[ VIGRA Homepage | Function Index | Class Index | Namespaces | File List | Main Page ]
NonlinearLSQOptions Class Reference |
Pass options to nonlinearLeastSquares(). More...
#include <vigra/regression.hxx>
Public Member Functions | |
NonlinearLSQOptions & | dampingParamters (double lambda, double v) |
Set damping parameters for Levenberg-Marquardt algorithm. More... | |
NonlinearLSQOptions & | maxIterations (int iter) |
Set maximum number of iterations. More... | |
NonlinearLSQOptions () | |
Initialize options with default values. | |
NonlinearLSQOptions & | tolerance (double eps) |
Set minimum relative improvement in residual. More... | |
Pass options to nonlinearLeastSquares().
#include <vigra/regression.hxx> Namespace: vigra
NonlinearLSQOptions& tolerance | ( | double | eps | ) |
Set minimum relative improvement in residual.
The algorithm stops when the relative improvement in residuals between consecutive iterations is less than this value.
Default: 0 (i.e. choose tolerance automatically, will be 10*epsilon of the numeric type)
NonlinearLSQOptions& maxIterations | ( | int | iter | ) |
Set maximum number of iterations.
Default: 50
NonlinearLSQOptions& dampingParamters | ( | double | lambda, |
double | v | ||
) |
Set damping parameters for Levenberg-Marquardt algorithm.
lambda determines by how much the diagonal is emphasized, and v is the factor by which lambda will be increased if more damping is needed for convergence (see Wikipedia for more explanations).
Default: lambda = 0.1, v = 1.4
© Ullrich Köthe (ullrich.koethe@iwr.uni-heidelberg.de) |
html generated using doxygen and Python
|