Vector fitting of a rational transfer function to frequency data. More...

#include <Vectorfit.h>

Public Member Functions
	VectorFit ()
	Sets the defaults: Fitoption 2, RelaxedConstraint false, FlipUnstable true, PoleIdentification true, ResidueIdentification true, ComplexStateSpace false.

int	Fitoption () const
	The terms of the fit: 1 fits with D = 0 and E = 0, 2 fits with D free and E = 0 (default), 3 fits with D and E free.

void	Fitoption (int val)

bool	RelaxedConstraint () const
	Use the relaxed non-triviality constraint in the pole identification. Default false.

void	RelaxedConstraint (bool val)

bool	FlipUnstable () const
	Mirror poles with a positive real part to the left half plane. Default true.

void	FlipUnstable (bool val)

bool	PoleIdentification () const
	Do the pole identification. When false, that part of Fit() is skipped and the returned C, D and E are zero (see Fit() in Vectorfit.cpp). Default true.

void	PoleIdentification (bool val)

bool	ResidueIdentification () const
	Do the residue identification (C, D and E). When false, only the poles are identified. Default true.

void	ResidueIdentification (bool val)

bool	ComplexStateSpace () const
	Stored as an option; Fit() does not read it.

void	ComplexStateSpace (bool val)

mat	MatrixAreal () const

mat	MatrixBreal () const

mat	MatrixCreal () const

double	MatrixDreal () const

double	MatrixEreal () const

mat_complex	Diff () const
	Fit minus data at the points of the last Fit() (when the poles were identified).

double	Rmserr () const
	Root mean square of Diff().

void	Fit (const vec_complex &f, const vec_complex &s, vec_complex &poles, const vec &weight)
	One iteration of vector fitting.

vec_complex	LinSolveLS (mat &A, mat &B)

vec_complex	LinSolve (mat &A, mat &B)

vec_complex	MatrixPolynomial (const mat_complex &A)

vec_complex	EigenValues (const mat_complex &ZER)

void	QRFactor (mat &A_new, mat &Q, mat &R)

void	TransferFunction (vec_complex &num, vec_complex &den)

Result of the last Fit(): the model \f$ f(s) = C (sI - A)^{-1} B + D + E s \f$ with A diagonal (the poles) and B all ones.
mat_complex	MatrixA () const

void	MatrixA (mat_complex val)

mat_complex	MatrixB () const

void	MatrixB (mat_complex val)

mat_complex	MatrixC () const

void	MatrixC (mat_complex val)

std::complex< double >	MatrixD () const

void	MatrixD (std::complex< double > val)

std::complex< double >	MatrixE () const

void	MatrixE (std::complex< double > val)

Public Attributes
	EIGEN_MAKE_ALIGNED_OPERATOR_NEW

Detailed Description

Fits $ f(s) \approx \sum_i c_i/(s - a_i) + d + e s $ to complex data $ f $ given at the complex points $ s $, by relocating a set of start poles $ a_i $ (pole identification) and then finding the residues $ c_i $ and the terms $ d $ and $ e $ (residue identification). It is used by FreqDomainIdent to identify the radiation model from the frequency domain data of the hull. Set the options, call Fit() (several times to iterate the poles), and read the state-space model with MatrixA() to MatrixE() or the polynomial form with TransferFunction().

Note: The names vec, mat, vec_complex and mat_complex are typedefs of Eigen types in namespace Ship.

Member Function Documentation

◆ Fit()

void Ship::VectorFit::Fit	(	const vec_complex &	f,
		const vec_complex &	s,
		vec_complex &	poles,
		const vec &	weight
	)

Relocates the poles and then identifies the residues. Calling it again with the poles it returns iterates the fit. A start pole that is exactly 0 is replaced by -i/100 (i is its index).

Parameters

[in]	f	Complex data to fit, one value per point.
[in]	s	Complex Laplace variable at each point. FreqDomainIdent passes $ s = j\omega $ with $ \omega $ the frequency vector of the hull data, with no factor 2 pi: the poles and the model are in the unit of that vector. The tests (tests/Vectorfit_Test.cpp) use rad/s. The code does not check the unit.
[in,out]	poles	Start poles on input (complex pairs next to each other); the new poles on output.
[in]	weight	One weight per point, or one value for all points.

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

webfhsim/reloadrepos/fhsim_maritime/src/ship/Ship/PotentialData/Identification/Vectorfit.h

Public Member Functions

Public Attributes

Detailed Description

Member Function Documentation

◆ Fit()