|
Ipopt Documentation
|
|
|
Ipopt Documentation
|
|
This is a wrapper around a given TNLP class that takes out a list of constraints that are given to the constructor. More...
#include <IpTNLPReducer.hpp>
Inheritance diagram for Ipopt::TNLPReducer:Private Member Functions | |
Default Compiler Generated Methods | |
(Hidden to avoid implicit creation/calling). These methods are not implemented and we do not want the compiler to implement them for us, so we declare them private and do not define them. This ensures that they will not be implicitly created/called. | |
| TNLPReducer () | |
| Default Constructor. | |
| TNLPReducer (const TNLPReducer &) | |
| Copy Constructor. | |
| void | operator= (const TNLPReducer &) |
| Default Assignment Operator. | |
Private Attributes | |
| Index | n_g_skip_ |
| Number of constraints to be skipped. | |
| Index * | index_g_skip_ |
| Array of indices of the constraints that are to be skipped. | |
| IndexStyleEnum | index_style_orig_ |
| Index style for original problem. | |
| Index * | g_keep_map_ |
| Map from original constraints to new constraints. | |
| Index | m_reduced_ |
| Number of constraints in reduced NLP. | |
| Index | nnz_jac_g_reduced_ |
| Number of Jacobian nonzeros in the reduced NLP. | |
| Index | nnz_jac_g_skipped_ |
| Number of Jacobian nonzeros that are skipped. | |
| Index * | jac_g_skipped_ |
| Array of Jacobian elements that are to be skipped in increasing order. | |
| Index | n_xL_skip_ |
| Number of lower variable bounds to be skipped. | |
| Index * | index_xL_skip_ |
| Array of indices of the lower variable bounds to be skipped. | |
| Index | n_xU_skip_ |
| Number of upper variable bounds to be skipped. | |
| Index * | index_xU_skip_ |
| Array of indices of the upper variable bounds to be skipped. | |
| Index | n_x_fix_ |
| Number of variables that are to be fixed to initial value. | |
| Index * | index_x_fix_ |
| Array of indices of the variables that are to be fixed. | |
original TNLP | |
| SmartPtr< TNLP > | tnlp_ |
| Index | m_orig_ |
| Index | nnz_jac_g_orig_ |
Additional Inherited Members | |
| Public Types inherited from Ipopt::TNLP | |
| enum | LinearityType { LINEAR , NON_LINEAR } |
| Linearity-types of variables and constraints. More... | |
| enum | IndexStyleEnum { C_STYLE = 0 , FORTRAN_STYLE = 1 } |
| typedef std::map< std::string, std::vector< std::string > > | StringMetaDataMapType |
| typedef std::map< std::string, std::vector< Index > > | IntegerMetaDataMapType |
| typedef std::map< std::string, std::vector< Number > > | NumericMetaDataMapType |
This is a wrapper around a given TNLP class that takes out a list of constraints that are given to the constructor.
It is provided for convenience, if one wants to experiment with problems that consist of only a subset of the constraints. But keep in mind that this is not efficient, since behind the scenes we are still evaluation all functions and derivatives, and are making copies of the original data.
Definition at line 23 of file IpTNLPReducer.hpp.
| Ipopt::TNLPReducer::TNLPReducer | ( | TNLP & | tnlp, |
| Index | n_g_skip, | ||
| const Index * | index_g_skip, | ||
| Index | n_xL_skip, | ||
| const Index * | index_xL_skip, | ||
| Index | n_xU_skip, | ||
| const Index * | index_xU_skip, | ||
| Index | n_x_fix, | ||
| const Index * | index_f_fix | ||
| ) |
Constructor is given the indices of the constraints that should be taken out of the problem statement, as well as the original TNLP.
|
virtual |
Default destructor.
|
private |
Default Constructor.
|
private |
Copy Constructor.
|
virtual |
Method to request the initial information about the problem.
Ipopt uses this information when allocating the arrays that it will later ask you to fill with values. Be careful in this method since incorrect values will cause memory bugs which may be very difficult to find.
| n | (out) Storage for the number of variables \(x\) |
| m | (out) Storage for the number of constraints \(g(x)\) |
| nnz_jac_g | (out) Storage for the number of nonzero entries in the Jacobian |
| nnz_h_lag | (out) Storage for the number of nonzero entries in the Hessian |
| index_style | (out) Storage for the index style, the numbering style used for row/col entries in the sparse matrix format (TNLP::C_STYLE: 0-based, TNLP::FORTRAN_STYLE: 1-based; see also Triplet Format for Sparse Matrices) |
Implements Ipopt::TNLP.
|
virtual |
Method to request bounds on the variables and constraints.
| n | (in) the number of variables \(x\) in the problem |
| x_l | (out) the lower bounds \(x^L\) for the variables \(x\) |
| x_u | (out) the upper bounds \(x^U\) for the variables \(x\) |
| m | (in) the number of constraints \(g(x)\) in the problem |
| g_l | (out) the lower bounds \(g^L\) for the constraints \(g(x)\) |
| g_u | (out) the upper bounds \(g^U\) for the constraints \(g(x)\) |
The values of n and m that were specified in TNLP::get_nlp_info are passed here for debug checking. Setting a lower bound to a value less than or equal to the value of the option nlp_lower_bound_inf will cause Ipopt to assume no lower bound. Likewise, specifying the upper bound above or equal to the value of the option nlp_upper_bound_inf will cause Ipopt to assume no upper bound. These options are set to -1019 and 1019, respectively, by default, but may be modified by changing these options.
Implements Ipopt::TNLP.
|
virtual |
Method to request scaling parameters.
This is only called if the options are set to retrieve user scaling, that is, if nlp_scaling_method is chosen as "user-scaling". The method should provide scaling factors for the objective function as well as for the optimization variables and/or constraints. The return value should be true, unless an error occurred, and the program is to be aborted.
The value returned in obj_scaling determines, how Ipopt should internally scale the objective function. For example, if this number is chosen to be 10, then Ipopt solves internally an optimization problem that has 10 times the value of the original objective function provided by the TNLP. In particular, if this value is negative, then Ipopt will maximize the objective function instead of minimizing it.
The scaling factors for the variables can be returned in x_scaling, which has the same length as x in the other TNLP methods, and the factors are ordered like x. use_x_scaling needs to be set to true, if Ipopt should scale the variables. If it is false, no internal scaling of the variables is done. Similarly, the scaling factors for the constraints can be returned in g_scaling, and this scaling is activated by setting use_g_scaling to true.
As a guideline, we suggest to scale the optimization problem (either directly in the original formulation, or after using scaling factors) so that all sensitivities, i.e., all non-zero first partial derivatives, are typically of the order 0.1-10.
Reimplemented from Ipopt::TNLP.
|
virtual |
Method to request the variables linearity.
This method is never called by Ipopt, but is used by Bonmin to get information about which variables occur only in linear terms. Ipopt passes the array var_types of length at least n, which should be filled with the appropriate linearity type of the variables (TNLP::LINEAR or TNLP::NON_LINEAR).
The default implementation just returns false and does not fill the array.
Reimplemented from Ipopt::TNLP.
|
virtual |
Method to request the constraints linearity.
This method is never called by Ipopt, but is used by Bonmin to get information about which constraints are linear. Ipopt passes the array const_types of size m, which should be filled with the appropriate linearity type of the constraints (TNLP::LINEAR or TNLP::NON_LINEAR).
The default implementation just returns false and does not fill the array.
Reimplemented from Ipopt::TNLP.
|
virtual |
Method to request the starting point before iterating.
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| init_x | (in) if true, this method must provide an initial value for \(x\) |
| x | (out) the initial values for the primal variables \(x\) |
| init_z | (in) if true, this method must provide an initial value for the bound multipliers \(z^L\) and \(z^U\) |
| z_L | (out) the initial values for the bound multipliers \(z^L\) |
| z_U | (out) the initial values for the bound multipliers \(z^U\) |
| m | (in) the number of constraints \(g(x)\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| init_lambda | (in) if true, this method must provide an initial value for the constraint multipliers \(\lambda\) |
| lambda | (out) the initial values for the constraint multipliers, \(\lambda\) |
The boolean variables indicate whether the algorithm requires to have x, z_L/z_u, and lambda initialized, respectively. If, for some reason, the algorithm requires initializations that cannot be provided, false should be returned and Ipopt will stop. The default options only require initial values for the primal variables \(x\).
Note, that the initial values for bound multiplier components for absent bounds ( \(x^L_i=-\infty\) or \(x^U_i=\infty\)) are ignored.
Implements Ipopt::TNLP.
|
virtual |
Method to provide an Ipopt warm start iterate which is already in the form Ipopt requires it internally for warm starts.
This method is only for expert users. The default implementation does not provide a warm start iterate and returns false.
| warm_start_iterate | storage for warm start iterate in the form Ipopt requires it internally |
Reimplemented from Ipopt::TNLP.
|
virtual |
Method to request the value of the objective function.
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| x | (in) the values for the primal variables \(x\) at which the objective function \(f(x)\) is to be evaluated |
| new_x | (in) false if any evaluation method (eval_*) was previously called with the same values in x, true otherwise. This can be helpful when users have efficient implementations that calculate multiple outputs at once. Ipopt internally caches results from the TNLP and generally, this flag can be ignored. |
| obj_value | (out) storage for the value of the objective function \(f(x)\) |
Implements Ipopt::TNLP.
|
virtual |
Method to request the gradient of the objective function.
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| x | (in) the values for the primal variables \(x\) at which the gradient \(\nabla f(x)\) is to be evaluated |
| new_x | (in) false if any evaluation method (eval_*) was previously called with the same values in x, true otherwise; see also TNLP::eval_f |
| grad_f | (out) array to store values of the gradient of the objective function \(\nabla f(x)\). The gradient array is in the same order as the \(x\) variables (i.e., the gradient of the objective with respect to x[2] should be put in grad_f[2]). |
Implements Ipopt::TNLP.
|
virtual |
Method to request the constraint values.
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| x | (in) the values for the primal variables \(x\) at which the constraint functions \(g(x)\) are to be evaluated |
| new_x | (in) false if any evaluation method (eval_*) was previously called with the same values in x, true otherwise; see also TNLP::eval_f |
| m | (in) the number of constraints \(g(x)\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| g | (out) array to store constraint function values \(g(x)\), do not add or subtract the bound values \(g^L\) or \(g^U\). |
Implements Ipopt::TNLP.
|
virtual |
Method to request either the sparsity structure or the values of the Jacobian of the constraints.
The Jacobian is the matrix of derivatives where the derivative of constraint function \(g_i\) with respect to variable \(x_j\) is placed in row \(i\) and column \(j\). See Triplet Format for Sparse Matrices for a discussion of the sparse matrix format used in this method.
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| x | (in) first call: NULL; later calls: the values for the primal variables \(x\) at which the constraint Jacobian \(\nabla g(x)^T\) is to be evaluated |
| new_x | (in) false if any evaluation method (eval_*) was previously called with the same values in x, true otherwise; see also TNLP::eval_f |
| m | (in) the number of constraints \(g(x)\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| nele_jac | (in) the number of nonzero elements in the Jacobian; it will have the same value that was specified in TNLP::get_nlp_info |
| iRow | (out) first call: array of length nele_jac to store the row indices of entries in the Jacobian of the constraints; later calls: NULL |
| jCol | (out) first call: array of length nele_jac to store the column indices of entries in the Jacobian of the constraints; later calls: NULL |
| values | (out) first call: NULL; later calls: array of length nele_jac to store the values of the entries in the Jacobian of the constraints |
x and values will be NULL. If the arguments x and values are not NULL, then Ipopt expects that the value of the Jacobian as calculated from array x is stored in array values (using the same order as used when specifying the sparsity structure). At this call, the arguments iRow and jCol will be NULL. Implements Ipopt::TNLP.
|
virtual |
Method to request either the sparsity structure or the values of the Hessian of the Lagrangian.
The Hessian matrix that Ipopt uses is
\[ \sigma_f \nabla^2 f(x_k) + \sum_{i=1}^m\lambda_i\nabla^2 g_i(x_k) \]
for the given values for \(x\), \(\sigma_f\), and \(\lambda\). See Triplet Format for Sparse Matrices for a discussion of the sparse matrix format used in this method.
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| x | (in) first call: NULL; later calls: the values for the primal variables \(x\) at which the Hessian is to be evaluated |
| new_x | (in) false if any evaluation method (eval_*) was previously called with the same values in x, true otherwise; see also TNLP::eval_f |
| obj_factor | (in) factor \(\sigma_f\) in front of the objective term in the Hessian |
| m | (in) the number of constraints \(g(x)\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| lambda | (in) the values for the constraint multipliers \(\lambda\) at which the Hessian is to be evaluated |
| new_lambda | (in) false if any evaluation method was previously called with the same values in lambda, true otherwise |
| nele_hess | (in) the number of nonzero elements in the Hessian; it will have the same value that was specified in TNLP::get_nlp_info |
| iRow | (out) first call: array of length nele_hess to store the row indices of entries in the Hessian; later calls: NULL |
| jCol | (out) first call: array of length nele_hess to store the column indices of entries in the Hessian; later calls: NULL |
| values | (out) first call: NULL; later calls: array of length nele_hess to store the values of the entries in the Hessian |
x, lambda, and values will be NULL. If the arguments x, lambda, and values are not NULL, then Ipopt expects that the value of the Hessian as calculated from arrays x and lambda are stored in array values (using the same order as used when specifying the sparsity structure). At this call, the arguments iRow and jCol will be NULL.A default implementation is provided, in case the user wants to set quasi-Newton approximations to estimate the second derivatives and doesn't not need to implement this method.
Reimplemented from Ipopt::TNLP.
|
virtual |
This method is called when the algorithm has finished (successfully or not) so the TNLP can digest the outcome, e.g., store/write the solution, if any.
| status | (in) gives the status of the algorithm
|
| n | (in) the number of variables \(x\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| x | (in) the final values for the primal variables |
| z_L | (in) the final values for the lower bound multipliers |
| z_U | (in) the final values for the upper bound multipliers |
| m | (in) the number of constraints \(g(x)\) in the problem; it will have the same value that was specified in TNLP::get_nlp_info |
| g | (in) the final values of the constraint functions |
| lambda | (in) the final values of the constraint multipliers |
| obj_value | (in) the final value of the objective function |
| ip_data | (in) provided for expert users |
| ip_cq | (in) provided for expert users |
Implements Ipopt::TNLP.
|
virtual |
Intermediate Callback method for the user.
This method is called once per iteration (during the convergence check), and can be used to obtain information about the optimization status while Ipopt solves the problem, and also to request a premature termination.
The information provided by the entities in the argument list correspond to what Ipopt prints in the iteration summary (see also Ipopt Output), except for inf_pr, which by default corresponds to the original problem in the log but to the scaled internal problem in this callback. Further information can be obtained from the ip_data and ip_cq objects. The current iterate and violations of feasibility and optimality can be accessed via the methods Ipopt::TNLP::get_curr_iterate() and Ipopt::TNLP::get_curr_violations(). These methods translate values for the internal representation of the problem from ip_data and ip_cq objects into the TNLP representation.
It is not required to implement (overload) this method. The default implementation always returns true.
Reimplemented from Ipopt::TNLP.
Return the number of variables that appear nonlinearly in the objective function or in at least one constraint function.
If -1 is returned as number of nonlinear variables, Ipopt assumes that all variables are nonlinear. Otherwise, it calls get_list_of_nonlinear_variables with an array into which the indices of the nonlinear variables should be written - the array has the length num_nonlin_vars, which is identical with the return value of get_number_of_nonlinear_variables(). It is assumed that the indices are counted starting with 1 in the FORTRAN_STYLE, and 0 for the C_STYLE.
The default implementation returns -1, i.e., all variables are assumed to be nonlinear.
Reimplemented from Ipopt::TNLP.
|
virtual |
Return the indices of all nonlinear variables.
This method is called only if limited-memory quasi-Newton option is used and get_number_of_nonlinear_variables() returned a positive number. This number is provided in parameter num_nonlin_var.
The method must store the indices of all nonlinear variables in pos_nonlin_vars, where the numbering starts with 0 order 1, depending on the numbering style determined in get_nlp_info.
Reimplemented from Ipopt::TNLP.
|
private |
Default Assignment Operator.
Definition at line 215 of file IpTNLPReducer.hpp.
|
private |
Definition at line 216 of file IpTNLPReducer.hpp.
|
private |
Definition at line 217 of file IpTNLPReducer.hpp.
|
private |
Number of constraints to be skipped.
Definition at line 221 of file IpTNLPReducer.hpp.
|
private |
Array of indices of the constraints that are to be skipped.
This is provided at the beginning in the constructor.
Definition at line 227 of file IpTNLPReducer.hpp.
|
private |
Index style for original problem.
Internally, we use C-Style now.
Definition at line 233 of file IpTNLPReducer.hpp.
|
private |
Map from original constraints to new constraints.
A -1 means that a constraint is skipped.
Definition at line 239 of file IpTNLPReducer.hpp.
|
private |
Number of constraints in reduced NLP.
Definition at line 242 of file IpTNLPReducer.hpp.
|
private |
Number of Jacobian nonzeros in the reduced NLP.
Definition at line 245 of file IpTNLPReducer.hpp.
|
private |
Number of Jacobian nonzeros that are skipped.
Definition at line 248 of file IpTNLPReducer.hpp.
|
private |
Array of Jacobian elements that are to be skipped in increasing order.
Definition at line 251 of file IpTNLPReducer.hpp.
|
private |
Number of lower variable bounds to be skipped.
Definition at line 254 of file IpTNLPReducer.hpp.
|
private |
Array of indices of the lower variable bounds to be skipped.
Definition at line 257 of file IpTNLPReducer.hpp.
|
private |
Number of upper variable bounds to be skipped.
Definition at line 260 of file IpTNLPReducer.hpp.
|
private |
Array of indices of the upper variable bounds to be skipped.
Definition at line 263 of file IpTNLPReducer.hpp.
|
private |
Number of variables that are to be fixed to initial value.
Definition at line 266 of file IpTNLPReducer.hpp.
|
private |
Array of indices of the variables that are to be fixed.
Definition at line 269 of file IpTNLPReducer.hpp.