sot::MeritWeightedDistance Class Reference

Merit function for choosing candidate points. More...

#include <merit_functions.h>

Inheritance diagram for sot::MeritWeightedDistance:

Public Member Functions
mat	pickPoints (const mat &cand, const std::shared_ptr< Surrogate > &surf, const mat &points, int newPoints, double distTol)
	Method for picking the next point. More...

Protected Attributes
vec	mWeights = {0.3, 0.5, 0.8, 0.95}
int	mNextWeight = 0

Detailed Description

Merit function for choosing candidate points.

This is the weighted distance merit function that chooses the next evaluation from a set of candidate point as the one that minimizes a criterion based on the Surrogate model prediction and the minimum distance to previously evaluated points.

Author

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

Member Function Documentation

mat sot::MeritWeightedDistance::pickPoints ( const mat &  cand, const std::shared_ptr< Surrogate > &  surf, const mat &  points, int  newPoints, double  distTol  )

inlinevirtual

Method for picking the next point.

Parameters

cand	The candidate points to choose from
surf	The surrogate model that predicts the value of the objective function
points	Previous evaluation that we compute the minimum distance to
newPoints	Number of candidate points that we are picking
distTol	Candidate points that are closer than distTol to a previously evaluated point are discarded

Returns

The proposed evaluations

Implements sot::MeritFunction.

Member Data Documentation

int sot::MeritWeightedDistance::mNextWeight = 0

protected

Next weight (modulo length of mWeights) to be used

vec sot::MeritWeightedDistance::mWeights = {0.3, 0.5, 0.8, 0.95}

protected

Weights that are cycled

---

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

• include/sot_bits/merit_functions.h