Header-only Nelder–Mead simplex minimiser with templated cost and report functors. More...

#include <simplex_algorithm.h>

Static Public Member Functions
template<typename T, typename CostFunc, typename ReportFunc>
static bool	simplex_minimize (Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic > &p, Eigen::Matrix< T, Eigen::Dynamic, 1 > &y, T ftol, T stol, CostFunc &&func, int max_eval, int &neval, int report, ReportFunc &&report_func)
template<typename T, typename CostFunc>
static bool	simplex_minimize (Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic > &p, Eigen::Matrix< T, Eigen::Dynamic, 1 > &y, T ftol, T stol, CostFunc &&func, int max_eval, int &neval)

Protected Member Functions
	SimplexAlgorithm ()

Detailed Description

Header-only Nelder–Mead simplex minimiser with templated cost and report functors.

Header-only Nelder–Mead simplex minimiser templated on the scalar type and on the cost / report callables, implementing the Strategy pattern at zero runtime cost (the functors are inlined by the compiler).

Note: Implements the Strategy pattern — the cost function and report function are injected as callable template parameters (zero-overhead type erasure).

Definition at line 67 of file simplex_algorithm.h.

Constructor & Destructor Documentation

◆ SimplexAlgorithm()

UTILSLIB::SimplexAlgorithm::SimplexAlgorithm ( )

protected

Protected constructor to make class non-instantiable.

Member Function Documentation

◆ simplex_minimize() [1/2]

template<typename T, typename CostFunc>

bool UTILSLIB::SimplexAlgorithm::simplex_minimize	(	Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic > &	p,
		Eigen::Matrix< T, Eigen::Dynamic, 1 > &	y,
		T	ftol,
		T	stol,
		CostFunc &&	func,
		int	max_eval,
		int &	neval )

static

Overload without report function (no reporting).

Definition at line 178 of file simplex_algorithm.h.

◆ simplex_minimize() [2/2]

template<typename T, typename CostFunc, typename ReportFunc>

bool UTILSLIB::SimplexAlgorithm::simplex_minimize	(	Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic > &	p,
		Eigen::Matrix< T, Eigen::Dynamic, 1 > &	y,
		T	ftol,
		T	stol,
		CostFunc &&	func,
		int	max_eval,
		int &	neval,
		int	report,
		ReportFunc &&	report_func )

static

mne_simplex_fit.c Refactored: mne_simplex_minimize

Minimization with the simplex algorithm. Modified from Numerical Recipes. Supports an optional absolute spatial tolerance (stol) to detect simplex collapse.

Template Parameters

T	Scalar type (float or double).
CostFunc	Callable with signature T(const VectorX<T>&). Evaluates the cost at a simplex vertex.
ReportFunc	Callable with signature bool(int loop, const VectorX<T>& fitpar, double fval_lo, double fval_hi, double par_diff).

Parameters

[in,out]	p	The initial simplex (npar+1 x npar). On return, row 0 is the best vertex.
[in,out]	y	Function values at the vertices.
[in]	ftol	Relative convergence tolerance.
[in]	stol	Absolute spatial convergence tolerance (0 to disable).
[in]	func	The cost function to be evaluated.
[in]	max_eval	Maximum number of function evaluations.
[out]	neval	Number of function evaluations performed.
[in]	report	How often to report (-1 = no reporting).
[in]	report_func	The function to be called when reporting (may be nullptr).

Returns: True when minimization succeeded, false otherwise.

Definition at line 193 of file simplex_algorithm.h.

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

src/libraries/math/simplex_algorithm.h