---

rdFSQP Class Reference

This class provides methods for finding the optimal controls of a redundant system by applying sequential quadratic programming techniques. More...

#include <rdFSQP.h>

List of all members.

Public Member Functions
virtual	~rdFSQP ()
	Destructor.
	rdFSQP (rdOptimizationTarget *aTarget)
	Construct an rdFSQP instance based on an optimization target.
rdOptimizationTarget *	setTarget (rdOptimizationTarget *aTarget)
	Set the optimization target.
rdOptimizationTarget *	getTarget ()
	Get the current optimization target.
void	setMode (int aMode)
	Set the mode of fsqp.
int	getMode ()
	Get the mode of fsqp.
void	setPrintLevel (int aLevel)
	Set the print level.
int	getPrintLevel ()
	Get the print level.
void	setMaxIterations (int aMaxIter)
	Set the maximum number of iterations.
int	getMaxIterations ()
	Get the maximum number of iterations.
void	setInfinity (double aInfinity)
	Set the value considered to be infinity.
double	getInfinity ()
	Set the value considered to be infinity.
void	setConvergenceCriterion (double aEPS)
	Set the convergence criterion.
double	getConvergenceCriterion ()
	Get the convergence criterion.
void	setNonlinearEqualityConstraintTolerance (double aEPSEQN)
	Set the tolerance allowed in meeting nonlinear equality constraints.
double	getNonlinearEqualityConstraintTolerance ()
	Get the tolerance allowed in meeting nonlinear equality constraints.
void	setLowerBound (double aLower)
	Set a lower bound on the controls.
void	setLowerBound (int aIndex, double aLower)
	Set the lower bound for a specified control.
void	setLowerBound (double aLower[])
	Set lower bounds on the controls.
double *	getLowerBound ()
	Get lower bounds on the controls.
void	setUpperBound (double aUpper)
	Set a upper bound on the controls.
void	setUpperBound (int aIndex, double aUpper)
	Set the upper bound for a specified control.
void	setUpperBound (double aUpper[])
	Set upper bounds on the controls.
double *	getUpperBound ()
	Get upper bounds on the controls.
int	computeOptimalControls (const double *xstart, double *x)
	Compute a set of optimal controls, given the input controls (xin) and the current state of the optimization target.
Static Public Member Functions
int	CentralDifferences (rdOptimizationTarget *aTarget, double *dx, double *x, double *dpdx, double *dcdx)
	Compute derivatives of performance and constraints with respect to the controls by central differences.
int	CentralDifferences (rdOptimizationTarget *aTarget, double *dx, double *x, double *dpdx)
	Compute derivatives of performance with respect to the controls by central differences.
int	CentralDifferencesConstraint (rdOptimizationTarget *aTarget, double *dx, double *x, int ic, double *dcdx)
	Compute derivatives of a constraint with respect to the controls by central differences.
void	pFunc (int nparam, int j, double *x, double *pj, void *cd)
	Compute the performance criterion.
void	cFunc (int nparam, int j, double *x, double *cj, void *cd)
	Compute the constraints.
void	dpdxFunc (int nparam, int j, double *x, double *dpdx, void(*dummy)(int, int, double *, double *, void *), void *cd)
	Compute the derivatives of the performance criterion.
void	dcdxFunc (int nparam, int j, double *x, double *dcdx, void(*dummy)(int, int, double *, double *, void *), void *cd)
	Compute the gradient of the constraints.
Private Member Functions
void	setNull ()
	Set NULL values for the member variables.
Private Attributes
rdOptimizationTarget *	_target
int	_mode
	Mode.
int	_printLevel
	Print level.
int	_maxIter
	Maximum number of iterations.
int	_inform
	Variable that contains information about the status of an optimization.
double	_infinity
	Value used for infinity or a very large number.
double	_eps
	Convergence criterion.
double	_epseqn
	Convergence criterion for nonlinear equality constraints.
double	_udelta
	I don't know.
int	_ncsrl
int	_ncsrn
int	_nfsr
int *	_mesh
double *	_bl
	Lower bounds on controls.
double *	_bu
	Upper bounds on controls.
double *	_x
	Controls.
double *	_p
	Array of performance criteria.
double *	_c
	Array of constraints.
double *	_lambda
	Lagrange multipliers for constraints.

---

Detailed Description

This class provides methods for finding the optimal controls of a redundant system by applying sequential quadratic programming techniques.

The core algorithm is called fsqp, "Fast Sequential Quadratic Programming".

---

Member Function Documentation

int rdFSQP::CentralDifferences (  rdOptimizationTarget *  aTarget, double *  dx, double *  x, double *  dpdx )  [static]

Compute derivatives of performance with respect to the controls by central differences. Note that the gradient array should be allocated as dpdx[nx]. Parameters: dx  An array of control perturbation values. x  Values of the controls at time t. dpdx  The derivatives of the performance criterion. Returns: -1 if an error is encountered, 0 otherwize.

int rdFSQP::CentralDifferences (  rdOptimizationTarget *  aTarget, double *  dx, double *  x, double *  dpdx, double *  dcdx )  [static]

Compute derivatives of performance and constraints with respect to the controls by central differences. Note that the gradient arrays should be allocated as dpdx[nx] and dcdx[nc][nx]. Parameters: dx  An array of control perturbation values. x  Values of the controls at time t. dpdx  The derivatives of the performance criterion. dcdx  The derivatives of the constraints. Returns: -1 if an error is encountered, 0 otherwize.

int rdFSQP::CentralDifferencesConstraint (  rdOptimizationTarget *  aTarget, double *  dx, double *  x, int  ic, double *  dcdx )  [static]

Compute derivatives of a constraint with respect to the controls by central differences. Parameters: dx  An array of control perturbation values. x  Values of the controls at time t. ic  Index of the constraint. dcdx  The derivatives of the constraints. Returns: -1 if an error is encountered, 0 otherwize.

int rdFSQP::computeOptimalControls (  const double *  xstart, double *  x )

Compute a set of optimal controls, given the input controls (xin) and the current state of the optimization target. Whether or not the optimization terminates normally, the latest value of the controls are copied to xout. It is safe to use the same pointer for xin and xout. However, in all cases the calling routine must allocate enough space for xin and xout. Parameters: xin  Values of the controls. xout  Optimal values of the controls. Returns: Parameter inform of cfsqp. -1 means a fatal error.

double * rdFSQP::getLowerBound (   )

Get lower bounds on the controls. Note that the address is returned and not a copy; Returns: Pointer the the array of lower bounds.

rdOptimizationTarget * rdFSQP::getTarget (   )

Get the current optimization target. Returns: Optimization target

double * rdFSQP::getUpperBound (   )

Get upper bounds on the controls. Note that the address is returned and not a copy; Returns: Pointer to the array of upper bounds.

void rdFSQP::setLowerBound (  double  aLower[]  )

Set lower bounds on the controls. Parameters: aLower  Array of lower bounds. The size of aLower should be at least the number of controls.

void rdFSQP::setLowerBound (  int  aIndex, double  aLower )

Set the lower bound for a specified control. Parameters: aIndex  Index of the control. aLower  Lower bound.

void rdFSQP::setLowerBound (  double  aLower  )

Set a lower bound on the controls. Parameters: aLower  Lower bound for all controls.

rdOptimizationTarget * rdFSQP::setTarget (  rdOptimizationTarget *  aTarget  )

Set the optimization target. It is permissible to change the target at any time. Calling this method, however, will destroy any information about the lower and upper bounds for the controls. The upper and lower bounds therefore need to be set again. Parameters: aTarget  New optimization target. Returns: Previous optimiztion target.

void rdFSQP::setUpperBound (  double  aUpper[]  )

Set upper bounds on the controls. Parameters: aUpper  Array of upper bounds. The size of aUpper should be at least the the number of controls

void rdFSQP::setUpperBound (  int  aIndex, double  aUpper )

Set the upper bound for a specified control. Parameters: aIndex  Index of the control. aLower  Upper bound.

void rdFSQP::setUpperBound (  double  aUpper  )

Set a upper bound on the controls. Parameters: aUpper  Upper bound for all controls.

---

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

• rdFSQP.h
• rdFSQP.cpp

---