|
Tpetra parallel linear algebra
Version of the Day
|
All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Pages
|
Tpetra parallel linear algebra
Version of the Day
|
Abstract base class for objects that can be the source of an Import or Export operation, and that also know how to pack their data to send to the target object. More...
#include <Tpetra_Packable.hpp>
Public Member Functions | |
| virtual void | pack (const Teuchos::ArrayView< const LocalOrdinal > &exportLIDs, Teuchos::Array< Packet > &exports, const Teuchos::ArrayView< size_t > &numPacketsPerLID, size_t &constantNumPackets, Distributor &distor) const =0 |
| Pack the object's data for an Import or Export. More... | |
| virtual | ~Packable () |
| Destructor (virtual for memory safety of derived classes). More... | |
Abstract base class for objects that can be the source of an Import or Export operation, and that also know how to pack their data to send to the target object.
| Packet | The type of each entry of the array of packed data to be sent in the Import or Export operation. The type of packed data may differ from the type of actual data stored in the object. For example, a sparse matrix might need to pack both column indices (an integer type) and values (typically, but not always, a floating-point type), and it might choose any of various types to represent packed data. |
| LocalOrdinal | The type of local indices in the object. This is a template parameter because the pack() method includes as input a list of local indices to pack. See the documentation of Map for requirements. |
If an object implements Packable, then that object acknowledges that it knows how to pack its data as the source object of an Import or Export operation. The target object in general assumes responsibility for packing the source object's data. However, the target object (in its packAndPrepare method) may ask the source object to pack its own data, if the source object implements Packable.
It might make sense for Packable to inherit from SrcDistObject. However, that sets up the possibility of ambiguous multiple inheritance. For example, RowGraph inherits from Packable, and CrsGraph inherits from both RowGraph and DistObject. Furthermore, it is not necessary for a source object of an Import or Export to know how to pack itself. The ability to pack oneself is independent of the ability to be the source of an Import or Export. Packable exists mainly for syntactic enforcement of the interface needed for an object to know how to pack itself for an Import or Export.
Definition at line 99 of file Tpetra_Packable.hpp.
|
inlinevirtual |
Destructor (virtual for memory safety of derived classes).
Definition at line 133 of file Tpetra_Packable.hpp.
|
pure virtual |
Pack the object's data for an Import or Export.
| exportLIDs | [in] List of the local indices of the entries which the source object will send out. |
| exports | [out] On exit, the buffer packed with data to send. This object may resize the array if necessary. |
| numPacketsPerLID | [out] On exit, the implementation of this method must do one of two things: set numPacketsPerLID[i] to contain the number of packets to be exported for exportLIDs[i] and set constantNumPackets to zero, or set constantNumPackets to a nonzero value. If the latter, the implementation need not fill numPacketsPerLID. |
| constantNumPackets | [out] On exit, zero if numPacketsPerLID has variable contents (different size for each local index). If nonzero, then it is expected that the number of packets per local index is constant, and that constantNumPackets is that value. |
| distor | [in] The Distributor object we are using. Implementations may ignore this object. We provide it for consistency with DistObject's packAndPrepare method. |
Implemented in Tpetra::CrsGraph< LocalOrdinal, GlobalOrdinal, Node >, Tpetra::CrsGraph< LO, GO, node_type >, Tpetra::RowMatrix< Scalar, LocalOrdinal, GlobalOrdinal, Node >, Tpetra::RowMatrix< Scalar, LO, GO, Node >, Tpetra::RowGraph< LocalOrdinal, GlobalOrdinal, Node >, and Tpetra::RowGraph< LO, GO, node_type >.
1.8.5