters:

filename Name of file to dump grid to

Scalar getLogValue	(	const std::string &	quantity,
		unsigned int	timestep
	)

inline

Obtain the value of a specific log quantity

Parameters:

quantity	The quantity to obtain the value of
timestep	The current value of the time step

std::vector< std::string > getProvidedLogQuantities ( )

inline

Returns the names of log quantitites provided by this integrator

bool isInitialized ( )

inline

Returns true if the integration has been initialized (i.e. the simulation has been run at least once)

void prepRun ( unsigned int timestep )

virtual

Prepare for the run

Parameters:

timestep The current value of the timestep

virtual void printStats ( )

inlinevirtual

Output statistics at end of run (unimplemented as of now)

void registerCollectiveVariable	(	boost::shared_ptr< CollectiveVariable >	cv,
		Scalar	sigma,
		Scalar	cv_min = `Scalar(0.0)`,
		Scalar	cv_max = `Scalar(0.0)`,
		int	num_points = `0`
	)

inline

Register a new collective variable

Parameters:

cv	The collective variable
sigma	The standard deviation of Gaussians for this collective variable
cv_min	Minimum value, if using grid
cv_max	Maximum value, if using grid
num_points	Number of grid points to use for interpolation

void removeAllVariables ( )

inline

Remove all collective variables

void restartFromGridFile ( const std::string & filename )

inline

Restart from a grid file. Upon running the simulation, the information will be read in.

Parameters:

filename The name of the file that contains the grid information

void setAddHills ( bool add_hills )

inline

Set a flag that controls deposition of new Gaussian hills

Parameters:

add_hills True if hills should be generated during the simulation

void setGrid ( bool use_grid )

Set up grid interpolation

Parameters:

use_grid True if the grid is to be used

void update ( unsigned int timestep )

virtual

Sample collective variables, update bias potential and derivatives

Parameters:

timestep The current value of the timestep

The documentation for this class was generated from the following files:

Detailed Description

Public Member Functions

Private Member Functions

Private Attributes

Constructor & Destructor Documentation

Member Function Documentation