nuri::NelderMead Class Reference

Nelder-Mead simplex algorithm for function minimization. More...

#include <nuri/algo/optim.h>

Public Member Functions
	NelderMead (MutRef< ArrayXXd > data)
	Prepare Nelder-Mead simplex algorithm.
auto	n () const
int	argmax () const
auto	min () const
double	minf () const
auto	max ()
auto	max () const
double	maxf () const
double	max2f () const
template<class Func>
NMResult	minimize (Func f, int maxiter=-1, const double ftol=1e-6, const double alpha=1, const double gamma=2, const double rho=0.5, const double sigma=0.5)
	Minimize a function using Nelder-Mead simplex algorithm.

Static Public Member Functions
static constexpr int	argmin ()

Detailed Description

Nelder-Mead simplex algorithm for function minimization.

See also

nelder_mead

References:

• "Nelder-Mead method", Wikipedia (Accessed 2025-05-21).
• F Gao and L Han. Comput. Optim. Appl. 2012, 53 (1), 259-277. DOI:10.1007/s10589-010-9329-3

Constructor & Destructor Documentation

NelderMead()

nuri::NelderMead::NelderMead ( MutRef< ArrayXXd > data )

explicit

Prepare Nelder-Mead simplex algorithm.

Parameters

data

Initial simplex data. When the problem is N-dimensional, the data should be (N + 1, N + 1) matrix, where the first N rows form the N + 1 simplex vertices of dimension N, which will be modified in-place. The last row need not be initialized, but will be populated on-demand with the function value of each vertex.

Member Function Documentation

argmax()

int nuri::NelderMead::argmax ( ) const

inline

argmin()

constexpr int nuri::NelderMead::argmin ( )

inlinestaticconstexpr

max() [1/2]

auto nuri::NelderMead::max ( )

inline

max() [2/2]

auto nuri::NelderMead::max ( ) const

inline

max2f()

double nuri::NelderMead::max2f ( ) const

inline

maxf()

double nuri::NelderMead::maxf ( ) const

inline

min()

auto nuri::NelderMead::min ( ) const

inline

minf()

double nuri::NelderMead::minf ( ) const

inline

minimize()

template<class Func>

NMResult nuri::NelderMead::minimize ( Func f, int maxiter = -1, const double ftol = 1e-6, const double alpha = 1, const double gamma = 2, const double rho = 0.5, const double sigma = 0.5 )

inline

Minimize a function using Nelder-Mead simplex algorithm.

Template Parameters

Func	Function object that computes the function value. The function value should be returned.

Parameters

f	Function object.
maxiter	Maximum number of iterations. If negative or zero, it will be set to N * 200, where N is the number of dimensions.
ftol	Tolerance for the absolute change in the function value.
alpha	Reflection coefficient. Must be > 0.
gamma	Expansion coefficient. Must be > 1.
rho	Contraction coefficient. Must be in range (0, 1).
sigma	Shrink coefficient. Must be in range (0, 1).

Returns

The result of the optimization, including the result code and the index of the best vertex.

Note

The data matrix will be modified in-place, and the last row will be populated with the function values of each vertex.

n()

auto nuri::NelderMead::n ( ) const

inline

---

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

• include/nuri/algo/optim.h