Solver for quadratic programming problems with box constraints (i.e., only upper and lower bounds). More...

#include <BoxQP.h>

Collaboration diagram for nmpc_ddp::BoxQP< VarDim >:

Classes
struct	Configuration
	Configuration. More...

struct	TraceData
	Data to trace optimization loop. More...

Public Types
using	VarDimVector = Eigen::Matrix< double, VarDim, 1 >
	Type of vector of variables dimension. More...

using	VarVarDimMatrix = Eigen::Matrix< double, VarDim, VarDim >
	Type of matrix of variables x variables dimension. More...

using	VarDimArray = Eigen::Array< bool, VarDim, 1 >
	Type of boolean array of variables dimension. More...

Public Member Functions
EIGEN_MAKE_ALIGNED_OPERATOR_NEW	BoxQP (int var_dim=VarDim)
	Constructor. More...

VarDimVector	solve (const VarVarDimMatrix &H, const VarDimVector &g, const VarDimVector &lower, const VarDimVector &upper)
	Solve optimization. More...

VarDimVector	solve (const VarVarDimMatrix &H, const VarDimVector &g, const VarDimVector &lower, const VarDimVector &upper, const VarDimVector &initial_x)
	Solve optimization. More...

Configuration &	config ()
	Accessor to configuration. More...

const Configuration &	config () const
	Const accessor to configuration. More...

const std::vector< TraceData > &	traceDataList () const
	Const accessor to trace data list. More...

Public Attributes
const int	var_dim_ = 0
	Dimension of decision variables. More...

int	retval_ = 0
	Return value. More...

const std::unordered_map< int, std::string >	retstr_
	Return string. More...

std::unique_ptr< Eigen::LLT< Eigen::MatrixXd > >	llt_free_
	Cholesky decomposition (LLT) of free block of objective Hessian matrix. More...

std::vector< int >	free_idxs_
	Indices of free dimensions in decision variables. More...

Protected Attributes
Configuration	config_
	Configuration. More...

std::vector< TraceData >	trace_data_list_
	Sequence of trace data. More...

Detailed Description

template<int VarDim>
class nmpc_ddp::BoxQP< VarDim >

Solver for quadratic programming problems with box constraints (i.e., only upper and lower bounds).

Template Parameters

VarDim dimension of decision variables

See the following for a detailed algorithm.

Y Tassa, N Mansard, E Todorov. Control-limited differential dynamic programming. ICRA, 2014.
https://www.mathworks.com/matlabcentral/fileexchange/52069-ilqg-ddp-trajectory-optimization

Definition at line 19 of file BoxQP.h.

Member Typedef Documentation

◆ VarDimArray

template<int VarDim>

using nmpc_ddp::BoxQP< VarDim >::VarDimArray = Eigen::Array<bool, VarDim, 1>

Type of boolean array of variables dimension.

Definition at line 29 of file BoxQP.h.

◆ VarDimVector

template<int VarDim>

using nmpc_ddp::BoxQP< VarDim >::VarDimVector = Eigen::Matrix<double, VarDim, 1>

Type of vector of variables dimension.

Definition at line 23 of file BoxQP.h.

◆ VarVarDimMatrix

template<int VarDim>

using nmpc_ddp::BoxQP< VarDim >::VarVarDimMatrix = Eigen::Matrix<double, VarDim, VarDim>

Type of matrix of variables x variables dimension.

Definition at line 26 of file BoxQP.h.

Constructor & Destructor Documentation

◆ BoxQP()

template<int VarDim>

EIGEN_MAKE_ALIGNED_OPERATOR_NEW nmpc_ddp::BoxQP< VarDim >::BoxQP ( int var_dim = VarDim )

inline

Constructor.

Parameters

var_dim dimension of decision variables

Note: dimensions in parameter can be omitted if a fixed value is given in the template value.

Definition at line 101 of file BoxQP.h.

Member Function Documentation

◆ config() [1/2]

template<int VarDim>

Configuration& nmpc_ddp::BoxQP< VarDim >::config ( )

inline

Accessor to configuration.

Definition at line 350 of file BoxQP.h.

◆ config() [2/2]

template<int VarDim>

const Configuration& nmpc_ddp::BoxQP< VarDim >::config ( ) const

inline

Const accessor to configuration.

Definition at line 356 of file BoxQP.h.

◆ solve() [1/2]

template<int VarDim>

VarDimVector nmpc_ddp::BoxQP< VarDim >::solve	(	const VarVarDimMatrix &	H,
		const VarDimVector &	g,
		const VarDimVector &	lower,
		const VarDimVector &	upper
	)

inline

Solve optimization.

Parameters

H	Hessian matrix of objective function
g	gradient vector of objective function
lower	lower limit of decision variables
upper	upper limit of decision variables

Definition at line 126 of file BoxQP.h.

◆ solve() [2/2]

template<int VarDim>

VarDimVector nmpc_ddp::BoxQP< VarDim >::solve	(	const VarVarDimMatrix &	H,
		const VarDimVector &	g,
		const VarDimVector &	lower,
		const VarDimVector &	upper,
		const VarDimVector &	initial_x
	)

inline

Solve optimization.

Parameters

H	Hessian matrix of objective function
g	gradient vector of objective function
lower	lower limit of decision variables
upper	upper limit of decision variables
initial_x	initial guess of decision variables

Definition at line 141 of file BoxQP.h.

◆ traceDataList()

template<int VarDim>

const std::vector<TraceData>& nmpc_ddp::BoxQP< VarDim >::traceDataList ( ) const

inline

Const accessor to trace data list.

Definition at line 362 of file BoxQP.h.

Member Data Documentation

◆ config_

template<int VarDim>

Configuration nmpc_ddp::BoxQP< VarDim >::config_

protected

Configuration.

Definition at line 393 of file BoxQP.h.

◆ free_idxs_

template<int VarDim>

std::vector<int> nmpc_ddp::BoxQP< VarDim >::free_idxs_

Indices of free dimensions in decision variables.

Definition at line 389 of file BoxQP.h.

◆ llt_free_

template<int VarDim>

std::unique_ptr<Eigen::LLT<Eigen::MatrixXd> > nmpc_ddp::BoxQP< VarDim >::llt_free_

Cholesky decomposition (LLT) of free block of objective Hessian matrix.

Definition at line 386 of file BoxQP.h.

◆ retstr_

template<int VarDim>

const std::unordered_map<int, std::string> nmpc_ddp::BoxQP< VarDim >::retstr_

Initial value:

= {{-2, "Gradient of search direction is positive"},
                                                        {-1, "Hessian is not positive definite"},
                                                        {0, "Computation is not finished"},
                                                        {1, "Maximum main iterations exceeded"},
                                                        {2, "Maximum line-search iterations exceeded"},
                                                        {3, "No bounds, returning Newton point"},
                                                        {4, "Improvement smaller than tolerance"},
                                                        {5, "Gradient norm smaller than tolerance"},
                                                        {6, "All dimensions are clamped"}}

Return string.

Definition at line 375 of file BoxQP.h.

◆ retval_

template<int VarDim>

int nmpc_ddp::BoxQP< VarDim >::retval_ = 0

Return value.

Definition at line 372 of file BoxQP.h.

◆ trace_data_list_

template<int VarDim>

std::vector<TraceData> nmpc_ddp::BoxQP< VarDim >::trace_data_list_

protected

Sequence of trace data.

Definition at line 396 of file BoxQP.h.

◆ var_dim_

template<int VarDim>

const int nmpc_ddp::BoxQP< VarDim >::var_dim_ = 0

Dimension of decision variables.

Definition at line 369 of file BoxQP.h.

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

/home/runner/work/NMPC/NMPC/catkin_ws/src/NMPC/nmpc_ddp/include/nmpc_ddp/BoxQP.h

Classes

Public Types

Public Member Functions

Public Attributes

Protected Attributes

Detailed Description

template<int VarDim> class nmpc_ddp::BoxQP< VarDim >

Member Typedef Documentation

◆ VarDimArray

◆ VarDimVector

◆ VarVarDimMatrix

Constructor & Destructor Documentation

◆ BoxQP()

Member Function Documentation

◆ config() [1/2]

◆ config() [2/2]

◆ solve() [1/2]

◆ solve() [2/2]

◆ traceDataList()

Member Data Documentation

◆ config_

◆ free_idxs_

◆ llt_free_

◆ retstr_

◆ retval_

◆ trace_data_list_

◆ var_dim_

template<int VarDim>
class nmpc_ddp::BoxQP< VarDim >