Package ffx.algorithms.thermodynamics

Class MonteCarloOST

java.lang.Object

ffx.algorithms.mc.BoltzmannMC

ffx.algorithms.thermodynamics.MonteCarloOST

All Implemented Interfaces:

MetropolisMC

public class MonteCarloOST
extends BoltzmannMC

Sample a thermodynamic path using the OST method, with the time-dependent bias built up using Metropolis Monte Carlo steps.

The algorithm generates coordinate (X) MC moves using molecular dynamics at a fixed lambda value (i.e., using OpenMM), followed by MC lambda moves.

1.) At a fixed Lambda, run a defined length MD trajectory to "move" coordinates and dU/dL on an approximate potential U* (i.e., no OST Bias).

2.) Accept / Reject the MD move with probability exp[-Beta(dU - dU*)] where dU is the change in AMOEBA + Bias energy and dU* is the change in AMOEBA + Kinetic energy from the MD.

3.) Randomly change the value of Lambda.

4.) Accept / Reject the Lambda move using the AMOEBA + OST Bias energy.

5.) Add to the time dependent 2D bias using the current values of Lambda and dU/dL.

Since:

1.0

Author:

Michael J. Schnieders, Hernan Beranbe, Mallory R. Tollefson, Jacob Litman

Field Summary

Fields inherited from class ffx.algorithms.mc.BoltzmannMC

random

Constructor Summary

Constructors

Constructor	Description
MonteCarloOST(Potential potentialEnergy, OrthogonalSpaceTempering orthogonalSpaceTempering, MolecularAssembly molecularAssembly, org.apache.commons.configuration2.CompositeConfiguration properties, AlgorithmListener listener, DynamicsOptions dynamics, boolean verbose, int cycleLength, File dynRestartFile)	Constructor for MonteCarloOST.

Method Summary

Modifier and Type	Method	Description
protected double	currentEnergy()	Must return the current energy of the system.
double	getLambda()	Returns the current value of lambda
MolecularDynamics	getMD()
void	revertStep()	If possible, reverts the last successful Monte Carlo step taken.
void	sampleOneStep()	The goal is to sample lambda and coordinates (X) simultaneously to converge the ensemble average dU/dL for every state (lambda) along the thermodynamic path.
void	sampleTwoStep()	The goal is to sample lambda and coordinates (X) separately to converge the ensemble average dU/dL for every state (lambda) along the thermodynamic path.
void	setAutomaticWriteouts(boolean autoWrite)
void	setEquilibration(boolean equilibration)	Sets the value of the boolean equilibration variables to true or false to either allow an equilibration step or skip it.
void	setLambda(double lambda)	Calls on the OST method set lambda to update lambda to the current value in this class
void	setLambdaStdDev(double stdDev)	Calls on LambdaMove class method setLambdaStdDev to update the lambda standard deviation to the current value in this class
void	setTotalSteps(long totalSteps)	Sets the next number of steps MC-OST should run.
protected void	storeState()	Store the state for reverting a move.

Methods inherited from class ffx.algorithms.mc.BoltzmannMC

acceptChance, evaluateMove, evaluateMove, getAccept, getE1, getE2, getTemperature, lastEnergy, mcStep, mcStep, mcStep, mcStep, setPrint, setRandomSeed, setTemperature

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

MonteCarloOST

public MonteCarloOST(Potential potentialEnergy,
 OrthogonalSpaceTempering orthogonalSpaceTempering,
 MolecularAssembly molecularAssembly,
 org.apache.commons.configuration2.CompositeConfiguration properties,
 AlgorithmListener listener,
 DynamicsOptions dynamics,
 boolean verbose,
 int cycleLength,
 File dynRestartFile)

Constructor for MonteCarloOST.

Parameters:

potentialEnergy - a Potential object.

orthogonalSpaceTempering - a OrthogonalSpaceTempering object.

molecularAssembly - a MolecularAssembly object.

properties - a CompositeConfiguration object.

listener - a AlgorithmListener object.

dynamics - CLI object containing key information.

verbose - Whether to be verbose.

cycleLength - Length of an MC cycle in MD steps.

Method Details

getLambda

public double getLambda()

Returns the current value of lambda

Returns:

lambda

setLambda

public void setLambda(double lambda)

Calls on the OST method set lambda to update lambda to the current value in this class

Parameters:

lambda - a double.

getMD

public MolecularDynamics getMD()

sampleOneStep

public void sampleOneStep()

The goal is to sample lambda and coordinates (X) simultaneously to converge the ensemble average dU/dL for every state (lambda) along the thermodynamic path. Note that the order of 1 and 2 below can be swapped (i.e. run MD and then change lambda). Here the order is random for each trial.

1.) Randomly change the value of Lambda.

2.) At the proposed lambda, run a am MD trajectory to "move" coordinates and dU/dL.

3.) Accept / Reject the Lambda + MD move using the total Hamiltonian (Kinetic energy + OST energy).

4.) Add to the bias.

sampleTwoStep

public void sampleTwoStep()

The goal is to sample lambda and coordinates (X) separately to converge the ensemble average dU/dL for every state (lambda) along the thermodynamic path.

1.) At a fixed lambda, run a defined length MD trajectory to "move" coordinates and dU/dL.

2.) Accept / Reject the MD move using the total Hamiltonian (Kinetic energy + OST energy).

3.) Randomly change the value of Lambda.

4.) Accept / Reject the Lambda move using the OST energy.

5.) Add a hill to the histogram.

setAutomaticWriteouts

public void setAutomaticWriteouts(boolean autoWrite)

setEquilibration

public void setEquilibration(boolean equilibration)

Sets the value of the boolean equilibration variables to true or false to either allow an equilibration step or skip it.

Parameters:

equilibration - a boolean.

setLambdaStdDev

public void setLambdaStdDev(double stdDev)

Calls on LambdaMove class method setLambdaStdDev to update the lambda standard deviation to the current value in this class

Parameters:

stdDev - a double.

setTotalSteps

public void setTotalSteps(long totalSteps)

Sets the next number of steps MC-OST should run.

Parameters:

totalSteps - The length of the next MC-OST run.

revertStep

public void revertStep()

If possible, reverts the last successful Monte Carlo step taken.

currentEnergy

protected double currentEnergy()

Must return the current energy of the system.

Specified by:

currentEnergy in class BoltzmannMC

Returns:

Current system energy

storeState

protected void storeState()

Store the state for reverting a move. Must be properly implemented for revertStep() to function properly; otherwise, the implementation of revertStep() should throw an OperationNotSupportedException.

Specified by:

storeState in class BoltzmannMC