xerus::ADFVariant Class Reference

Wrapper class for all ADF variants. More...

#include <adf.h>

Classes
class	InternalSolver

Public Member Functions
	ADFVariant (const size_t _maxIteration, const double _targetResidual, const double _minimalResidualDecrease)
	fully defining constructor. alternatively ALSVariants can be created by copying a predefined variant and modifying it More...
template<class MeasurmentSet >
double	operator() (TTTensor &_x, const MeasurmentSet &_measurments, PerformanceData &_perfData) const
	Tries to reconstruct the (low rank) tensor _x from the given measurments. More...
template<class MeasurmentSet >
double	operator() (TTTensor &_x, const MeasurmentSet &_measurments, const std::vector< size_t > &_maxRanks, PerformanceData &_perfData) const
	Tries to reconstruct the (low rank) tensor _x from the given measurments. More...

Public Attributes
size_t	maxIterations
	Maximum number of sweeps to perform. Set to 0 for infinite. More...
double	minimalResidualNormDecrease
double	targetResidualNorm
	Target residual. The algorithm will stop upon reaching a residual smaller than this value. More...

Detailed Description

Wrapper class for all ADF variants.

By creating a new object of this class and modifying the member variables, the behaviour of the solver can be modified. This algorithm is a modified implementation of the alternating directional fitting algrothim, first introduced by Grasedyck, Kluge and Kraemer (2015).

Definition at line 39 of file adf.h.

Constructor & Destructor Documentation

ADFVariant()

xerus::ADFVariant::ADFVariant ( const size_t  _maxIteration, const double  _targetResidual, const double  _minimalResidualDecrease  )

inline

fully defining constructor. alternatively ALSVariants can be created by copying a predefined variant and modifying it

Definition at line 238 of file adf.h.

Member Function Documentation

operator()() [1/2]

template<class MeasurmentSet >

double xerus::ADFVariant::operator() ( TTTensor &  _x, const MeasurmentSet &  _measurments, PerformanceData &  _perfData  ) const

inline

Tries to reconstruct the (low rank) tensor _x from the given measurments.

Parameters

[in,out]	_x	On input: an initial guess of the solution, also defining the ranks. On output: The reconstruction found by the algorithm.
	_measurments	the available measurments, can be either a SinglePointMeasurementSet or RankOneMeasurementSet.
	_perfData	optinal performanceData object to be used.

Returns

the residual \(|P_\Omega(x-b)|_2\) of the final _x.

Definition at line 249 of file adf.h.

operator()() [2/2]

template<class MeasurmentSet >

double xerus::ADFVariant::operator() ( TTTensor &  _x, const MeasurmentSet &  _measurments, const std::vector< size_t > &  _maxRanks, PerformanceData &  _perfData  ) const

inline

Tries to reconstruct the (low rank) tensor _x from the given measurments.

Parameters

[in,out]	_x	On input: an initial guess of the solution, may be of smaller rank. On output: The reconstruction found by the algorithm.
	_measurments	the available measurments, can be either a SinglePointMeasurementSet or RankOneMeasurementSet.
	_maxRanks	the maximal ranks the algorithm may use to decrease the resdiual.
	_perfData	optinal performanceData object to be used.

Returns

the residual \(|P_\Omega(x-b)|_2\) of the final _x.

Definition at line 263 of file adf.h.

Member Data Documentation

maxIterations

size_t xerus::ADFVariant::maxIterations

Maximum number of sweeps to perform. Set to 0 for infinite.

Definition at line 233 of file adf.h.

minimalResidualNormDecrease

double xerus::ADFVariant::minimalResidualNormDecrease

Definition at line 235 of file adf.h.

targetResidualNorm

double xerus::ADFVariant::targetResidualNorm

Target residual. The algorithm will stop upon reaching a residual smaller than this value.

Definition at line 234 of file adf.h.

---

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

• /home/gitlab-runner/builds/9071116c/0/xerus/xerus/include/xerus/algorithms/adf.h