DUNE PDELab (git)

Dune::PDELab::DisjointPartitioningDataHandle< GFS, V > Class Template Reference

GatherScatter data handle for creating a disjoint DOF partitioning. More...

#include <dune/pdelab/gridfunctionspace/genericdatahandle.hh>

Public Member Functions

 DisjointPartitioningDataHandle (const GFS &gfs_, V &v_, bool init_vector=true)
 Creates a new DisjointPartitioningDataHandle. More...
 
bool contains (int dim, int codim) const
 returns true if data for this codim should be communicated
 
bool fixedSize (int dim, int codim) const
 returns true if size per entity of given dim and codim is a constant
 
size_type size (const Entity &e) const
 how many objects of type DataType have to be sent for a given entity More...
 
size_t size (const EntityType &e) const
 how many objects of type DataType have to be sent for a given entity More...
 
std::enable_if< CommunicationDescriptor::wrap_buffer &&AlwaysTrue< Entity >::value >::type gather (MessageBuffer &buff, const Entity &e) const
 pack data from user to message buffer - version with support for sending leaf ordering sizes
 
std::enable_if<!CommunicationDescriptor::wrap_buffer &&AlwaysTrue< Entity >::value >::type gather (MessageBuffer &buff, const Entity &e) const
 pack data from user to message buffer - version without support for sending leaf ordering sizes
 
void gather (MessageBufferImp &buff, const EntityType &e) const
 pack data from user to message buffer More...
 
std::enable_if< CommunicationDescriptor::wrap_buffer &&AlwaysTrue< Entity >::value >::type scatter (MessageBuffer &buff, const Entity &e, size_type n)
 unpack data from message buffer to user More...
 
std::enable_if<!CommunicationDescriptor::wrap_buffer &&AlwaysTrue< Entity >::value >::type scatter (MessageBuffer &buff, const Entity &e, size_type n)
 unpack data from message buffer to user More...
 
void scatter (MessageBufferImp &buff, const EntityType &e, size_t n)
 unpack data from message buffer to user. More...
 

Detailed Description

template<class GFS, class V>
class Dune::PDELab::DisjointPartitioningDataHandle< GFS, V >

GatherScatter data handle for creating a disjoint DOF partitioning.

This data handle will associate each DOF with a unique rank, creating a nonoverlapping partitioning of the unknowns. The rank for a DOF is chosen by finding the lowest rank on which the associated grid entity belongs to either the interior or the border partition.

Note
In order to work correctly, the data handle must be communicated on the Dune::InteriorBorder_All_Interface and the result vector must be initialized with the MPI rank value.

Constructor & Destructor Documentation

◆ DisjointPartitioningDataHandle()

template<class GFS , class V >
Dune::PDELab::DisjointPartitioningDataHandle< GFS, V >::DisjointPartitioningDataHandle ( const GFS &  gfs_,
V &  v_,
bool  init_vector = true 
)
inline

Creates a new DisjointPartitioningDataHandle.

Creates a new DisjointPartitioningDataHandle and by default initializes the result vector with the current MPI rank. If you have already done that externally, you can skip the initialization.

Parameters
gfs_The GridFunctionSpace to operate on.
v_The result vector.
init_vectorFlag to control whether the result vector will be initialized.

Member Function Documentation

◆ gather()

void Dune::CommDataHandleIF< GFSDataHandle< GFS, V, DisjointPartitioningGatherScatter< V::ElementType >, EntityDataCommunicationDescriptor< V::ElementType > > , CommunicationDescriptor::DataType >::gather ( MessageBufferImp &  buff,
const EntityType &  e 
) const
inlineinherited

pack data from user to message buffer

Parameters
buffmessage buffer provided by the grid
eentity for which date should be packed to buffer

◆ scatter() [1/3]

std::enable_if< CommunicationDescriptor::wrap_buffer &&AlwaysTrue< Entity >::value >::type Dune::PDELab::GFSDataHandle< GFS, V, DisjointPartitioningGatherScatter< V::ElementType > , EntityDataCommunicationDescriptor< V::ElementType > >::scatter ( MessageBuffer &  buff,
const Entity e,
size_type  n 
)
inlineinherited

unpack data from message buffer to user

n is the number of objects sent by the sender

This is the version with support for receiving leaf ordering sizes

◆ scatter() [2/3]

std::enable_if<!CommunicationDescriptor::wrap_buffer &&AlwaysTrue< Entity >::value >::type Dune::PDELab::GFSDataHandle< GFS, V, DisjointPartitioningGatherScatter< V::ElementType > , EntityDataCommunicationDescriptor< V::ElementType > >::scatter ( MessageBuffer &  buff,
const Entity e,
size_type  n 
)
inlineinherited

unpack data from message buffer to user

n is the number of objects sent by the sender

This is the version without support for receiving leaf ordering sizes

◆ scatter() [3/3]

void Dune::CommDataHandleIF< GFSDataHandle< GFS, V, DisjointPartitioningGatherScatter< V::ElementType >, EntityDataCommunicationDescriptor< V::ElementType > > , CommunicationDescriptor::DataType >::scatter ( MessageBufferImp &  buff,
const EntityType &  e,
size_t  n 
)
inlineinherited

unpack data from message buffer to user.

Parameters
buffmessage buffer provided by the grid. This is not const, because the buffer has an internal iterator that gets advanced when reading from the buffer.
eentity for which date should be unpacked from buffer
nnumber of data written to buffer for this entity before

◆ size() [1/2]

size_type Dune::PDELab::GFSDataHandle< GFS, V, DisjointPartitioningGatherScatter< V::ElementType > , EntityDataCommunicationDescriptor< V::ElementType > >::size ( const Entity e) const
inlineinherited

how many objects of type DataType have to be sent for a given entity

Note: Only the sender side needs to know this size.

◆ size() [2/2]

size_t Dune::CommDataHandleIF< GFSDataHandle< GFS, V, DisjointPartitioningGatherScatter< V::ElementType >, EntityDataCommunicationDescriptor< V::ElementType > > , CommunicationDescriptor::DataType >::size ( const EntityType &  e) const
inlineinherited

how many objects of type DataType have to be sent for a given entity

Note
Only the sender side needs to know this size.
Parameters
eentity for which the size should be determined

The documentation for this class was generated from the following file:
Creative Commons License   |  Legal Statements / Impressum  |  Hosted by TU Dresden  |  generated with Hugo v0.111.3 (Jul 15, 22:36, 2024)