sot::DYCORS< MeritFunction > Class Template Reference

DYnamic COordinate search using Response Surface models. More...

#include <adaptive_sampling.h>

Inheritance diagram for sot::DYCORS< MeritFunction >:

Public Member Functions
	DYCORS (const std::shared_ptr< Problem > &data, const std::shared_ptr< Surrogate > &surf, int numCand, int budget)
	Constructor. More...
void	reset (int budget)
	Resets the object for a new budget (useful if a strategy restarts) More...
mat	makePoints (const vec &xBest, const mat &points, const vec &sigma, int newPoints)
	Proposes new evaluations. More...

Protected Attributes
std::shared_ptr< Problem >	mData
std::shared_ptr< Surrogate >	mSurf
int	mNumCand
int	mDim
vec	mxLow
vec	mxUp
double	mDistTol
int	mNumEvals = 0
int	mBudget
MeritFunction	mMerit

Detailed Description

template<class MeritFunction = MeritWeightedDistance>
class sot::DYCORS< MeritFunction >

DYnamic COordinate search using Response Surface models.

This is an implementation of the DYCORS method that perturbs fewer and fewer variables as the optimization proceeds. The candidate points are generated by perturbing each variable using the probability proposed by DYCORS.

Template Parameters

MeritFunction

The merit function is used to pick the most promising out of the generated candidate points.

Todo:

Should use SRBF as a Base class

Author

David Eriksson, dme65.nosp@m.@cor.nosp@m.nell..nosp@m.edu

Constructor & Destructor Documentation

template<class MeritFunction = MeritWeightedDistance>

sot::DYCORS< MeritFunction >::DYCORS ( const std::shared_ptr< Problem > &  data, const std::shared_ptr< Surrogate > &  surf, int  numCand, int  budget  )

inline

Constructor.

Parameters

data	A shared pointer to the optimization problem
surf	A shared pointer to the surrogate model
numCand	Number of candidate points that are generated in makePoints
budget	Evaluation budget for the adaptive sampling phase

Member Function Documentation

template<class MeritFunction = MeritWeightedDistance>

mat sot::DYCORS< MeritFunction >::makePoints ( const vec &  xBest, const mat &  points, const vec &  sigma, int  newPoints  )

inlinevirtual

Proposes new evaluations.

Parameters

xBest	The best solution found so far
points	Previously evaluated points
sigma	The sampling radius
newPoints	Number of new evaluations to be generated

Returns

The proposed points

Implements sot::Sampling.

template<class MeritFunction = MeritWeightedDistance>

void sot::DYCORS< MeritFunction >::reset ( int  budget )

inlinevirtual

Resets the object for a new budget (useful if a strategy restarts)

Parameters

budget

New evaluation budget

Implements sot::Sampling.

Member Data Documentation

template<class MeritFunction = MeritWeightedDistance>

int sot::DYCORS< MeritFunction >::mBudget

protected

Evaluation budget for the adaptive sampling phase

template<class MeritFunction = MeritWeightedDistance>

std::shared_ptr<Problem> sot::DYCORS< MeritFunction >::mData

protected

A shared pointer to the optimization problem

template<class MeritFunction = MeritWeightedDistance>

int sot::DYCORS< MeritFunction >::mDim

protected

Number of dimensions (extracted from mData)

template<class MeritFunction = MeritWeightedDistance>

double sot::DYCORS< MeritFunction >::mDistTol

protected

Distance tolerance

template<class MeritFunction = MeritWeightedDistance>

MeritFunction sot::DYCORS< MeritFunction >::mMerit

protected

Merit function that is used for picking candidate points

template<class MeritFunction = MeritWeightedDistance>

int sot::DYCORS< MeritFunction >::mNumCand

protected

Number of candidate points that are generated in makePoints

template<class MeritFunction = MeritWeightedDistance>

int sot::DYCORS< MeritFunction >::mNumEvals = 0

protected

Current evaluation count

template<class MeritFunction = MeritWeightedDistance>

std::shared_ptr<Surrogate> sot::DYCORS< MeritFunction >::mSurf

protected

A shared pointer to the surrogate model

template<class MeritFunction = MeritWeightedDistance>

vec sot::DYCORS< MeritFunction >::mxLow

protected

Lower variable bounds (extracted from mData)

template<class MeritFunction = MeritWeightedDistance>

vec sot::DYCORS< MeritFunction >::mxUp

protected

Upper variable bounds (extracted from mData)

---

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

• include/sot_bits/adaptive_sampling.h