#include <Nonlinreg.hh>
Inheritance diagram for RazorBack::Nonlinreg_:
Public Types | |
enum | Verbosity_ { SILENT, TERSE, VERBOSE } |
Verbosity_: controls std output during the iteration. More... | |
Public Methods | |
Nonlinreg_ (size_t NX, size_t NY, size_t NP, size_t NM) | |
Init to desired sizes if known in advance. More... | |
void | report_level (Verbosity_ V) |
report_level(V): sets the verbosity level of the progress report. | |
double | nonlin_reg (const Rmatdbl_ &Xmeas, const Rmatdbl_ &Ymeas, const Rmatdbl_ &Userwgt, const Vecdbl_ &Initpar, Nonlinfunc_ &Funcobj, int &Itmax, double Steplim=1e-6) |
nonlin_reg(Xmeas, Ymeas [,Userwgt], Initpar, Funcobj, Itmax, Steplim): Least-squares multiple nonlinear regression for the function Y=F(X, P) where Y is a Ny-dim. More... | |
double | nonlin_reg (const Rmatdbl_ &Xmeas, const Rmatdbl_ &Ymeas, const Vecdbl_ &Initpar, Nonlinfunc_ &Funcobj, int &Itmax, double Steplim=1e-6) |
Protected Methods | |
void | adjust_dimensions (const Rmatdbl_ &Xmeas, const Rmatdbl_ &Ymeas, const Rmatdbl_ &Userwgt, const Vecdbl_ &Initpar, const Nonlinfunc_ &Funcobj) |
void | setup_weight (const Rmatdbl_ &YW) |
double | quality (const Rmatdbl_ &Xmeas, const Rmatdbl_ &Ymeas, const Vecdbl_ &Par, Nonlinfunc_ &Funcobj) const |
void | report (ostream &Out, int Itno, double Q, double Lambda, const Vecdbl_ &Par) const |
Protected Attributes | |
size_t | Nx |
size_t | Ny |
X and Y vector sizes. | |
Rmatdbl_ | W |
weight matrix Nm x Ny. | |
Verbosity_ | Verb |
progress report verbosity. |
Definition at line 85 of file Nonlinreg.hh.
|
Verbosity_: controls std output during the iteration.
Definition at line 91 of file Nonlinreg.hh. |
|
Init to desired sizes if known in advance. Actual regression calls adjust sizes automatically. Definition at line 107 of file Nonlinreg.hh. |
|
report_level(V): sets the verbosity level of the progress report.
Definition at line 111 of file Nonlinreg.hh. |
|
nonlin_reg(Xmeas, Ymeas [,Userwgt], Initpar, Funcobj, Itmax, Steplim): Least-squares multiple nonlinear regression for the function Y=F(X, P) where Y is a Ny-dim. vector, X is a Nx-dim vector, and P is a Np-dim. vector of parameters, F a nonlinear vector-vector function. Routine parameters:- Xmeas: Nm x Nx matrix of independent variables Ymeas: Nm x Ny matrix of dependent variables Userwgt: optional Nm x Ny matrix of weights can be omitted if equal or relative weights are used Initpar: Np-dim. vector of the initial parameter estimates Funcobj: a nonlinear function object (instance of class Nonlinfunc_) that can calculate F and the Jacobian dF/dP Itmax: on input, the maximal number of iterations, on output, the actual number of iterations Steplim: minimal stepsize (default 1e-6). Return value: the residual deviation. |
|
|
|
|
|
|
|
|
|
|
|
Definition at line 96 of file Nonlinreg.hh. |
|
X and Y vector sizes.
Definition at line 96 of file Nonlinreg.hh. |
|
weight matrix Nm x Ny.
Definition at line 97 of file Nonlinreg.hh. |
|
progress report verbosity.
Definition at line 98 of file Nonlinreg.hh. |