SG++-Doxygen-Documentation
|
A coarsening functor, removing points according to the minimal absolute values in a DataVector provided, weighted with the corresponding basis function's surplus, i.e., with \(2^{-|\vec{l}|_1} = 2^{\sum_{k=1}^d l_d}\). More...
#include <SurplusVolumeCoarseningFunctor.hpp>
Public Member Functions | |
double | getCoarseningThreshold () const override |
Returns the threshold value. More... | |
size_t | getRemovementsNum () const override |
Returns the maximal number of points that should be removed. More... | |
double | operator() (GridStorage &storage, size_t seq) override |
This should be returning a coarsening value for every grid point. More... | |
double | start () const override |
This should return the initial value of coarsening criterion (e.g. More... | |
SurplusVolumeCoarseningFunctor (DataVector &alpha, size_t removements_num=1, double threshold=0.0) | |
Constructor. More... | |
~SurplusVolumeCoarseningFunctor () override | |
Destructor. More... | |
Public Member Functions inherited from sgpp::base::CoarseningFunctor | |
CoarseningFunctor () | |
Constructor. More... | |
virtual | ~CoarseningFunctor () |
Destructor. More... | |
Protected Attributes | |
DataVector & | alpha |
pointer to the vector that stores the alpha values More... | |
size_t | removements_num |
number of grid points to remove More... | |
double | threshold |
threshold, only the points with greater to equal absolute values of the refinement criterion (e.g. More... | |
Additional Inherited Members | |
Public Types inherited from sgpp::base::CoarseningFunctor | |
typedef double | value_type |
A coarsening functor, removing points according to the minimal absolute values in a DataVector provided, weighted with the corresponding basis function's surplus, i.e., with \(2^{-|\vec{l}|_1} = 2^{\sum_{k=1}^d l_d}\).
sgpp::base::SurplusVolumeCoarseningFunctor::SurplusVolumeCoarseningFunctor | ( | DataVector & | alpha, |
size_t | removements_num = 1 , |
||
double | threshold = 0.0 |
||
) |
Constructor.
alpha | DataVector that is basis for coarsening decisions. The i-th entry corresponds to the i-th grid point. |
removements_num | Number of grid points which should be removed (if possible - there could be less removable grid points) |
threshold | The absolute value of the entries have to be less or equal than the threshold to be considered for coarsening |
|
override |
Destructor.
|
overridevirtual |
Returns the threshold value.
Only the grid points with absolute value of coarsening criterion (e.g. alpha) less or equal to this threshold will be removed
Implements sgpp::base::CoarseningFunctor.
References threshold.
|
overridevirtual |
Returns the maximal number of points that should be removed.
The maximal number of points to removed is set in the constructor of implementation class.
Reimplemented from sgpp::base::CoarseningFunctor.
References removements_num.
|
overridevirtual |
This should be returning a coarsening value for every grid point.
The point with the lowest value will be removed first.
storage | reference to the grids storage object |
seq | sequence number in the coefficients array |
Implements sgpp::base::CoarseningFunctor.
References alpha, sgpp::base::HashGridPoint::getLevelSum(), sgpp::base::HashGridStorage::getPoint(), and sgpp::combigrid::pow().
|
overridevirtual |
This should return the initial value of coarsening criterion (e.g.
alpha or error).
Implements sgpp::base::CoarseningFunctor.
|
protected |
pointer to the vector that stores the alpha values
Referenced by python.learner.Learner.Learner::doLearningIteration(), python.uq.learner.Regressor.Regressor::learnData(), python.learner.Learner.Learner::learnData(), python.uq.learner.Interpolant.Interpolant::learnDataWithTest(), python.learner.Learner.Learner::learnDataWithTest(), operator()(), python.uq.dists.SGDEdist.SGDEdist::pdf(), python.learner.Classifier.Classifier::refineGrid(), and python.uq.dists.SGDEdist.SGDEdist::toJson().
|
protected |
number of grid points to remove
Referenced by getRemovementsNum().
|
protected |
threshold, only the points with greater to equal absolute values of the refinement criterion (e.g.
alpha or error) will be refined
Referenced by getCoarseningThreshold().