#include <MoochoPack_NLPAlgoState.hpp>
Inheritance diagram for MoochoPack::NLPAlgoState:
Public Types | |
| typedef Teuchos::RefCountPtr< const VectorSpace > | vec_space_ptr_t |
| | |
Constructors/initializers | |
| void | set_decomp_sys (const Teuchos::RefCountPtr< DecompositionSystem > &decomp_sys) |
| Set the DecompositionSystem object that all share. | |
| void | set_space_x (const Teuchos::RefCountPtr< const VectorSpace > &space_x) |
| Set the VectorSpace of x. | |
| void | set_space_c (const Teuchos::RefCountPtr< const VectorSpace > &space_c) |
| Set the VectorSpace of c. | |
| void | set_space_range (const vec_space_ptr_t &space_range) |
| Set the VectorSpace of the range space (py). | |
| void | set_space_null (const vec_space_ptr_t &space_null) |
| Set the VectorSpace of the null space (pz). | |
| NLPAlgoState (const decomp_sys_ptr_t &decomp_sys=Teuchos::null, const vec_space_ptr_t &space_x=Teuchos::null, const vec_space_ptr_t &space_c=Teuchos::null, const vec_space_ptr_t &space_range=Teuchos::null, const vec_space_ptr_t &space_null=Teuchos::null) | |
| Construct. | |
| virtual | ~NLPAlgoState () |
| | |
Basis Pivot Info (variable reduction decompositions only) | |
| void | set_P_var_current (const Teuchos::RefCountPtr< Permutation > &P_var_current) |
| Current permutation for variables. | |
| void | set_P_var_last (const Teuchos::RefCountPtr< Permutation > &P_var_last) |
| Previous permutation for variables. | |
| void | set_P_equ_current (const Teuchos::RefCountPtr< Permutation > &P_equ_current) |
| Current permutation for equality constraints. | |
| void | set_P_equ_last (const Teuchos::RefCountPtr< Permutation > &P_equ_last) |
| Previous permutation for equality constraints. | |
Decomposition information | |
| Scaled KKT error for optimality ||rGL|| Scaled KKT error for feasibility ||c|| and ||hl <= h <= hu|| Scaled KKT error for complementarity (bounds) GL: Gradient of the Lagrangian ( n x 1 ) rGL: Reduced gradient of the Lagrangian ( (n-m) x 1 ) lambda: Lagrange multipliers for the equality constraints 'c' ( m x 1 ) nu: Difference between Lagrange multipiers for the upper and lower bounds ( n x 1 ) | |
| void | equ_decomp (const Range1D &equ_decomp) |
| Range of decomposed equality constraints [1,r]. | |
| const Range1D & | equ_decomp () const |
| void | equ_undecomp (const Range1D &equ_undecomp) |
| Range of undecomposed equality constraints [r+1,m]. | |
| const Range1D & | equ_undecomp () const |
Protected Member Functions | |
| void | update_iq_id (const std::string &iq_name, iq_id_encap *iq_id) const |
| | |
| void | update_index_type_iq_id (const std::string &iq_name, iq_id_encap *iq_id) |
| | |
| void | update_value_type_iq_id (const std::string &iq_name, iq_id_encap *iq_id) |
| | |
| void | update_vector_iq_id (const std::string &iq_name, const VectorSpace::space_ptr_t &vec_space, EVecSpaceType vec_space_type, iq_id_encap *iq_id) |
| | |
This in an interface to a set of data specific to a reduced space SQP algorithms. The iteration quantites are abstracted within IterQuantityAccess<> objects. A set of boilerplate macros are used to add the necessary declarations and implemetations of these iteration quantity access functions. As shown by these macros the access methods are declared virtual so that subclasses can override these methods. Otherwise, much of these could have been declared inline.
The implementation defined in this class uses IterQuantityAccessContiguous<> for iteration quantities of type index_type, value_type and VectorMutable with a default of one storage location. The default implementation is able to create the VectorMutable iteration quantities by using VectorSpace objects that the client sets this up with.
For all other types of iteration quantities (i.e. MatrixOp etc.) the client is responsible for setting the iteration quantity object of type IterQuantityAccess<>. The client can also change the type of class used for any iteration quantity by simply calling AlgorithmState::set_iter_quant(...).
The number of storage locations for any iteration quantity of type IterQuantityAccessContiguous<> can be changed by fetching the iteration quantity using the access methods defined here and then using dynamic_cast<> and calling the IterQuantityAccessContiguous<>::resize(...) method.
Note that the underlying AlgorithmState object will not know about the iteration quantity objects with default implementations until the access functions have been called at least once.
ToDo: Finish documentation.
Definition at line 251 of file MoochoPack_NLPAlgoState.hpp.
|
|
Definition at line 264 of file MoochoPack_NLPAlgoState.hpp. |
|
||||||||||||||||||||||||
|
Construct. Initializes num_basis() == 0 Definition at line 102 of file MoochoPack_NLPAlgoState.cpp. |
|
|
Definition at line 327 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Set the DecompositionSystem object that all share.
Definition at line 288 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Set the VectorSpace of x.
Definition at line 290 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Set the VectorSpace of c.
Definition at line 292 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Set the VectorSpace of the range space (py). Calling this method will cause all of the vector iteration quantity objects set in this space to be updated with this vector space (factory) object. Definition at line 90 of file MoochoPack_NLPAlgoState.cpp. |
|
|
Set the VectorSpace of the null space (pz). Calling this method will cause all of the vector iteration quantity objects set in this space to be updated with this vector space (factory) object. Definition at line 96 of file MoochoPack_NLPAlgoState.cpp. |
|
|
Range of decomposed equality constraints [1,r].
Definition at line 445 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Range of undecomposed equality constraints [r+1,m].
Definition at line 447 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Current permutation for variables.
Definition at line 455 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Previous permutation for variables.
Definition at line 457 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Current permutation for equality constraints.
Definition at line 459 of file MoochoPack_NLPAlgoState.hpp. |
|
|
Previous permutation for equality constraints.
Definition at line 461 of file MoochoPack_NLPAlgoState.hpp. |
|
||||||||||||
|
|
|
||||||||||||
|
Definition at line 188 of file MoochoPack_NLPAlgoState.cpp. |
|
||||||||||||
|
Definition at line 218 of file MoochoPack_NLPAlgoState.cpp. |
|
||||||||||||||||||||
|
Definition at line 248 of file MoochoPack_NLPAlgoState.cpp. |
1.3.9.1