TRTK::Optimization::Options< ValueType > Struct Template Reference

Structure used by several functions to control their operation. More...

#include <Optimization.hpp>

List of all members.

Public Types
enum	Algorithm { LEVENBERG_MARQUARDT, NEWTON_RAPHSON }
typedef ValueType	value_type
Public Attributes
value_type	error_tolerance
	This value is used by some algorithms as an abortion criterion.
value_type	lambda
	Damping parameter used by the Levenberg-Marquardt algorithm.
unsigned	max_number_iterations
	Maximum number of iterations performed during an iteration.
value_type	nu
	Damping parameter used by the Levenberg-Marquardt algorithm.
value_type	spacing
	Spacing used while computing a finit difference (approximation of the derivative).
enum TRTK::Optimization::Options::Algorithm	algorithm
	Algorithm used for the computations.

---

Detailed Description

template<class ValueType>
struct TRTK::Optimization::Options< ValueType >

Structure used by several functions to control their operation.

Default values:

• error_tolerance: 1e-15
• lambda: 1e-2
• max_number_iterations: 25
• nu: 10
• spacing: 1e-6
• algorithm: LEVENBERG_MARQUARDT

Note:

In order to obtain small rounding errors while computing finit differences (approximations of the derivatives) you might want to set the spacing to \( \sqrt{\epsilon}x \) where the machine epsilon \( \epsilon \) is typically of the order 2.2e-16.

Note:

Not all fields are necessarily used by all functions.

Definition at line 237 of file Optimization.hpp.

---

Member Enumeration Documentation

template<class ValueType >

enum TRTK::Optimization::Options::Algorithm

Enumerator:

LEVENBERG_MARQUARDT	Levenberg-Marquardt Algorithm.
NEWTON_RAPHSON	Newton-Raphson Method.

Definition at line 247 of file Optimization.hpp.

---

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

• D:/Programmierung/C/TRTK/include/TRTK/Optimization.hpp