fitter - Constructor

Construct fitting tool

Synopsis
fitter(n, type, colfac, lmfac, host, forcenewserver)

Arguments

n	in	number of unknowns
		Allowed:	integer
		Default:	0
type	in	type of solution
		Allowed:	coded value as string. Allowed: real, complex, separable, asreal, conjugate
		Default:	'real'
colfac	in	collinearity factor
		Allowed:	positive float
		Default:	1e-8
lmfac	in	Levenberg-Marquardt factor
		Allowed:	positive float
		Default:	1e-3
host	in	name of host to run on
		Allowed:	string
		Default:	''
forcenewserver	in	is a new process needed
		Allowed:	bool
		Default:	F

Returns
fitting tool or fail

Description
Create a fitting tool. The tool can be created without arguments (in which case it is assumed to be a real fitter), or with the arguments specifying the number of unknowns to be solved for (a number not relevant in practice); and the type of solution: real, complex, conjugate (complex with both the unknown and its conjugate in the condition equations), separable complex, asreal complex with the real and imaginary part seen as independent unknowns. All solutions need a model (specified as a functional. All solutions are done using an SVD type method. A collinearity factor can be specified, which is in essence the sine squared of the minimum angle between two normal equation columns that are still to be considered independent. For automatic non-linear solutions, a Levenberg-Marquardt factor (see Note 224) is used, which can be specified as well.

In the case of non-linear solutions that have to be handled by the system, an initial estimate for the model parameters is necessary.

A default fitter is always created, with an id that does not have to be specified. (see fitter method for sub constructors).

Next: fitter.id - Function Up: fitter - Tool Previous: fitter - Tool Contents Index

	Package	utility
	Module	fitting
	Tool	fitter