DownhillSimplexMinimization.h File Reference

#include "Stroika/Foundation/StroikaPreComp.h"
#include <optional>
#include "Stroika/Foundation/Characters/String.h"
#include "Stroika/Foundation/Containers/Sequence.h"
#include "Stroika/Foundation/Traversal/Iterable.h"
#include "DownhillSimplexMinimization.inl"

Go to the source code of this file.

Classes
struct	Stroika::Foundation::Math::Optimization::DownhillSimplexMinimization::Options< FLOAT_TYPE >
struct	Stroika::Foundation::Math::Optimization::DownhillSimplexMinimization::Results< FLOAT_TYPE >

Namespaces
namespace	Stroika::Foundation

Functions
template<typename FLOAT_TYPE >
Results< FLOAT_TYPE >	Stroika::Foundation::Math::Optimization::DownhillSimplexMinimization::Run (const TargetFunction< FLOAT_TYPE > &function2Minimize, const Sequence< FLOAT_TYPE > &initialValues, const Options< FLOAT_TYPE > &options=Options< FLOAT_TYPE >{})
	Downhill Simplex Minimization, AKA Nelder-Mead algorithm, to compute minimization.

Detailed Description

Note

Code-Status: Beta

Definition in file DownhillSimplexMinimization.h.

Function Documentation

Run()

template<typename FLOAT_TYPE >

Results< FLOAT_TYPE > Stroika::Foundation::Math::Optimization::DownhillSimplexMinimization::Run	(	const TargetFunction< FLOAT_TYPE > &	function2Minimize,
		const Sequence< FLOAT_TYPE > &	initialValues,
		const Options< FLOAT_TYPE > &	options = Options<FLOAT_TYPE>{}
	)

Downhill Simplex Minimization, AKA Nelder-Mead algorithm, to compute minimization.

Take an argument function with argument 'initialValues' and try many values to find where the targetFunction is least (ie minimize it).

If the function is naturally periodic - you can add checks in the minimization function for the desired target range and produce LARGE answers there so they wont appear as minima of the 'targetFunction'.

Reference: https://en.wikipedia.org/wiki/Nelder%E2%80%93Mead_method
Reference: https://github.com/fchollet/nelder-mead/blob/master/nelder_mead.py
Reference: http://www.aip.de/groups/soe/local/numres/bookcpdf/c10-4.pdf
Reference: https://www.mathworks.com/matlabcentral/mlc-downloads/downloads/submissions/20398/versions/1/previews/fminsearch2.m/index.html

Note

The size of the Sequence<> in initialValues will be the same as the size of the Sequence passed to 'function2Minimize', which in turn will the the size of the Sequence<> in the Results.

Example Usage

TargetFunction<double> f = [](const Sequence<double>& x) {
    double d = x[0];
     if (d < 0 or d >= numbers::pi) {     // avoid falling off ends of ranges - periodic function
        return 100.0;
     }
     return -std::cos (d);
 };
 Results<double> result = Run (f, {.1});
 EXPECT_TRUE (Math::NearlyEquals (result.fOptimizedParameters[0], 0.0, 1e-10));

Definition at line 183 of file DownhillSimplexMinimization.inl.