SQPstep.hpp File Reference

Provides the infrastructure for calculation of the RSVS equations. More...

#include <Eigen>
#include <fstream>
#include <iostream>
#include <vector>
#include "RSVScalc.hpp"
#include "arraystructures.hpp"
#include "mesh.hpp"
#include "snake.hpp"
#include "triangulate.hpp"
#include "warning.hpp"

Go to the source code of this file.

Functions
void	ResizeLagrangianMultiplier (const RSVScalc &calcobj, Eigen::VectorXd &lagMultAct, bool &isLarge, bool &isNan)
	Resizes the lagrangian multiplier LagMultAct based on whether any of its values are nan or too large. More...
template<class T >
bool	SQPstep (const RSVScalc &calcobj, const Eigen::MatrixXd &dConstrAct, const Eigen::RowVectorXd &dObjAct, const Eigen::VectorXd &constrAct, Eigen::VectorXd &lagMultAct, Eigen::VectorXd &deltaDVAct, bool &isNan, bool &isLarge, bool attemptConstrOnly=true)
	Template for calculation of an SQP step. More...
template<class T >
bool	SQPsens_sparse (const Eigen::MatrixXd &sensMult, const MatrixXd_sparse &sensInv, Eigen::MatrixXd &sensRes)
	Calculation for the sparse SQP sensitivity. More...
template<class T >
bool	SQPsens (const Eigen::MatrixXd &sensMult, const Eigen::MatrixXd &sensInv, Eigen::MatrixXd &sensRes)
template<class T >
bool	SQPstep_sparse (const RSVScalc &calcobj, const MatrixXd_sparse &dConstrAct, const Eigen::RowVectorXd &dObjAct, const Eigen::VectorXd &constrAct, Eigen::VectorXd &lagMultAct, Eigen::VectorXd &deltaDVAct, bool &isNan, bool &isLarge, bool attemptConstrOnly)

Detailed Description

Provides the infrastructure for calculation of the RSVS equations.

Definition in file SQPstep.hpp.

Function Documentation

ResizeLagrangianMultiplier()

void ResizeLagrangianMultiplier	(	const RSVScalc &	calcobj,
		Eigen::VectorXd &	lagMultAct,
		bool &	isLarge,
		bool &	isNan
	)

Resizes the lagrangian multiplier LagMultAct based on whether any of its values are nan or too large.

This uses the RSVScalc object to guide the resizing operation if it is needed.

Parameters

[in]	calcobj	The calculation object.
[in,out]	lagMultAct	The std::vector of active lagrangian multipliers.
[out]	isLarge	Returns if lagMultAct is too large.
[out]	isNan	Returns if lagMultAct has Nan values.

SQPsens_sparse()

template<class T >

bool SQPsens_sparse	(	const Eigen::MatrixXd &	sensMult,
		const MatrixXd_sparse &	sensInv,
		Eigen::MatrixXd &	sensRes
	)

Calculation for the sparse SQP sensitivity.

Parameters

	sensMult	The sensitivity multiplier.
[in]	sensInv	The Matrix to invert.
	sensRes	The sensitivity result.

Template Parameters

T	The Eigen object type to use. Should take a RSVScalc::HLag as a constructor and support a solve method.

Returns

true.

Definition at line 257 of file SQPstep.hpp.

SQPstep()

template<class T >

bool SQPstep	(	const RSVScalc &	calcobj,
		const Eigen::MatrixXd &	dConstrAct,
		const Eigen::RowVectorXd &	dObjAct,
		const Eigen::VectorXd &	constrAct,
		Eigen::VectorXd &	lagMultAct,
		Eigen::VectorXd &	deltaDVAct,
		bool &	isNan,
		bool &	isLarge,
		bool	attemptConstrOnly = true
	)

Template for calculation of an SQP step.

This template cannot be deduced and needs the developer to pass the required solver class when it is called.

Instantiation options: Eigen::HouseholderQR Eigen::ColPivHouseholderQR Eigen::LLT<Eigen::MatrixXd> (*) <- needs a full type to be defined (see below) Eigen::PartialPivLU

For stability info https://eigen.tuxfamily.org/dox/group__TutorialLinearAlgebra.html https://eigen.tuxfamily.org/dox/group__DenseDecompositionBenchmark.html

Parameters

[in]	calcobj	The calculation object
[in]	dConstrAct	Active constraint jacobian dh/dx
[in]	dObjAct	Active objective jacobian dJ/dx
[in]	constrAct	Active constraint vector
	lagMultAct	The active lagrangian multipliers
	deltaDVAct	The active SQP step to take
[out]	isNan	Indicates if lagMult is nan
[out]	isLarge	Indicates if lagMult is large
[in]	attemptConstrOnly	Should the step algorithm attempt using only the constraint to step.

Template Parameters

T	The Eigen object template type to use. A full type will be defined using T<Eigen::MatrixXd>.

Returns

(isLarge || isNan), if true some form of failure was detected.

This template cannot be deduced and needs the developer to pass the required solver class when it is called.

Instantiation options: Eigen::HouseholderQR<Eigen::MatrixXd> Eigen::ColPivHouseholderQR<Eigen::MatrixXd> Eigen::LLT<Eigen::MatrixXd> Eigen::PartialPivLU<Eigen::MatrixXd>

For stability info https://eigen.tuxfamily.org/dox/group__TutorialLinearAlgebra.html https://eigen.tuxfamily.org/dox/group__DenseDecompositionBenchmark.html

Parameters

[in]	calcobj	The calculation object
[in]	dConstrAct	Active constraint jacobian dh/dx
[in]	dObjAct	Active objective jacobian dJ/dx
[in]	constrAct	Active constraint vector
	lagMultAct	The active lagrangian multipliers
	deltaDVAct	The active SQP step to take
[out]	isNan	Indicates if lagMult is nan
[out]	isLarge	Indicates if lagMult is large
[in]	attemptConstrOnly	Should the step algorithm attempt using only the constraint to step.

Template Parameters

T	The Eigen object type to use. Should take a RSVScalc::HLag as a constructor and support a solve method.

Returns

(isLarge || isNan), if true some form of failure was detected.

Definition at line 142 of file SQPstep.hpp.