meshi.optimizers
Class SteepestDecent

java.lang.Object
  meshi.optimizers.Optimizer
      meshi.optimizers.Minimizer
          meshi.optimizers.SteepestDecent

public class SteepestDecent
extends Minimizer

This class implements a simple steepest descent minimizer, using a simple back tracking line search. The search direction is given by the gradient of the energy function, at the minimizer position. The step length first tried is the previous iteration step length, multiplied by some expansion factor. If it leads to energy reduction it is choosen, otherwise it is shortened by some factor and tried again. This repeats until energy reduction is achieved. To run this minimizer: a) Instantiate this class with the desired minimization parameters. b) Put the initial coordinates in the 'coordinates' variable at the 'energy' class. c) Activate SteepestDecent.run(). d) Check for thrown errors to see if the minimization succeeded. e) The minimized position is in the 'coordinates' variable at the 'energy' class. Parameters in the full constructor: ---------------------------------- - energy - pointer to a TotalEnergy object, where the energy function is. - tolerance - Minimization stops when the magnitude of the maximal gradient component drops below tolerance. - maxIteration - The maximal number of iteration steps allowed - reoprtEvery - The frequency of the minimization reports. - initialStepLength - parameter of the line search. The first step length to be tried after the calculation of the first gradient. This parameter should normally be 1 unless very large gradients (such as clashhing of VDW atoms) are expected in the first steps. In that case it should be a much smaller number. - stepSizeReduction - parameter of the line search. The step length is multiplied by this factor if no reduction in energy is achieved. - stepSizeExpansion - parameter of the line search. The first step length tried is the step length from previous line search multiplied by this factor. (Note that non-positive values to this paramater cause special options to be called (see the SimpleStepLength class help).

Nested Class Summary

Nested classes/interfaces inherited from class meshi.optimizers.Optimizer

Optimizer.OptimizerStatus

Field Summary

private double[][]	bufferCoordinates
private double[][]	coordinates
private static double	DEFAULT_INITIAL_STEP_LENGTH
private static int	DEFAULT_MAX_ITERATION
private static int	DEFAULT_REPORT_EVERY
private static double	DEFAULT_STEP_SIZE_EXPENTION
private static double	DEFAULT_STEP_SIZE_REDUCTION
private static double	DEFAULT_TOLERANCE
private double	initialStepLength
private int	iterationNum
private double	lastStepLength
private SimpleStepLength	lineSearch
private double	magnitudeForce
private int	reportEvery
private double	stepSizeExpansion
private double	stepSizeReduction

Fields inherited from class meshi.optimizers.Minimizer

MAX_KICKSTARTS, terminator, tolerance

Fields inherited from class meshi.optimizers.Optimizer

energy, maxSteps, optimizerTerminator

Constructor Summary

SteepestDecent(TotalEnergy energy)

SteepestDecent(TotalEnergy energy, double tolerance, int maxSteps, int reportEvery)

SteepestDecent(TotalEnergy energy, double tolerance, int maxSteps, int reportEvery, double initialStepLength, double stepSizeReduction, double stepSizeExpansion)

Method Summary

protected void	init()
protected void	kickStart()
double	lastStepLength()
protected boolean	minimizationStep()
java.lang.String	toString()

Methods inherited from class meshi.optimizers.Minimizer

run, run

Methods inherited from class meshi.optimizers.Optimizer

energy

Methods inherited from class java.lang.Object

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

Field Detail

lineSearch

private SimpleStepLength lineSearch

coordinates

private double[][] coordinates

bufferCoordinates

private double[][] bufferCoordinates

iterationNum

private int iterationNum

lastStepLength

private double lastStepLength

magnitudeForce

private double magnitudeForce

DEFAULT_TOLERANCE

private static final double DEFAULT_TOLERANCE

See Also:

Constant Field Values

DEFAULT_MAX_ITERATION

private static final int DEFAULT_MAX_ITERATION

See Also:

Constant Field Values

DEFAULT_REPORT_EVERY

private static final int DEFAULT_REPORT_EVERY

See Also:

Constant Field Values

reportEvery

private int reportEvery

DEFAULT_INITIAL_STEP_LENGTH

private static final double DEFAULT_INITIAL_STEP_LENGTH

See Also:

Constant Field Values

DEFAULT_STEP_SIZE_REDUCTION

private static final double DEFAULT_STEP_SIZE_REDUCTION

See Also:

Constant Field Values

DEFAULT_STEP_SIZE_EXPENTION

private static final double DEFAULT_STEP_SIZE_EXPENTION

See Also:

Constant Field Values

initialStepLength

private double initialStepLength

stepSizeReduction

private double stepSizeReduction

stepSizeExpansion

private double stepSizeExpansion

Constructor Detail

SteepestDecent

public SteepestDecent(TotalEnergy energy,
                      double tolerance,
                      int maxSteps,
                      int reportEvery,
                      double initialStepLength,
                      double stepSizeReduction,
                      double stepSizeExpansion)

SteepestDecent

public SteepestDecent(TotalEnergy energy,
                      double tolerance,
                      int maxSteps,
                      int reportEvery)

SteepestDecent

public SteepestDecent(TotalEnergy energy)

Method Detail

init

protected void init()
             throws OptimizerException

Specified by:

init in class Minimizer

Throws:

OptimizerException

minimizationStep

protected boolean minimizationStep()
                            throws OptimizerException

Specified by:

minimizationStep in class Minimizer

Throws:

OptimizerException

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

lastStepLength

public double lastStepLength()

kickStart

protected void kickStart()
                  throws OptimizerException

Specified by:

kickStart in class Minimizer

Throws:

OptimizerException