Reference documentation for deal.II version 9.1.0-pre
|
#include <deal.II/multigrid/mg_transfer.h>
Public Member Functions | |
void | clear () |
template<int dim, typename Number2 , int spacedim> | |
void | copy_to_mg (const DoFHandler< dim, spacedim > &mg_dof, MGLevelObject< LinearAlgebra::distributed::Vector< Number >> &dst, const LinearAlgebra::distributed::Vector< Number2 > &src) const |
template<int dim, typename Number2 , int spacedim> | |
void | copy_from_mg (const DoFHandler< dim, spacedim > &mg_dof, LinearAlgebra::distributed::Vector< Number2 > &dst, const MGLevelObject< LinearAlgebra::distributed::Vector< Number >> &src) const |
template<int dim, typename Number2 , int spacedim> | |
void | copy_from_mg_add (const DoFHandler< dim, spacedim > &mg_dof, LinearAlgebra::distributed::Vector< Number2 > &dst, const MGLevelObject< LinearAlgebra::distributed::Vector< Number >> &src) const |
void | set_component_to_block_map (const std::vector< unsigned int > &map) |
std::size_t | memory_consumption () const |
void | print_indices (std::ostream &os) const |
Public Member Functions inherited from MGTransferBase< LinearAlgebra::distributed::Vector< Number > > | |
virtual | ~MGTransferBase () override=default |
virtual void | prolongate (const unsigned int to_level, LinearAlgebra::distributed::Vector< Number > &dst, const LinearAlgebra::distributed::Vector< Number > &src) const =0 |
virtual void | restrict_and_add (const unsigned int from_level, LinearAlgebra::distributed::Vector< Number > &dst, const LinearAlgebra::distributed::Vector< Number > &src) const =0 |
Public Member Functions inherited from Subscriptor | |
Subscriptor () | |
Subscriptor (const Subscriptor &) | |
Subscriptor (Subscriptor &&) noexcept | |
virtual | ~Subscriptor () |
Subscriptor & | operator= (const Subscriptor &) |
Subscriptor & | operator= (Subscriptor &&) noexcept |
void | subscribe (const char *identifier=nullptr) const |
void | unsubscribe (const char *identifier=nullptr) const |
unsigned int | n_subscriptions () const |
template<typename StreamType > | |
void | list_subscribers (StreamType &stream) const |
void | list_subscribers () const |
template<class Archive > | |
void | serialize (Archive &ar, const unsigned int version) |
Protected Member Functions | |
template<int dim, typename Number2 , int spacedim> | |
void | copy_to_mg (const DoFHandler< dim, spacedim > &mg_dof, MGLevelObject< LinearAlgebra::distributed::Vector< Number >> &dst, const LinearAlgebra::distributed::Vector< Number2 > &src, const bool solution_transfer) const |
template<int dim, int spacedim> | |
void | fill_and_communicate_copy_indices (const DoFHandler< dim, spacedim > &mg_dof) |
Additional Inherited Members | |
Static Public Member Functions inherited from Subscriptor | |
static::ExceptionBase & | ExcInUse (int arg1, std::string arg2, std::string arg3) |
static::ExceptionBase & | ExcNoSubscriber (std::string arg1, std::string arg2) |
Implementation of transfer between the global vectors and the multigrid levels for use in the derived class MGTransferPrebuilt and other classes. This class is a specialization for the case of LinearAlgebra::distributed::Vector that requires a few different calling routines as compared to the parallel vectors in the PETScWrappers and TrilinosWrappers namespaces.
Definition at line 404 of file mg_transfer.h.
void MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::clear | ( | ) |
Reset the object to the state it had right after the default constructor.
Definition at line 370 of file mg_level_global_transfer.cc.
void MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::copy_to_mg | ( | const DoFHandler< dim, spacedim > & | mg_dof, |
MGLevelObject< LinearAlgebra::distributed::Vector< Number >> & | dst, | ||
const LinearAlgebra::distributed::Vector< Number2 > & | src | ||
) | const |
Transfer from a vector on the global grid to vectors defined on each of the levels separately for the active degrees of freedom. In particular, for a globally refined mesh only the finest level in dst
is filled as a plain copy of src
. All the other level objects are left untouched.
void MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::copy_from_mg | ( | const DoFHandler< dim, spacedim > & | mg_dof, |
LinearAlgebra::distributed::Vector< Number2 > & | dst, | ||
const MGLevelObject< LinearAlgebra::distributed::Vector< Number >> & | src | ||
) | const |
Transfer from multi-level vector to normal vector.
Copies data from active portions of an MGVector into the respective positions of a Vector<number>
. In order to keep the result consistent, constrained degrees of freedom are set to zero.
void MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::copy_from_mg_add | ( | const DoFHandler< dim, spacedim > & | mg_dof, |
LinearAlgebra::distributed::Vector< Number2 > & | dst, | ||
const MGLevelObject< LinearAlgebra::distributed::Vector< Number >> & | src | ||
) | const |
Add a multi-level vector to a normal vector.
Works as the previous function, but probably not for continuous elements.
void MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::set_component_to_block_map | ( | const std::vector< unsigned int > & | map | ) |
If this object operates on BlockVector objects, we need to describe how the individual vector components are mapped to the blocks of a vector. For example, for a Stokes system, we have dim+1 vector components for velocity and pressure, but we may want to use block vectors with only two blocks for all velocities in one block, and the pressure variables in the other.
By default, if this function is not called, block vectors have as many blocks as the finite element has vector components. However, this can be changed by calling this function with an array that describes how vector components are to be grouped into blocks. The meaning of the argument is the same as the one given to the DoFTools::count_dofs_per_component function.
std::size_t MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::memory_consumption | ( | ) | const |
Memory used by this object.
Definition at line 419 of file mg_level_global_transfer.cc.
void MGLevelGlobalTransfer< LinearAlgebra::distributed::Vector< Number > >::print_indices | ( | std::ostream & | os | ) | const |
Print the copy index fields for debugging purposes.
Definition at line 389 of file mg_level_global_transfer.cc.
|
protected |
Internal function to perform transfer of residuals or solutions basesd on the flag solution_transfer
.
|
protected |
Internal function to fill
copy_indices*. Called by derived classes.
Definition at line 297 of file mg_level_global_transfer.cc.
|
protected |
Sizes of the multi-level vectors.
Definition at line 504 of file mg_transfer.h.
|
protected |
Mapping for the copy_to_mg() and copy_from_mg() functions. Here only index pairs locally owned is stored.
The data is organized as follows: one vector per level. Each element of these vectors contains first the global index, then the level index.
Definition at line 513 of file mg_transfer.h.
|
protected |
Same as above, but used to transfer solution vectors.
Definition at line 520 of file mg_transfer.h.
|
protected |
Additional degrees of freedom for the copy_to_mg() function. These are the ones where the global degree of freedom is locally owned and the level degree of freedom is not.
Organization of the data is like for copy_indices_mine
.
Definition at line 530 of file mg_transfer.h.
|
protected |
Same as above, but used to transfer solution vectors.
Definition at line 536 of file mg_transfer.h.
|
protected |
Additional degrees of freedom for the copy_from_mg() function. These are the ones where the level degree of freedom is locally owned and the global degree of freedom is not.
Organization of the data is like for copy_indices_mine
.
Definition at line 546 of file mg_transfer.h.
|
protected |
Same as above, but used to transfer solution vectors.
Definition at line 552 of file mg_transfer.h.
|
protected |
Stores whether the copy operation from the global to the level vector is actually a plain copy to the finest level. This means that the grid has no adaptive refinement and the numbering on the finest multigrid level is the same as in the global case.
Definition at line 560 of file mg_transfer.h.
|
protected |
Stores whether the copy operation from the global to the level vector is actually a plain copy to the finest level except for a renumbering within the finest level of the degrees of freedom. This means that the grid has no adaptive refinement.
Definition at line 568 of file mg_transfer.h.
|
protected |
The vector that stores what has been given to the set_component_to_block_map() function.
Definition at line 574 of file mg_transfer.h.
|
protected |
The mg_constrained_dofs of the level systems.
Definition at line 582 of file mg_transfer.h.
|
mutableprotected |
In the function copy_to_mg, we need to access ghosted entries of the global vector for inserting into the level vectors. This vector is populated with those entries.
Definition at line 589 of file mg_transfer.h.
|
mutableprotected |
Same as above but used when working with solution vectors.
Definition at line 595 of file mg_transfer.h.
|
mutableprotected |
In the function copy_from_mg, we access all level vectors with certain ghost entries for inserting the result into a global vector.
Definition at line 602 of file mg_transfer.h.
|
mutableprotected |
Same as above but used when working with solution vectors.
Definition at line 608 of file mg_transfer.h.