Package gov.nih.mipav.model.algorithms

Class FitGaussian

java.lang.Object

gov.nih.mipav.model.algorithms.NLConstrainedEngine

gov.nih.mipav.model.algorithms.NLFittedFunction

gov.nih.mipav.model.algorithms.FitGaussian

public class FitGaussian
extends NLFittedFunction

FitGaussian -fits an array of points to a normal curve, general from f = a*exp(-(x-b)^2/2sigma^2) Will also perform thresholding techniques to determine useful data points for fitting

Version:

0.1

Author:

senseneyj

See Also:

NLConstrainedEngine

Field Summary

Fields

Modifier and Type	Field	Description
private double	amp	Amplitude parameter
private int	dataEnd	Location in xSeries where Gaussian data ends
private int	dataStart	Location in xSeries where Gaussian data starts
private int	iters
private double	rSquared	R squared
private double	sigma	Sigma parameter
private double	xInit	Center parameter

Fields inherited from class gov.nih.mipav.model.algorithms.NLFittedFunction

chisq, EPSILON, MAX_ITR, MIN_ITR, xSeries, yDataFitted, ySeries

Fields inherited from class gov.nih.mipav.model.algorithms.NLConstrainedEngine

a, absoluteConvergence, analyticalJacobian, bl, bounds, bu, ctrlMat, dyda, gues, internalScaling, jacobian, maxIterations, nPts, outputMes, param, parameterConvergence, relativeConvergence, residuals, secondAllowed, stdv, tolerance

Constructor Summary

Constructors

Constructor	Description
FitGaussian()	FitGaussian - test constructor, builds and test function.
FitGaussian(int nPoints, double[] xData, double[] yData)	FitGaussian.
FitGaussian(int nPoints, float[] xData, float[] yData)	Constructs new fit gaussian.

Method Summary

Modifier and Type	Method	Description
private double[]	applyKernel()	Apply small kernel to smooth out data.
protected void	calculateChiSq()	Calculates chi squared
protected void	calculateFittedY()	Calculates yDataFitted
private double	dgdA(double x)	Partial derivative of gaussian with respect to A.
private double	dgdsigma(double x)	Partial derivative of gaussian with respect to sigma.
private double	dgdx(double x)	Partial derivative of gaussian with respect to x.
void	displayResults()	Display results of displaying exponential fitting parameters.
void	driver()	Starts the analysis.
private void	estimateInitial()
void	fitToFunction(double[] a, double[] residuals, double[][] covarMat)	fitToFunction communicates with 3 protected variables param, nPts, and ctrlMat ctrlMat is used as a wrapper for ctrl or lctrl.
private double	gauss(double x)	Gaussian evaluated at a point with given parameters
protected Jama.Matrix	generateJacobian()	Jacobian used for non-linear least squares fitting.
protected Jama.Matrix	generateResiduals()	Calculates the residuals for a given function, not implemented since some functions might prefer to only use a subset of data points, or not use yDataFitted if working during an iteration
double	getRSquared()
private void	testData()	Test data to test fitting of gaussian.

Methods inherited from class gov.nih.mipav.model.algorithms.NLFittedFunction

getChisq, getFittedY, getMedian

Methods inherited from class gov.nih.mipav.model.algorithms.NLConstrainedEngine

dumpTestResults, fitToTestFunction, getChiSquared, getExitStatus, getIterations, getParameters, getResiduals, statusMessage

Methods inherited from class java.lang.Object

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

Field Detail

dataStart

private int dataStart

Location in xSeries where Gaussian data starts

dataEnd

private int dataEnd

Location in xSeries where Gaussian data ends

amp

private double amp

Amplitude parameter

xInit

private double xInit

Center parameter

sigma

private double sigma

Sigma parameter

rSquared

private double rSquared

R squared

iters

private int iters

Constructor Detail

FitGaussian

public FitGaussian()

FitGaussian - test constructor, builds and test function.

FitGaussian

public FitGaussian(int nPoints,
                   double[] xData,
                   double[] yData)

FitGaussian.

Parameters:

nPoints - number of points in the function

xData - DOCUMENT ME!

yData - DOCUMENT ME!

FitGaussian

public FitGaussian(int nPoints,
                   float[] xData,
                   float[] yData)

Constructs new fit gaussian.

Parameters:

nPoints - Number of points in the function

xData - DOCUMENT ME!

yData - DOCUMENT ME!

Method Detail

applyKernel

private double[] applyKernel()

Apply small kernel to smooth out data. Often helpful when too many bins have been applied by user.

estimateInitial

private void estimateInitial()

driver

public void driver()

Starts the analysis. For some reason a guess with the wrong sign for a2 will not converge. Therefore, try both sign and take the one with the lowest chi-squared value.

Overrides:

driver in class NLConstrainedEngine

calculateChiSq

protected void calculateChiSq()

Description copied from class: NLFittedFunction

Calculates chi squared

Specified by:

calculateChiSq in class NLFittedFunction

calculateFittedY

protected void calculateFittedY()

Description copied from class: NLFittedFunction

Calculates yDataFitted

Specified by:

calculateFittedY in class NLFittedFunction

displayResults

public void displayResults()

Display results of displaying exponential fitting parameters.

Specified by:

displayResults in class NLFittedFunction

getRSquared

public double getRSquared()

fitToFunction

public void fitToFunction(double[] a,
                          double[] residuals,
                          double[][] covarMat)

Description copied from class: NLConstrainedEngine

fitToFunction communicates with 3 protected variables param, nPts, and ctrlMat ctrlMat is used as a wrapper for ctrl or lctrl. Evaluates the residuals or the Jacobian at a certain a[]

Specified by:

fitToFunction in class NLConstrainedEngine

Parameters:

a - DOCUMENT ME!

residuals - DOCUMENT ME!

covarMat - DOCUMENT ME!

testData

private void testData()

Test data to test fitting of gaussian.

gauss

private double gauss(double x)

Gaussian evaluated at a point with given parameters

dgdA

private double dgdA(double x)

Partial derivative of gaussian with respect to A.

dgdx

private double dgdx(double x)

Partial derivative of gaussian with respect to x.

dgdsigma

private double dgdsigma(double x)

Partial derivative of gaussian with respect to sigma.

generateJacobian

protected Jama.Matrix generateJacobian()

Jacobian used for non-linear least squares fitting.

generateResiduals

protected Jama.Matrix generateResiduals()

Description copied from class: NLFittedFunction

Calculates the residuals for a given function, not implemented since some functions might prefer to only use a subset of data points, or not use yDataFitted if working during an iteration

Specified by:

generateResiduals in class NLFittedFunction

Returns: