C# Class Cureos.Numerics.IpoptProblem

Managed wrapper and base class for setup and solution of an Ipopt problem.

Inheritance: IDisposable

ファイルを表示 Open project: cureos/csipopt Class Usage Examples

Public Methods

Method	Description
AddOption ( string keyword, double val ) : bool	Function for adding a floating point option.
AddOption ( string keyword, int val ) : bool	Function for adding an integer option.
AddOption ( string keyword, string val ) : bool	Function for adding a string option.
Dispose ( ) : void	Implement the IDisposable interface to release the Ipopt DLL resources held by this class
IpoptProblem ( int n, double x_L, double x_U, int m, double g_L, double g_U, int nele_jac, int nele_hess, Eval_F_CB eval_f_cb, Eval_G_CB eval_g_cb, Eval_Grad_F_CB eval_grad_f_cb, Eval_Jac_G_CB eval_jac_g_cb, Eval_H_CB eval_h_cb ) : System	Constructor for creating a new Ipopt Problem object using native function delegates. This function initializes an object that can be passed to the IpoptSolve call. It contains the basic definition of the optimization problem, such as number of variables and constraints, bounds on variables and constraints, information about the derivatives, and the callback function for the computation of the optimization problem functions and derivatives. During this call, the options file ipopt.opt is read as well.
IpoptProblem ( int n, double x_L, double x_U, int m, double g_L, double g_U, int nele_jac, int nele_hess, EvaluateObjectiveDelegate eval_f_cb, EvaluateConstraintsDelegate eval_g_cb, EvaluateObjectiveGradientDelegate eval_grad_f_cb, EvaluateJacobianDelegate eval_jac_g_cb, EvaluateHessianDelegate eval_h_cb ) : System	Constructor for creating a new Ipopt Problem object using managed function delegates. This function initializes an object that can be passed to the IpoptSolve call. It contains the basic definition of the optimization problem, such as number of variables and constraints, bounds on variables and constraints, information about the derivatives, and the callback function for the computation of the optimization problem functions and derivatives. During this call, the options file ipopt.opt is read as well.
OpenOutputFile ( string file_name, int print_level ) : bool	Method for opening an output file for a given name with given print level.
SetIntermediateCallback ( IntermediateDelegate intermediate_cb ) : bool	Setting a callback function for the "intermediate callback" method in the optimizer. This gives control back to the user once per iteration. If set, it provides the user with some information on the state of the optimization. This can be used to print some user-defined output. It also gives the user a way to terminate the optimization prematurely. If the callback method returns false, Ipopt will terminate the optimization. Calling this set method to set the CB pointer to null disables the intermediate callback functionality.
SetIntermediateCallback ( Intermediate_CB intermediate_cb ) : bool	Setting a callback function for the "intermediate callback" method in the optimizer. This gives control back to the user once per iteration. If set, it provides the user with some information on the state of the optimization. This can be used to print some user-defined output. It also gives the user a way to terminate the optimization prematurely. If the callback method returns false, Ipopt will terminate the optimization. Calling this set method to set the CB pointer to null disables the intermediate callback functionality.
SetScaling ( double obj_scaling, double x_scaling, double g_scaling ) : bool	Optional function for setting scaling parameter for the NLP. This corresponds to the get_scaling_parameters method in TNLP. If the pointers x_scaling or g_scaling are null, then no scaling for x resp. g is done.
SolveProblem ( double x, double &obj_val, double g = null, double mult_g = null, double mult_x_L = null, double mult_x_U = null ) : IpoptReturnCode	Function calling the IPOPT optimization algorithm for a problem previously defined with the constructor. IPOPT will use the options previously specified with AddOption for this problem.
eval_f ( int n, double x, bool new_x, double &obj_value ) : bool	Dummy implementation of the managed objective function evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.
eval_g ( int n, double x, bool new_x, int m, double g ) : bool	Dummy implementation of the managed constraints evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.
eval_grad_f ( int n, double x, bool new_x, double grad_f ) : bool	Dummy implementation of the managed objective function gradient evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.
eval_h ( int n, double x, bool new_x, double obj_factor, int m, double lambda, bool new_lambda, int nele_hess, int iRow, int jCol, double values ) : bool	Dummy implementation of the managed Hessian evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.
eval_jac_g ( int n, double x, bool new_x, int m, int nele_jac, int iRow, int jCol, double values ) : bool	Dummy implementation of the managed Jacobian evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.
intermediate ( IpoptAlgorithmMode alg_mod, int iter_count, double obj_value, double inf_pr, double inf_du, double mu, double d_norm, double regularization_size, double alpha_du, double alpha_pr, int ls_trials ) : bool	Dummy implementation of the managed intermediate callback method delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

Protected Methods

Method	Description
Dispose ( bool disposing ) : void	Dispose(bool disposing) executes in two distinct scenarios. If disposing equals true, the method has been called directly or indirectly by a user's code. Managed and unmanaged resources can be disposed. If disposing equals false, the method has been called by the runtime from inside the finalizer and you should not reference other objects. Only unmanaged resources can be disposed.
IpoptProblem ( int n, double x_L, double x_U, int m, double g_L, double g_U, int nele_jac, int nele_hess, bool useNativeCallbackFunctions = false, bool useHessianApproximation = false, bool useIntermediateCallback = false ) : System	Constructor for creating a subclassed Ipopt Problem object using managed or native function delegates. This is the preferred constructor when subclassing IpoptProblem. Prerequisite is that the managed/native optimization function delegates are implemented in the inheriting class. This function initializes an object that can be passed to the IpoptSolve call. It contains the basic definition of the optimization problem, such as number of variables and constraints, bounds on variables and constraints, information about the derivatives, and the callback function for the computation of the optimization problem functions and derivatives. During this call, the options file ipopt.opt is read as well.

Private Methods

Method	Description
eval_f ( int n, double x, IpoptBoolType new_x, double &obj_value, IntPtr p_user_data ) : IpoptBoolType
eval_g ( int n, double x, IpoptBoolType new_x, int m, double g, IntPtr p_user_data ) : IpoptBoolType
eval_grad_f ( int n, double x, IpoptBoolType new_x, double grad_f, IntPtr p_user_data ) : IpoptBoolType
eval_h ( int n, IntPtr p_x, IpoptBoolType new_x, double obj_factor, int m, IntPtr p_lambda, IpoptBoolType new_lambda, int nele_hess, IntPtr p_iRow, IntPtr p_jCol, IntPtr p_values, IntPtr p_user_data ) : IpoptBoolType
eval_h ( int n, double x, IpoptBoolType new_x, double obj_factor, int m, double lambda, IpoptBoolType new_lambda, int nele_hess, int iRow, int jCol, double values, IntPtr p_user_data ) : IpoptBoolType
eval_jac_g ( int n, IntPtr p_x, IpoptBoolType new_x, int m, int nele_jac, IntPtr p_iRow, IntPtr p_jCol, IntPtr p_values, IntPtr p_user_data ) : IpoptBoolType
eval_jac_g ( int n, double x, IpoptBoolType new_x, int m, int nele_jac, int iRow, int jCol, double values, IntPtr p_user_data ) : IpoptBoolType
intermediate ( IpoptAlgorithmMode alg_mod, int iter_count, double obj_value, double inf_pr, double inf_du, double mu, double d_norm, double regularization_size, double alpha_du, double alpha_pr, int ls_trials, IntPtr p_user_data ) : IpoptBoolType

Method Details

AddOption() public method

Function for adding a floating point option.

public AddOption ( string keyword, double val ) : bool
keyword	string	Name of option
val	double	Floating point value of option
return	bool

AddOption() public method

Function for adding an integer option.

public AddOption ( string keyword, int val ) : bool
keyword	string	Name of option
val	int	Integer value of option
return	bool

AddOption() public method

Function for adding a string option.

public AddOption ( string keyword, string val ) : bool
keyword	string	Name of option
val	string	String value of option
return	bool

Dispose() public method

Implement the IDisposable interface to release the Ipopt DLL resources held by this class

public Dispose ( ) : void
return	void

Dispose() protected method

Dispose(bool disposing) executes in two distinct scenarios. If disposing equals true, the method has been called directly or indirectly by a user's code. Managed and unmanaged resources can be disposed. If disposing equals false, the method has been called by the runtime from inside the finalizer and you should not reference other objects. Only unmanaged resources can be disposed.

protected Dispose ( bool disposing ) : void
disposing	bool	true if Dispose method is explicitly called, false otherwise.
return	void

IpoptProblem() public method

Constructor for creating a new Ipopt Problem object using native function delegates. This function initializes an object that can be passed to the IpoptSolve call. It contains the basic definition of the optimization problem, such as number of variables and constraints, bounds on variables and constraints, information about the derivatives, and the callback function for the computation of the optimization problem functions and derivatives. During this call, the options file ipopt.opt is read as well.

public IpoptProblem ( int n, double x_L, double x_U, int m, double g_L, double g_U, int nele_jac, int nele_hess, Eval_F_CB eval_f_cb, Eval_G_CB eval_g_cb, Eval_Grad_F_CB eval_grad_f_cb, Eval_Jac_G_CB eval_jac_g_cb, Eval_H_CB eval_h_cb ) : System
n	int	Number of optimization variables
x_L	double	Lower bounds on variables. This array of size n is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// less or equal than the number specified by option 'nlp_lower_bound_inf' is interpreted to be minus infinity.
x_U	double	Upper bounds on variables. This array of size n is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// greater or equal than the number specified by option 'nlp_upper_bound_inf' is interpreted to be plus infinity.
m	int	Number of constraints.
g_L	double	Lower bounds on constraints. This array of size m is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// less or equal than the number specified by option 'nlp_lower_bound_inf' is interpreted to be minus infinity.
g_U	double	Upper bounds on constraints. This array of size m is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// greater or equal than the number specified by option 'nlp_upper_bound_inf' is interpreted to be plus infinity.
nele_jac	int	Number of non-zero elements in constraint Jacobian.
nele_hess	int	Number of non-zero elements in Hessian of Lagrangian.
eval_f_cb	Eval_F_CB	Native callback function for evaluating objective function
eval_g_cb	Eval_G_CB	Native callback function for evaluating constraint functions
eval_grad_f_cb	Eval_Grad_F_CB	Native callback function for evaluating gradient of objective function
eval_jac_g_cb	Eval_Jac_G_CB	Native callback function for evaluating Jacobian of constraint functions
eval_h_cb	Eval_H_CB	Native callback function for evaluating Hessian of Lagrangian function
return	System

IpoptProblem() public method

Constructor for creating a new Ipopt Problem object using managed function delegates. This function initializes an object that can be passed to the IpoptSolve call. It contains the basic definition of the optimization problem, such as number of variables and constraints, bounds on variables and constraints, information about the derivatives, and the callback function for the computation of the optimization problem functions and derivatives. During this call, the options file ipopt.opt is read as well.

public IpoptProblem ( int n, double x_L, double x_U, int m, double g_L, double g_U, int nele_jac, int nele_hess, EvaluateObjectiveDelegate eval_f_cb, EvaluateConstraintsDelegate eval_g_cb, EvaluateObjectiveGradientDelegate eval_grad_f_cb, EvaluateJacobianDelegate eval_jac_g_cb, EvaluateHessianDelegate eval_h_cb ) : System
n	int	Number of optimization variables
x_L	double	Lower bounds on variables. This array of size n is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// less or equal than the number specified by option 'nlp_lower_bound_inf' is interpreted to be minus infinity.
x_U	double	Upper bounds on variables. This array of size n is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// greater or equal than the number specified by option 'nlp_upper_bound_inf' is interpreted to be plus infinity.
m	int	Number of constraints.
g_L	double	Lower bounds on constraints. This array of size m is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// less or equal than the number specified by option 'nlp_lower_bound_inf' is interpreted to be minus infinity.
g_U	double	Upper bounds on constraints. This array of size m is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// greater or equal than the number specified by option 'nlp_upper_bound_inf' is interpreted to be plus infinity.
nele_jac	int	Number of non-zero elements in constraint Jacobian.
nele_hess	int	Number of non-zero elements in Hessian of Lagrangian.
eval_f_cb	EvaluateObjectiveDelegate	Managed callback function for evaluating objective function
eval_g_cb	EvaluateConstraintsDelegate	Managed callback function for evaluating constraint functions
eval_grad_f_cb	EvaluateObjectiveGradientDelegate	Managed callback function for evaluating gradient of objective function
eval_jac_g_cb	EvaluateJacobianDelegate	Managed callback function for evaluating Jacobian of constraint functions
eval_h_cb	EvaluateHessianDelegate	Managed callback function for evaluating Hessian of Lagrangian function
return	System

IpoptProblem() protected method

Constructor for creating a subclassed Ipopt Problem object using managed or native function delegates. This is the preferred constructor when subclassing IpoptProblem. Prerequisite is that the managed/native optimization function delegates are implemented in the inheriting class. This function initializes an object that can be passed to the IpoptSolve call. It contains the basic definition of the optimization problem, such as number of variables and constraints, bounds on variables and constraints, information about the derivatives, and the callback function for the computation of the optimization problem functions and derivatives. During this call, the options file ipopt.opt is read as well.

protected IpoptProblem ( int n, double x_L, double x_U, int m, double g_L, double g_U, int nele_jac, int nele_hess, bool useNativeCallbackFunctions = false, bool useHessianApproximation = false, bool useIntermediateCallback = false ) : System
n	int	Number of optimization variables
x_L	double	Lower bounds on variables. This array of size n is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// less or equal than the number specified by option 'nlp_lower_bound_inf' is interpreted to be minus infinity.
x_U	double	Upper bounds on variables. This array of size n is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// greater or equal than the number specified by option 'nlp_upper_bound_inf' is interpreted to be plus infinity.
m	int	Number of constraints.
g_L	double	Lower bounds on constraints. This array of size m is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// less or equal than the number specified by option 'nlp_lower_bound_inf' is interpreted to be minus infinity.
g_U	double	Upper bounds on constraints. This array of size m is copied internally, so that the /// caller can change the incoming data after return without that IpoptProblem is modified. Any value /// greater or equal than the number specified by option 'nlp_upper_bound_inf' is interpreted to be plus infinity.
nele_jac	int	Number of non-zero elements in constraint Jacobian.
nele_hess	int	Number of non-zero elements in Hessian of Lagrangian.
useNativeCallbackFunctions	bool	If set to true, native callback functions are used to setup /// the Ipopt problem; if set to false, managed callback functions are used.
useHessianApproximation	bool	If set to true, the Ipopt optimizer creates a limited memory /// Hessian approximation and the eval_h (managed or native) method need not be implemented. /// If set to false, an exact Hessian should be evaluated using the appropriate Hessian evaluation method.
useIntermediateCallback	bool	If set to true, the intermediate method (managed or native) will be called /// after each full iteration. If false, the intermediate callback function will not be called.
return	System

OpenOutputFile() public method

Method for opening an output file for a given name with given print level.

public OpenOutputFile ( string file_name, int print_level ) : bool
file_name	string	Name of output file
print_level	int	Level of printed information
return	bool

SetIntermediateCallback() public method

Setting a callback function for the "intermediate callback" method in the optimizer. This gives control back to the user once per iteration. If set, it provides the user with some information on the state of the optimization. This can be used to print some user-defined output. It also gives the user a way to terminate the optimization prematurely. If the callback method returns false, Ipopt will terminate the optimization. Calling this set method to set the CB pointer to null disables the intermediate callback functionality.

public SetIntermediateCallback ( IntermediateDelegate intermediate_cb ) : bool
intermediate_cb	IntermediateDelegate	Managed intermediate callback function
return	bool

SetIntermediateCallback() public method

public SetIntermediateCallback ( Intermediate_CB intermediate_cb ) : bool
intermediate_cb	Intermediate_CB	Native intermediate callback function
return	bool

SetScaling() public method

Optional function for setting scaling parameter for the NLP. This corresponds to the get_scaling_parameters method in TNLP. If the pointers x_scaling or g_scaling are null, then no scaling for x resp. g is done.

public SetScaling ( double obj_scaling, double x_scaling, double g_scaling ) : bool
obj_scaling	double	Scaling of the objective function
x_scaling	double	Scaling of the problem variables
g_scaling	double	Scaling of the constraint functions
return	bool

SolveProblem() public method

Function calling the IPOPT optimization algorithm for a problem previously defined with the constructor. IPOPT will use the options previously specified with AddOption for this problem.

public SolveProblem ( double x, double &obj_val, double g = null, double mult_g = null, double mult_x_L = null, double mult_x_U = null ) : IpoptReturnCode
x	double	Input: Starting point; Output: Optimal solution
obj_val	double	Final value of objective function (output only)
g	double	Values of constraint at final point (output only - ignored if null on input)
mult_g	double	Final multipliers for constraints (output only - ignored if null on input)
mult_x_L	double	Final multipliers for lower variable bounds (output only - ignored if null on input)
mult_x_U	double	Final multipliers for upper variable bounds (output only - ignored if null on input)
return	IpoptReturnCode

eval_f() public method

Dummy implementation of the managed objective function evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

public eval_f ( int n, double x, bool new_x, double &obj_value ) : bool
n	int
x	double
new_x	bool
obj_value	double
return	bool

eval_g() public method

Dummy implementation of the managed constraints evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

public eval_g ( int n, double x, bool new_x, int m, double g ) : bool
n	int
x	double
new_x	bool
m	int
g	double
return	bool

eval_grad_f() public method

Dummy implementation of the managed objective function gradient evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

public eval_grad_f ( int n, double x, bool new_x, double grad_f ) : bool
n	int
x	double
new_x	bool
grad_f	double
return	bool

eval_h() public method

Dummy implementation of the managed Hessian evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

public eval_h ( int n, double x, bool new_x, double obj_factor, int m, double lambda, bool new_lambda, int nele_hess, int iRow, int jCol, double values ) : bool
n	int
x	double
new_x	bool
obj_factor	double
m	int
lambda	double
new_lambda	bool
nele_hess	int
iRow	int
jCol	int
values	double
return	bool

eval_jac_g() public method

Dummy implementation of the managed Jacobian evaluation delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

public eval_jac_g ( int n, double x, bool new_x, int m, int nele_jac, int iRow, int jCol, double values ) : bool
n	int
x	double
new_x	bool
m	int
nele_jac	int
iRow	int
jCol	int
values	double
return	bool

intermediate() public method

Dummy implementation of the managed intermediate callback method delegate. Recommended action is to override this method in a subclass and use the protected constructor of this base class to initialize the subclassed object.

public intermediate ( IpoptAlgorithmMode alg_mod, int iter_count, double obj_value, double inf_pr, double inf_du, double mu, double d_norm, double regularization_size, double alpha_du, double alpha_pr, int ls_trials ) : bool
alg_mod	IpoptAlgorithmMode
iter_count	int
obj_value	double
inf_pr	double
inf_du	double
mu	double
d_norm	double
regularization_size	double
alpha_du	double
alpha_pr	double
ls_trials	int
return	bool