casacore::Fit2D Class Reference

Fit 2-D objects to 2-D Lattices or Arrays. More...

#include <Fit2D.h>

Public Types
enum	Types { GAUSSIAN, DISK, LEVEL, PLANE, nTypes }
	Enum describing the different models you can fit. More...
enum	ErrorTypes { OK, NOCONVERGE, FAILED, NOGOOD, NOMODELS, nErrorTypes }
	Enum describing output error conditions. More...
Public Member Functions
	Fit2D (LogIO &logger)
	Constructor.
	~Fit2D ()
	Destructor.
	Fit2D (const Fit2D &other)
	Copy constructor.
Fit2D &	operator= (const Fit2D &other)
	Assignment operator.
uInt	addModel (Fit2D::Types type, const Vector< Double > &parameters, const Vector< Bool > &parameterMask)
	Add a model to the list to be simultaneously fit and return its index.
uInt	addModel (Fit2D::Types type, const Vector< Double > &parameters)
void	setIncludeRange (Double minVal, Double maxVal)
	Set a pixel selection range.
void	setExcludeRange (Double minVal, Double maxVal)
void	resetRange ()
uInt	nModels () const
	Recover number of models.
Vector< Double >	estimate (Fit2D::Types type, const MaskedLattice< Float > &data)
	Determine an initial estimate for the solution of the specified model type to the given data - no compound models are allowable in this function.
Vector< Double >	estimate (Fit2D::Types type, const Lattice< Float > &data)
Vector< Double >	estimate (Fit2D::Types type, const Array< Float > &data)
Vector< Double >	estimate (Fit2D::Types type, const Array< Float > &data, const Array< Bool > &mask)
Fit2D::ErrorTypes	fit (const MaskedLattice< Float > &data, const Lattice< Float > &sigma)
	Do the fit.
Fit2D::ErrorTypes	fit (const Lattice< Float > &data, const Lattice< Float > &sigma)
Fit2D::ErrorTypes	fit (const Array< Float > &data, const Array< Float > &sigma)
Fit2D::ErrorTypes	fit (const Array< Float > &data, const Array< Bool > &mask, const Array< Float > &sigma)
Fit2D::ErrorTypes	residual (Array< Float > &resid, Array< Float > &model, const Array< Float > &data, Int xOffset=0, int yOffset=0) const
	Find the residuals to the fit.
Fit2D::ErrorTypes	residual (Array< Float > &resid, Array< Float > &model, const MaskedLattice< Float > &data)
Fit2D::ErrorTypes	residual (Array< Float > &resid, Array< Float > &model, const Lattice< Float > &data)
String	errorMessage () const
	If function fit failed, you will find a message here saying why it failed.
Vector< Double >	availableSolution () const
	Recover solution for either all model components or a specific one.
Vector< Double >	availableSolution (uInt which) const
Vector< Double >	availableErrors () const
	The errors.
Vector< Double >	availableErrors (uInt which) const
uInt	numberIterations () const
	The number of iterations that the fitter finished with.
Double	chiSquared () const
	The chi squared of the fit.
uInt	numberPoints () const
	The number of points used for the last fit.
Fit2D::Types	type (uInt which)
	Find type of specific model.
Static Public Member Functions
static Vector< Bool >	convertMask (const String fixedmask, Fit2D::Types type)
	Convert mask from a string to a vector.
static uInt	nParameters (Fit2D::Types type)
	Return number of parameters for this type of model.
static String	type (Fit2D::Types type)
	Return type as a string.
static Fit2D::Types	type (const String &type)
	Return string type as enum (min match).
static Double	paToGauss2D (Double pa)
	Convert p.a.
static Double	paFromGauss2D (Double pa)
	Convert p.a.
Private Member Functions
Fit2D::ErrorTypes	fitData (const Vector< Double > &values, const Matrix< Double > &pos, const Vector< Double > &sigma)
Vector< Double >	availableSolution (uInt &iStart, uInt which) const
	Returns available (adjustable + fixed) solution for model of interest and tells you where it began in the full solution vector Does no axial ratio nor position angle conversions from direct fit solution vector.
Vector< Double >	availableErrors (uInt &iStart, uInt which) const
Vector< Double >	getParams (uInt which) const
void	setParams (const Vector< Double > &params, uInt which)
Bool	includeIt (Float value, const Vector< Float > &range, Int includeIt) const
Bool	selectData (Matrix< Double > &pos, Vector< Double > &values, Vector< Double > &weights, const Array< Float > &pixels, const Array< Bool > &mask, const Array< Float > &sigma)
void	piRange (Double &pa) const
Private Attributes
LogIO	itsLogger
Bool	itsValid
Bool	itsValidSolution
Bool	itsHasSigma
Bool	itsInclude
Vector< Float >	itsPixelRange
CompoundFunction< AutoDiff < Double > >	itsFunction
NonLinearFitLM< Double >	itsFitter
Vector< Double >	itsSolution
Vector< Double >	itsErrors
Double	itsChiSquared
String	itsErrorMessage
uInt	itsNumberPoints
Vector< uInt >	itsTypeList

Detailed Description

Fit 2-D objects to 2-D Lattices or Arrays.

Intended use:

Public interface

Prerequisite

Lattice

Synopsis

This class allows you to fit different types of 2-D models to either Lattices or Arrays. These must be 2 dimensional; for Lattices, the appropriate 2-D Lattice can be made with the SubLattice class.

You may fit more than one model simultaneously to the data. Models are added with the addModel method. With this method, you also specify the initial guesses of the parameters of the model. Any parameters involving coordinates are expected in zero-relative absolute pixel coordinates (e.g. the centre of a model). Additionally with the addModel method, you may specify which parameters are to be held fixed during the fitting process. This is done with the parameterMask Vector which is in the same order as the parameter Vector. A value of True indicates the parameter will be fitted for. Presently, when you say fix the minor axis, you really end up fixing the axial ratio (internals). I don't have a solution for this presently.

For Gaussians, the parameter Vector (input or output) consists, in order, of the peak, x location, y location, FWHM of major axis, FWHM of minor axis, and position angle of the major axis (in radians). The position angle is positive +x to +y in the pixel coordinate system ([0,0] in center of image) and in the range -2pi to 2pi. When the solution is recovered, the position angle will be in the range 0 to pi.

Example

To Do

template it
Speed up some Array calculations indexed with IPositions
Don't handle Lattices simply by getting pixels into Arrays
make an addModel interface taking functionals

Definition at line 101 of file Fit2D.h.

Member Enumeration Documentation

enum casacore::Fit2D::ErrorTypes

Enum describing output error conditions.

Enumerator:

OK	ok
NOCONVERGE	Did not converge.
FAILED	Solution failed.
NOGOOD	There were no unmasked points.
NOMODELS	No models set.
nErrorTypes	Number of conditions.

Definition at line 115 of file Fit2D.h.

enum casacore::Fit2D::Types

Enum describing the different models you can fit.

Enumerator:

GAUSSIAN
DISK
LEVEL
PLANE
nTypes

Definition at line 106 of file Fit2D.h.

Constructor & Destructor Documentation

casacore::Fit2D::Fit2D ( LogIO & logger ) [explicit]

Constructor.

casacore::Fit2D::~Fit2D ( )

Destructor.

casacore::Fit2D::Fit2D ( const Fit2D & other )

Copy constructor.

Uses copy semantics except for the logger for which a reference copy is made

Member Function Documentation

uInt casacore::Fit2D::addModel	(	Fit2D::Types	type,
		const Vector< Double > &	parameters
	)

uInt casacore::Fit2D::addModel	(	Fit2D::Types	type,
		const Vector< Double > &	parameters,
		const Vector< Bool > &	parameterMask
	)

Add a model to the list to be simultaneously fit and return its index.

Specify the initial guesses for the model and a mask indicating whether the parameter is fixed (False) during the fit or not. Returns the the model number added (0, 1, 2 etc)

Vector<Double> casacore::Fit2D::availableErrors	(	uInt &	iStart,
		uInt	which
	)			const `[private]`

Vector<Double> casacore::Fit2D::availableErrors ( uInt which ) const

Vector<Double> casacore::Fit2D::availableErrors ( ) const

The errors.

All available parameters (fixed and adjustable) are included in the error vectors. Unsolved for parameters will have error 0.

Vector<Double> casacore::Fit2D::availableSolution	(	uInt &	iStart,
		uInt	which
	)			const `[private]`

Returns available (adjustable + fixed) solution for model of interest and tells you where it began in the full solution vector Does no axial ratio nor position angle conversions from direct fit solution vector.

Vector<Double> casacore::Fit2D::availableSolution ( uInt which ) const

Vector<Double> casacore::Fit2D::availableSolution ( ) const

Recover solution for either all model components or a specific one.

These functions will return an empty vector if there is no valid solution. All available parameters (fixed and adjustable) are included in the solution vectors.

Double casacore::Fit2D::chiSquared ( ) const

The chi squared of the fit.

Returns 0 if fit has been done.

static Vector<Bool> casacore::Fit2D::convertMask	(	const String	fixedmask,
		Fit2D::Types	type
	)			`[static]`

Convert mask from a string to a vector.

The string gives the parameters to keep fixed in the fit (f (flux), x (x position), y (y position), a (FWHM major axis), b (FWHM minor axis), p (position angle)

String casacore::Fit2D::errorMessage ( ) const

If function fit failed, you will find a message here saying why it failed.

Vector<Double> casacore::Fit2D::estimate	(	Fit2D::Types	type,
		const Array< Float > &	data,
		const Array< Bool > &	mask
	)

Vector<Double> casacore::Fit2D::estimate	(	Fit2D::Types	type,
		const Array< Float > &	data
	)

Vector<Double> casacore::Fit2D::estimate	(	Fit2D::Types	type,
		const Lattice< Float > &	data
	)

Vector<Double> casacore::Fit2D::estimate	(	Fit2D::Types	type,
		const MaskedLattice< Float > &	data
	)

Determine an initial estimate for the solution of the specified model type to the given data - no compound models are allowable in this function.

If you have specified an include or exclude pixel range to the fitter, that will be honoured. This function does not interact with the addModel function. Returns a zero length vector if it fails to make an estimate.

Fit2D::ErrorTypes casacore::Fit2D::fit	(	const Array< Float > &	data,
		const Array< Bool > &	mask,
		const Array< Float > &	sigma
	)

Fit2D::ErrorTypes casacore::Fit2D::fit	(	const Array< Float > &	data,
		const Array< Float > &	sigma
	)

Fit2D::ErrorTypes casacore::Fit2D::fit	(	const Lattice< Float > &	data,
		const Lattice< Float > &	sigma
	)

Fit2D::ErrorTypes casacore::Fit2D::fit	(	const MaskedLattice< Float > &	data,
		const Lattice< Float > &	sigma
	)

Do the fit.

Returns an enum value to tell you what happened if the fit failed for some reasons. A message can also be found with function errorMessage if the fit was not successful. For Array(i,j) i is x and j is y

Fit2D::ErrorTypes casacore::Fit2D::fitData	(	const Vector< Double > &	values,
		const Matrix< Double > &	pos,
		const Vector< Double > &	sigma
	)			`[private]`

Vector<Double> casacore::Fit2D::getParams ( uInt which ) const [private]

Bool casacore::Fit2D::includeIt	(	Float	value,
		const Vector< Float > &	range,
		Int	includeIt
	)			const `[inline, private]`

Definition at line 315 of file Fit2D.h.

References casacore::False, and casacore::True.

uInt casacore::Fit2D::nModels ( ) const

Recover number of models.

static uInt casacore::Fit2D::nParameters ( Fit2D::Types type ) [static]

Return number of parameters for this type of model.

uInt casacore::Fit2D::numberIterations ( ) const

The number of iterations that the fitter finished with.

uInt casacore::Fit2D::numberPoints ( ) const

The number of points used for the last fit.

Fit2D& casacore::Fit2D::operator= ( const Fit2D & other )

Assignment operator.

Uses copy semantics except for the logger for which a reference copy is made

static Double casacore::Fit2D::paFromGauss2D ( Double pa ) [inline, static]

Convert p.a.

(radians) from positive +y -> -x (Gaussian2D) to positive +x -> +y (Fit2D)

Definition at line 271 of file Fit2D.h.

References casacore::C::pi_2.

static Double casacore::Fit2D::paToGauss2D ( Double pa ) [inline, static]

Convert p.a.

(radians) from positive +x -> +y (Fit2D) to positive +y -> -x (Gaussian2D)

Definition at line 267 of file Fit2D.h.

References casacore::C::pi_2.

void casacore::Fit2D::piRange ( Double & pa ) const [private]

void casacore::Fit2D::resetRange ( )

Fit2D::ErrorTypes casacore::Fit2D::residual	(	Array< Float > &	resid,
		Array< Float > &	model,
		const Lattice< Float > &	data
	)

Fit2D::ErrorTypes casacore::Fit2D::residual	(	Array< Float > &	resid,
		Array< Float > &	model,
		const MaskedLattice< Float > &	data
	)

Fit2D::ErrorTypes casacore::Fit2D::residual	(	Array< Float > &	resid,
		Array< Float > &	model,
		const Array< Float > &	data,
		Int	xOffset = `0`,
		int	yOffset = `0`
	)			const

Find the residuals to the fit.

xOffset and yOffset allow one to provide a data array that is offset in space from the grid that was fit. In this way, one can fill out a larger image than the subimage that was fit, for example. A negative value of xOffset means the supplied data array represents a grid that has a y axis left of the grid of pixels that was fit. A negative yOffset value means the supplied data array represents a grid that has an x axis that is below the x axis of the grid of pixels that was fit.

Bool casacore::Fit2D::selectData	(	Matrix< Double > &	pos,
		Vector< Double > &	values,
		Vector< Double > &	weights,
		const Array< Float > &	pixels,
		const Array< Bool > &	mask,
		const Array< Float > &	sigma
	)			`[private]`