Epetra_CrsGraph: A class for constructing and using sparse compressed row graphs. More...

#include <Epetra_CrsGraph.h>

Inheritance diagram for Epetra_CrsGraph:

[legend]

Collaboration diagram for Epetra_CrsGraph:

[legend]

Public Member Functions
Constructors/Destructor
	Epetra_CrsGraph (Epetra_DataAccess CV, const Epetra_BlockMap &RowMap, const int *NumIndicesPerRow, bool StaticProfile=false)
	Epetra_CrsGraph constuctor with variable number of indices per row.
	Epetra_CrsGraph (Epetra_DataAccess CV, const Epetra_BlockMap &RowMap, int NumIndicesPerRow, bool StaticProfile=false)
	Epetra_CrsGraph constuctor with fixed number of indices per row.
	Epetra_CrsGraph (Epetra_DataAccess CV, const Epetra_BlockMap &RowMap, const Epetra_BlockMap &ColMap, const int *NumIndicesPerRow, bool StaticProfile=false)
	Epetra_CrsGraph constuctor with variable number of indices per row.
	Epetra_CrsGraph (Epetra_DataAccess CV, const Epetra_BlockMap &RowMap, const Epetra_BlockMap &ColMap, int NumIndicesPerRow, bool StaticProfile=false)
	Epetra_CrsGraph constuctor with fixed number of indices per row.
	Epetra_CrsGraph (const Epetra_CrsGraph &Graph)
	Copy constructor.
virtual	~Epetra_CrsGraph ()
	Epetra_CrsGraph Destructor.
Insertion/Removal methods
int	InsertGlobalIndices (int GlobalRow, int NumIndices, int *Indices)
	Enter a list of elements in a specified global row of the graph.
int	InsertGlobalIndices (long long GlobalRow, int NumIndices, long long *Indices)
int	RemoveGlobalIndices (int GlobalRow, int NumIndices, int *Indices)
	Remove a list of elements from a specified global row of the graph.
int	RemoveGlobalIndices (long long GlobalRow, int NumIndices, long long *Indices)
int	RemoveGlobalIndices (long long Row)
	Remove all indices from a specified global row of the graph.
int	InsertMyIndices (int LocalRow, int NumIndices, int *Indices)
	Enter a list of elements in a specified local row of the graph.
int	RemoveMyIndices (int LocalRow, int NumIndices, int *Indices)
	Remove a list of elements from a specified local row of the graph.
int	RemoveMyIndices (int Row)
	Remove all indices from a specified local row of the graph.
Transformation methods
int	FillComplete ()
	Tranform to local index space. Perform other operations to allow optimal matrix operations.
int	FillComplete (const Epetra_BlockMap &DomainMap, const Epetra_BlockMap &RangeMap)
	Transform to local index space using specified Domain/Range maps. Perform other operations to allow optimal matrix operations.
int	OptimizeStorage ()
	Make consecutive row index sections contiguous, minimize internal storage used for constructing graph.
Extraction methods
int	ExtractGlobalRowCopy (int GlobalRow, int LenOfIndices, int &NumIndices, int *Indices) const
	Extract a list of elements in a specified global row of the graph. Put into storage allocated by calling routine.
int	ExtractGlobalRowCopy (long long GlobalRow, int LenOfIndices, int &NumIndices, long long *Indices) const
int	ExtractMyRowCopy (int LocalRow, int LenOfIndices, int &NumIndices, int *Indices) const
	Extract a list of elements in a specified local row of the graph. Put into storage allocated by calling routine.
int	ExtractGlobalRowView (int GlobalRow, int &NumIndices, int *&Indices) const
	Get a view of the elements in a specified global row of the graph.
int	ExtractGlobalRowView (long long GlobalRow, int &NumIndices, long long *&Indices) const
int	ExtractMyRowView (int LocalRow, int &NumIndices, int *&Indices) const
	Get a view of the elements in a specified local row of the graph.
Graph Properties Query Methods
bool	Filled () const
	If FillComplete() has been called, this query returns true, otherwise it returns false.
bool	StorageOptimized () const
	If OptimizeStorage() has been called, this query returns true, otherwise it returns false.
bool	IndicesAreGlobal () const
	If column indices are in global range, this query returns true, otherwise it returns false.
bool	IndicesAreLocal () const
	If column indices are in local range, this query returns true, otherwise it returns false.
bool	LowerTriangular () const
	If graph is lower triangular in local index space, this query returns true, otherwise it returns false.
bool	UpperTriangular () const
	If graph is upper triangular in local index space, this query returns true, otherwise it returns false.
bool	NoDiagonal () const
	If graph has no diagonal entries in global index space, this query returns true, otherwise it returns false.
bool	MyGlobalRow (int GID) const
	Returns true of GID is owned by the calling processor, otherwise it returns false.
bool	MyGlobalRow (long long GID) const
bool	HaveColMap () const
	Returns true if we have a well-defined ColMap, and returns false otherwise.
Attribute access functions
int	NumMyRows () const
	Returns the number of matrix rows on this processor.
int	NumGlobalRows () const
	Returns the number of matrix rows in global matrix.
long long	NumGlobalRows64 () const
int	NumMyCols () const
	Returns the number of entries in the set of column-indices that appear on this processor.
int	NumGlobalCols () const
	Returns the number of matrix columns in global matrix.
long long	NumGlobalCols64 () const
int	NumGlobalNonzeros () const
	Returns the number of indices in the global graph.
long long	NumGlobalNonzeros64 () const
int	NumGlobalDiagonals () const
	Returns the number of diagonal entries in the global graph, based on global row/column index comparisons.
long long	NumGlobalDiagonals64 () const
int	NumMyDiagonals () const
	Returns the number of diagonal entries in the local graph, based on global row/column index comparisons.
int	NumMyBlockRows () const
	Returns the number of block matrix rows on this processor.
int	NumGlobalBlockRows () const
	Returns the number of Block matrix rows in global matrix.
long long	NumGlobalBlockRows64 () const
int	NumMyBlockCols () const
	Returns the number of Block matrix columns on this processor.
int	NumGlobalBlockCols () const
	Returns the number of Block matrix columns in global matrix.
long long	NumGlobalBlockCols64 () const
int	NumMyBlockDiagonals () const
	Returns the number of Block diagonal entries in the local graph, based on global row/column index comparisons.
int	NumGlobalBlockDiagonals () const
	Returns the number of Block diagonal entries in the global graph, based on global row/column index comparisons.
long long	NumGlobalBlockDiagonals64 () const
int	NumGlobalEntries () const
	Returns the number of entries in the global graph.
long long	NumGlobalEntries64 () const
int	NumMyEntries () const
	Returns the number of entries on this processor.
int	MaxRowDim () const
	Returns the max row dimension of block entries on the processor.
int	GlobalMaxRowDim () const
	Returns the max row dimension of block entries across all processors.
int	MaxColDim () const
	Returns the max column dimension of block entries on the processor.
int	GlobalMaxColDim () const
	Returns the max column dimension of block entries across all processors.
int	NumMyNonzeros () const
	Returns the number of indices in the local graph.
int	NumGlobalIndices (long long Row) const
	Returns the current number of nonzero entries in specified global row on this processor.
int	NumAllocatedGlobalIndices (long long Row) const
	Returns the allocated number of nonzero entries in specified global row on this processor.
int	MaxNumIndices () const
	Returns the maximum number of nonzero entries across all rows on this processor.
int	GlobalMaxNumIndices () const
	Returns the maximun number of nonzero entries across all rows across all processors.
int	MaxNumNonzeros () const
	Returns the maximum number of nonzero points across all rows on this processor.
int	GlobalMaxNumNonzeros () const
	Returns the maximun number of nonzero points across all rows across all processors.
int	NumMyIndices (int Row) const
	Returns the current number of nonzero entries in specified local row on this processor.
int	NumAllocatedMyIndices (int Row) const
	Returns the allocated number of nonzero entries in specified local row on this processor.
int	IndexBase () const
	Returns the index base for row and column indices for this graph.
long long	IndexBase64 () const
const Epetra_BlockMap &	RowMap () const
	Returns the RowMap associated with this graph.
int	ReplaceRowMap (const Epetra_BlockMap &newmap)
int	ReplaceColMap (const Epetra_BlockMap &newmap)
int	ReplaceDomainMapAndImporter (const Epetra_BlockMap &NewDomainMap, const Epetra_Import *NewImporter)
	Replaces the current DomainMap & Importer with the user-specified map object.
int	RemoveEmptyProcessesInPlace (const Epetra_BlockMap *NewMap)
	Remove processes owning zero rows from the Maps and their communicator.
const Epetra_BlockMap &	ColMap () const
	Returns the Column Map associated with this graph.
const Epetra_BlockMap &	DomainMap () const
	Returns the DomainMap associated with this graph.
const Epetra_BlockMap &	RangeMap () const
	Returns the RangeMap associated with this graph.
const Epetra_Import *	Importer () const
	Returns the Importer associated with this graph.
const Epetra_Export *	Exporter () const
	Returns the Exporter associated with this graph.
const Epetra_Comm &	Comm () const
	Returns a pointer to the Epetra_Comm communicator associated with this graph.
Local/Global ID methods
int	LRID (int GRID_in) const
	Returns the local row index for given global row index, returns -1 if no local row for this global row.
int	LRID (long long GRID_in) const
int	GRID (int LRID_in) const
	Returns the global row index for give local row index, returns IndexBase-1 if we don't have this local row.
long long	GRID64 (int LRID_in) const
int	LCID (int GCID_in) const
	Returns the local column index for given global column index, returns -1 if no local column for this global column.
int	LCID (long long GCID_in) const
int	GCID (int LCID_in) const
	Returns the global column index for give local column index, returns IndexBase-1 if we don't have this local column.
long long	GCID64 (int LCID_in) const
bool	MyGRID (int GRID_in) const
	Returns true if the GRID passed in belongs to the calling processor in this map, otherwise returns false.
bool	MyGRID (long long GRID_in) const
bool	MyLRID (int LRID_in) const
	Returns true if the LRID passed in belongs to the calling processor in this map, otherwise returns false.
bool	MyGCID (int GCID_in) const
	Returns true if the GCID passed in belongs to the calling processor in this map, otherwise returns false.
bool	MyGCID (long long GCID_in) const
bool	MyLCID (int LCID_in) const
	Returns true if the LRID passed in belongs to the calling processor in this map, otherwise returns false.
Inlined Operator Methods
int *	operator[] (int Loc)
	Inlined bracket operator for fast access to data. (Const and Non-const versions).
int *	operator[] (int Loc) const
Epetra_CrsGraph &	operator= (const Epetra_CrsGraph &Source)
	Assignment operator.
I/O Methods
virtual void	Print (std::ostream &os) const
	Print method.
void	PrintGraphData (std::ostream &os) const
void	PrintGraphData (std::ostream &os, int level) const
Deprecated methods: These methods still work, but will be removed in a future version
const Epetra_BlockMap &	ImportMap () const
	Use ColMap() instead.
int	TransformToLocal ()
	Use FillComplete() instead.
int	TransformToLocal (const Epetra_BlockMap DomainMap, const Epetra_BlockMap RangeMap)
	Use FillComplete(const Epetra_BlockMap& DomainMap, const Epetra_BlockMap& RangeMap) instead.
Public Member Functions inherited from Epetra_DistObject
	Epetra_DistObject (const Epetra_BlockMap &Map)
	Basic Epetra_DistObject constuctor.
	Epetra_DistObject (const Epetra_BlockMap &Map, const char *const Label)
	Epetra_DistObject (const Epetra_DistObject &Source)
	Epetra_DistObject copy constructor.
virtual	~Epetra_DistObject ()
	Epetra_DistObject destructor.
int	Import (const Epetra_SrcDistObject &A, const Epetra_Import &Importer, Epetra_CombineMode CombineMode, const Epetra_OffsetIndex *Indexor=0)
	Imports an Epetra_DistObject using the Epetra_Import object.
int	Import (const Epetra_SrcDistObject &A, const Epetra_Export &Exporter, Epetra_CombineMode CombineMode, const Epetra_OffsetIndex *Indexor=0)
	Imports an Epetra_DistObject using the Epetra_Export object.
int	Export (const Epetra_SrcDistObject &A, const Epetra_Import &Importer, Epetra_CombineMode CombineMode, const Epetra_OffsetIndex *Indexor=0)
	Exports an Epetra_DistObject using the Epetra_Import object.
int	Export (const Epetra_SrcDistObject &A, const Epetra_Export &Exporter, Epetra_CombineMode CombineMode, const Epetra_OffsetIndex *Indexor=0)
	Exports an Epetra_DistObject using the Epetra_Export object.
const Epetra_BlockMap &	Map () const
	Returns the address of the Epetra_BlockMap for this multi-vector.
const Epetra_Comm &	Comm () const
	Returns the address of the Epetra_Comm for this multi-vector.
bool	DistributedGlobal () const
	Returns true if this multi-vector is distributed global, i.e., not local replicated.
Public Member Functions inherited from Epetra_Object
	Epetra_Object (int TracebackModeIn=-1, bool set_label=true)
	Epetra_Object Constructor.
	Epetra_Object (const char *const Label, int TracebackModeIn=-1)
	Epetra_Object Constructor.
	Epetra_Object (const Epetra_Object &Object)
	Epetra_Object Copy Constructor.
virtual	~Epetra_Object ()
	Epetra_Object Destructor.
virtual int	ReportError (const std::string Message, int ErrorCode) const
	Error reporting method.
virtual void	SetLabel (const char *const Label)
	Epetra_Object Label definition using char *.
virtual const char *	Label () const
	Epetra_Object Label access funtion.
Public Member Functions inherited from Epetra_SrcDistObject
virtual	~Epetra_SrcDistObject ()
	Epetra_SrcDistObject destructor.

Expert Users and Developers Only
class	Epetra_CrsMatrix
class	Epetra_VbrMatrix
class	Epetra_FECrsGraph
class	Epetra_FECrsMatrix
class	Epetra_FEVbrMatrix
class	Epetra_OffsetIndex
int	ReferenceCount () const
	Returns the reference count of CrsGraphData.
const Epetra_CrsGraphData *	DataPtr () const
	Returns a pointer to the CrsGraphData instance this CrsGraph uses.
Epetra_IntSerialDenseVector &	ExpertExtractIndexOffset ()
	Returns a reference to the Epetra_IntSerialDenseVector used to hold the local IndexOffsets (CRS rowptr).
Epetra_IntSerialDenseVector &	ExpertExtractIndices ()
	Returns a reference to the Epetra_IntSerialDenseVector used to hold the local All_Indices (CRS colind).
void	SortGhostsAssociatedWithEachProcessor (bool Flag)
	Forces FillComplete() to locally order ghostnodes associated with each remote processor in ascending order.
int *	All_Indices () const
int *	IndexOffset () const
int *	NumIndicesPerRow () const
int *	NumAllocatedIndicesPerRow () const
int **	Indices () const
int *	Indices (int LocalRow) const
template<typename int_type>
int_type **	TIndices () const
template<typename int_type>
int_type *	TIndices (int LocalRow) const
bool	IndicesAreContiguous () const
bool	StaticProfile () const
bool	GlobalConstantsComputed () const
bool	FindGlobalIndexLoc (int LocalRow, int Index, int Start, int &Loc) const
bool	FindGlobalIndexLoc (int NumIndices, const int *Indices, int Index, int Start, int &Loc) const
bool	FindGlobalIndexLoc (int LocalRow, long long Index, int Start, int &Loc) const
bool	FindGlobalIndexLoc (int NumIndices, const long long *Indices, long long Index, int Start, int &Loc) const
bool	FindMyIndexLoc (int LocalRow, int Index, int Start, int &Loc) const
bool	FindMyIndexLoc (int NumIndices, const int *Indices, int Index, int Start, int &Loc) const
int	InsertIndices (int Row, int NumIndices, int *Indices)
int	InsertIndicesIntoSorted (int Row, int NumIndices, int *Indices)
int	InsertIndices (int Row, int NumIndices, long long *Indices)
int	InsertIndicesIntoSorted (int Row, int NumIndices, long long *Indices)
int	MakeIndicesLocal (const Epetra_BlockMap &DomainMap, const Epetra_BlockMap &RangeMap)
void	SetIndicesAreLocal (bool Flag)
void	SetIndicesAreGlobal (bool Flag)
void	SetSorted (bool Flag)
int	SortIndices ()
	Sort column indices, row-by-row, in ascending order.
bool	Sorted () const
	If SortIndices() has been called, this query returns true, otherwise it returns false.
int	RemoveRedundantIndices ()
	Removes any redundant column indices in the rows of the graph.
int	DetermineTriangular ()
bool	NoRedundancies () const
	If RemoveRedundantIndices() has been called, this query returns true, otherwise it returns false.

Additional Inherited Members
static void	SetTracebackMode (int TracebackModeValue)
	Set the value of the Epetra_Object error traceback report mode.
static int	GetTracebackMode ()
	Get the value of the Epetra_Object error report mode.
static std::ostream &	GetTracebackStream ()
	Get the output stream for error reporting.
static int	TracebackMode
Protected Member Functions inherited from Epetra_DistObject
virtual int	DoTransfer (const Epetra_SrcDistObject &A, Epetra_CombineMode CombineMode, int NumSameIDs, int NumPermuteIDs, int NumRemoteIDs, int NumExportIDs, int PermuteToLIDs, int PermuteFromLIDs, int RemoteLIDs, int ExportLIDs, int &LenExports, char &Exports, int &LenImports, char &Imports, Epetra_Distributor &Distor, bool DoReverse, const Epetra_OffsetIndex *Indexor)
	Perform actual transfer (redistribution) of data across memory images, using Epetra_Distributor object.
std::string	toString (const int &x) const
std::string	toString (const long long &x) const
std::string	toString (const double &x) const
Epetra_BlockMap	Map_
const Epetra_Comm *	Comm_
char *	Exports_
char *	Imports_
int	LenExports_
int	LenImports_
int *	Sizes_

Detailed Description

Epetra_CrsGraph: A class for constructing and using sparse compressed row graphs.

Epetra_CrsGraph enables the piecewise construction and use of sparse matrix graphs (the integer structure without values) where entries are intended for row access.

Epetra_CrsGraph is an attribute of all Epetra row-based matrix classes, defining their nonzero structure and also holding their Epetra_Map attributes.

Constructing Epetra_CrsGraph objects

Constructing Epetra_CrsGraph objects is a multi-step process. The basic steps are as follows:

Create Epetra_CrsGraph instance, including some initial storage, via constructor. In addition to the copy constructor, Epetra_CrsGraph has four different constructors. All four of these constructors have an argument, StaticProfile, which by default is set to false. If it is set to true, then the profile (the number of indices per row as defined by NumIndicesPerRow) will be rigidly enforced. Although this takes away flexibility, it allows a single array to be allocated for all indices. This decreases memory fragmentation and improves performance across many operations. A more detailed discussion of the StaticProfile option is found below.
1. User-provided row map, variable nonzero profile: This constructor is used to define the row distribution of the graph and specify a varying number of nonzero entries per row. It is best to use this constructor when the user will be inserting entries using global index values and wants every column index to be included in the graph. Note that in this case, the column map will be built for the user when FillComplete() is called. This constructor is also appropriate for when there is a large variation in the number of indices per row. If this is not the case, the next constructor may be more convenient to use.
2. User-provided row map, fixed nonzero profile: This constructor is used to define the row distribution of the graph and specify a fixed number of nonzero entries per row. It is best to use this constructor when the user will be inserting entries using global index values and wants every column index to be included in the graph. Note that in this case, the column map will be built for the user when FillComplete() is called. This constructor is also appropriate for when there is little or no variation in the number of indices per row.
3. User-provided row map, user-provided column map and variable nonzero profile: This constructor is used to define the row and column distribution of the graph, and specify a varying number of nonzero entries per row. It is best to use this constructor when the user will be inserting entries and already knows which columns of the matrix should be included on each processor. Note that in this case, the column map will not be built for the user when FillComplete() is called. Also, if the user attempts to insert a column index whose GID is not part of the column map on that process, the index will be discarded. This property can be used to "filter out" column entries that should be ignored. This constructor is also appropriate for when there is a large variation in the number of indices per row. If this is not the case, the next constructor may be more convenient to use.
4. User-provided row map, user-provided column map and fixed nonzero profile: This constructor is used to define the row and column distribution of the graph, and specify a fixed number of nonzero entries per row. It is best to use this constructor when the user will be inserting entries and already knows which columns of the matrix should be included on each processor. Note that in this case, the column map will not be built for the user when FillComplete() is called. Also, if the user attempts to insert a column index whose GID is not part of the column map on that process, the index will be discarded. This constructor is also appropriate for when there is little or no variation in the number of indices per row.
Enter row and column entry information via calls to the InsertGlobalIndices method.
Complete construction via FillComplete call, which performs the following tasks:
1. Transforms indices to local index space (after this, IndicesAreLocal()==true)
2. Sorts column-indices within each row
3. Compresses out any redundant indices within rows
4. Computes global data such as num-nonzeros, maximum row-lengths, etc.
(Optional) Optimize the graph storage via a call to OptimizeStorage.

Performance Enhancement Issues

The Epetra_CrsGraph class attempts to address four basic types of situations, depending on the user's primary concern:

Simple, flexible construction over minimal memory use or control of column indices: In this case the user wants to provide only a row distribution of the graph and insert indices without worrying about memory allocation performance. This type of user is best served by the constructor that requires only a row map, and a fixed number of indices per row. In fact, setting NumIndicesPerRow=0 is probably the best option.
Stronger control over memory allocation performance and use over flexibility and simplicity: In this case the user explicitly set StaticProfile to true and will provide values, either a single global int or an array of int's, for NumIndicesPerRow, such that the actual number of indices submitted to the graph will not exceed the estimates. Because we know that NumIndicesPerRow will not be exceeded, we can pre-allocate all of the storage for the graph as a single array. This is typically much more efficient.
Explicit control over column indices: In this case the user prescribes the column map. Given the column map, any index that is submitted for entry into the graph will be included only if they are present in the list of GIDs for the column map on the processor that submits the index. This feature allows the user to define a filter such that only certain columns will be kept. The user also prescribes the local ordering via this technique, since the ordering of GIDs in the column map imposes the local ordering.
Construction using local indices only: In some situations, users may want to build a graph using local index values only. In this case, the user must explicitly assign GIDs. This is done by prescribing the column map, in the same way as the previous situation.

Notes:

In all but the most advanced uses, users will typically not specify the column map. In other words, graph entries will be submitted using GIDs not LIDs and all entries that are submitted are intended to be inserted into the graph.
If a user is not particularly worried about performance, or really needs the flexibility associated with the first situation, then there is no need to explicitly manage the NumIndicesPerRow values or set StaticProfile to true. In this case, it is best to set NumIndicesPerRow to zero.
Users who are concerned about performance should carefully manage NumIndicesPerRow and set StaticProfile to true. This will give the best performance and use the least amount of memory.
A compromise approach would be to not set StaticProfile to true, giving the user flexibility, but then calling OptimizeStorage() once FillComplete() has been called. This approach requires additional temporary memory because the graph will be copied into an efficient data structure and the old memory deleted. However, once the copy has been made, the resulting data structure is as efficient as when StaticProfile is used.

Epetra_Map attributes

Epetra_CrsGraph objects have four Epetra_Map attributes.

The Epetra_Map attributes can be obtained via these accessor methods:

RowMap() Describes the numbering and distribution of the rows of the graph. The row-map exists and is valid for the entire life of the graph, having been passed in as a constructor argument. The set of graph rows is defined by the row-map and may not be changed. Rows may not be inserted or deleted by the user. The only change that may be made is that the user can replace the row-map with a compatible row-map (which is the same except for re-numbering) by calling the ReplaceRowMap() method.
ColMap() Describes the set of column-indices that appear in the rows in each processor's portion of the graph. Unless provided by the user at construction time, a valid column-map doesn't exist until FillComplete() is called.
RangeMap() Describes the range of the matrix operator. e.g., for a matrix-vector product operation, the result vector's map must be compatible with the range-map of the matrix operator. The range-map is usually the same as the row-map. The range-map is set equal to the row-map at graph creation time, but may be specified by the user when FillComplete() is called.
DomainMap() Describes the domain of the matrix operator. The domain-map can be specified by the user when FillComplete() is called. Until then, it is set equal to the row-map.

It is important to note that while the row-map and the range-map are often the same, the column-map and the domain-map are almost never the same. The set of entries in a distributed column-map almost always form overlapping sets, with entries being associated with more than one processor. A domain-map, on the other hand, must be a 1-to-1 map, with entries being associated with only a single processor.

Global versus Local indices

After creation and before FillComplete() has been called, the column-indices of the graph are in the global space as received from the user. One of the tasks performed by FillComplete() is to transform the indices to a local index space. The query methods IndicesAreGlobal() and IndicesAreLocal() return true or false depending on whether this transformation has been performed or not.

Note the behavior of several graph methods:

InsertGlobalIndices() returns an error if IndicesAreLocal()==true or StorageOptimized()==true
InsertMyIndices() returns an error if IndicesAreGlobal()==true or StorageOptimized()==true
RemoveGlobalIndices() returns an error if IndicesAreLocal()==true or if graph was constructed in View mode
RemoveMyIndices() returns an error if IndicesAreGlobal()==true or if graph was constructed in View mode
ExtractGlobalRowCopy() works regardless of state of indices
ExtractMyRowCopy() returns an error if IndicesAreGlobal()==true
ExtractGlobalRowView() returns an error if IndicesAreLocal()==true
ExtractMyRowView() returns an error if IndicesAreGlobal()==true

Note that even after a graph is constructed, it is possible to add or remove entries. However, FillComplete must then be called again to restore the graph to a consistent state.

Constructor & Destructor Documentation

◆ Epetra_CrsGraph() [1/5]

Epetra_CrsGraph::Epetra_CrsGraph	(	Epetra_DataAccess	CV,
		const Epetra_BlockMap &	RowMap,
		const int *	NumIndicesPerRow,
		bool	StaticProfile = false )

Epetra_CrsGraph constuctor with variable number of indices per row.

Creates a Epetra_CrsGraph object and allocates storage.

Parameters

CV	- (In) A Epetra_DataAccess enumerated type set to Copy or View.
RowMap	- (In) An Epetra_BlockMap (or Epetra_Map or Epetra_LocalMap) listing the rows that this processor will contribute to.In
NumIndicesPerRow	- (In) An integer array of length NumMyRows such that NumIndicesPerRow[i] indicates the (approximate if StaticProfile=false) number of entries in the ith row.
StaticProfile	- (In) Optional argument that indicates whether or not NumIndicesPerRow should be interpreted as an exact count of nonzeros, or should be used as an approximation. By default this value is false, allowing the profile to be determined dynamically. If the user sets it to true, then the memory allocation for the Epetra_CrsGraph object will be done in one large block, saving on memory fragmentation and generally improving the performance of matrix multiplication and solve kernels.

References RowMap().

Referenced by Epetra_CrsGraph(), and operator=().

◆ Epetra_CrsGraph() [2/5]

Epetra_CrsGraph::Epetra_CrsGraph	(	Epetra_DataAccess	CV,
		const Epetra_BlockMap &	RowMap,
		int	NumIndicesPerRow,
		bool	StaticProfile = false )

Epetra_CrsGraph constuctor with fixed number of indices per row.

Creates a Epetra_CrsGraph object and allocates storage.

Parameters

CV	- (In) A Epetra_DataAccess enumerated type set to Copy or View.
RowMap	- (In) An Epetra_BlockMap (or Epetra_Map or Epetra_LocalMap) listing the rows that this processor will contribute to.
NumIndicesPerRow	- (In) An integer that indicates the (approximate if StaticProfile=false) number of entries in the each row. Note that it is possible to use 0 for this value and let fill occur during the insertion phase.
StaticProfile	- (In) Optional argument that indicates whether or not NumIndicesPerRow should be interpreted as an exact count of nonzeros, or should be used as an approximation. By default this value is false, allowing the profile to be determined dynamically. If the user sets it to true, then the memory allocation for the Epetra_CrsGraph object will be done in one large block, saving on memory fragmentation and generally improving the performance of matrix multiplication and solve kernels.

References RowMap().

◆ Epetra_CrsGraph() [3/5]

Epetra_CrsGraph::Epetra_CrsGraph	(	Epetra_DataAccess	CV,
		const Epetra_BlockMap &	RowMap,
		const Epetra_BlockMap &	ColMap,
		const int *	NumIndicesPerRow,
		bool	StaticProfile = false )

Epetra_CrsGraph constuctor with variable number of indices per row.

Creates a Epetra_CrsGraph object and allocates storage.

Parameters

CV	- (In) A Epetra_DataAccess enumerated type set to Copy or View.
RowMap	- (In) An Epetra_BlockMap (or Epetra_Map or Epetra_LocalMap) listing the rows that this processor will contribute to.
ColMap	- (In) An Epetra_BlockMap (or Epetra_Map or Epetra_LocalMap) listing the columns that this processor will contribute to.
NumIndicesPerRow	- (In) An integer array of length NumMyRows such that NumIndicesPerRow[i] indicates the (approximate if StaticProfile=false) number of entries in the ith row.
StaticProfile	- (In) Optional argument that indicates whether or not NumIndicesPerRow should be interpreted as an exact count of nonzeros, or should be used as an approximation. By default this value is false, allowing the profile to be determined dynamically. If the user sets it to true, then the memory allocation for the Epetra_CrsGraph object will be done in one large block, saving on memory fragmentation and generally improving the performance of matrix multiplication and solve kernels.

References ColMap(), and RowMap().

◆ Epetra_CrsGraph() [4/5]

Epetra_CrsGraph::Epetra_CrsGraph	(	Epetra_DataAccess	CV,
		const Epetra_BlockMap &	RowMap,
		const Epetra_BlockMap &	ColMap,
		int	NumIndicesPerRow,
		bool	StaticProfile = false )

Epetra_CrsGraph constuctor with fixed number of indices per row.

Creates a Epetra_CrsGraph object and allocates storage.

Parameters

CV	- (In) A Epetra_DataAccess enumerated type set to Copy or View.
RowMap	- (In) An Epetra_BlockMap (or Epetra_Map or Epetra_LocalMap) listing the rows that this processor will contribute to.
ColMap	- (In) An Epetra_BlockMap (or Epetra_Map or Epetra_LocalMap) listing the columns that this processor will contribute to.
In	NumIndicesPerRow - An integer that indicates the (approximate if StaticProfile=false) number of entries in the each row. Note that it is possible to use 0 for this value and let fill occur during the insertion phase.
StaticProfile	- (In) Optional argument that indicates whether or not NumIndicesPerRow should be interpreted as an exact count of nonzeros, or should be used as an approximation. By default this value is false, allowing the profile to be determined dynamically. If the user sets it to true, then the memory allocation for the Epetra_CrsGraph object will be done in one large block, saving on memory fragmentation and generally improving the performance of matrix multiplication and solve kernels.

References ColMap(), and RowMap().

◆ Epetra_CrsGraph() [5/5]

Epetra_CrsGraph::Epetra_CrsGraph ( const Epetra_CrsGraph & Graph )

Copy constructor.

This will create a Level 1 deep copy. This Graph will share ownership of the CrsGraphData object with the right hand side Graph.

References Epetra_CrsGraph().

Member Function Documentation

◆ ColMap()

const Epetra_BlockMap & Epetra_CrsGraph::ColMap ( ) const

inline

Returns the Column Map associated with this graph.

Precondition: HaveColMap()==true

Referenced by Epetra_CrsGraph(), Epetra_CrsGraph(), Epetra_FECrsGraph::Epetra_FECrsGraph(), Epetra_FECrsGraph::Epetra_FECrsGraph(), and LCID().

◆ DataPtr()

const Epetra_CrsGraphData * Epetra_CrsGraph::DataPtr ( ) const

inline

Returns a pointer to the CrsGraphData instance this CrsGraph uses.

(Intended for developer use only for testing purposes.)

◆ DomainMap()

const Epetra_BlockMap & Epetra_CrsGraph::DomainMap ( ) const

inline

Returns the DomainMap associated with this graph.

Precondition: Filled()==true

Referenced by FillComplete(), and TransformToLocal().

◆ ExpertExtractIndexOffset()

Epetra_IntSerialDenseVector & Epetra_CrsGraph::ExpertExtractIndexOffset ( )

Returns a reference to the Epetra_IntSerialDenseVector used to hold the local IndexOffsets (CRS rowptr).

Warning: This method is intended for experts only, its use may require user code modifications in future versions of Epetra.

◆ ExpertExtractIndices()

Epetra_IntSerialDenseVector & Epetra_CrsGraph::ExpertExtractIndices ( )

Returns a reference to the Epetra_IntSerialDenseVector used to hold the local All_Indices (CRS colind).

Warning: This method is intended for experts only, its use may require user code modifications in future versions of Epetra.

◆ ExtractGlobalRowCopy()

int Epetra_CrsGraph::ExtractGlobalRowCopy	(	int	GlobalRow,
		int	LenOfIndices,
		int &	NumIndices,
		int *	Indices ) const

Extract a list of elements in a specified global row of the graph. Put into storage allocated by calling routine.

Parameters

Row	- (In) Global row number to get indices.
LenOfIndices	- (In) Length of Indices array.
NumIndices	- (Out) Number of Indices.
Indices	- (Out) Global column indices corresponding to values.

Returns: Integer error code, set to 0 if successful.

References ExtractGlobalRowCopy().

Referenced by ExtractGlobalRowCopy().

◆ ExtractGlobalRowView()

int Epetra_CrsGraph::ExtractGlobalRowView	(	int	GlobalRow,
		int &	NumIndices,
		int *&	Indices ) const

Get a view of the elements in a specified global row of the graph.

This function requires that the graph not be completed (FillComplete() was not called).

Parameters

Row	- (In) Global row number to get indices.
NumIndices	- (Out) Number of Indices.
Indices	- (Out) Global column indices corresponding to values.

Returns: Integer error code, set to 0 if successful. Returns -1 if invalid row. Returns -2 if graph is completed.

Precondition: IndicesAreLocal()==false

References ExtractGlobalRowView().

Referenced by ExtractGlobalRowView().

◆ ExtractMyRowCopy()

int Epetra_CrsGraph::ExtractMyRowCopy	(	int	LocalRow,
		int	LenOfIndices,
		int &	NumIndices,
		int *	Indices ) const

Extract a list of elements in a specified local row of the graph. Put into storage allocated by calling routine.

Parameters

Row	- (In) Local row number to get indices.
LenOfIndices	- (In) Length of Indices array.
NumIndices	- (Out) Number of Indices.
Indices	- (Out) Local column indices corresponding to values.

Returns: Integer error code, set to 0 if successful.

Precondition: IndicesAreLocal()==true

◆ ExtractMyRowView()

int Epetra_CrsGraph::ExtractMyRowView	(	int	LocalRow,
		int &	NumIndices,
		int *&	Indices ) const

Get a view of the elements in a specified local row of the graph.

This function requires that the graph be completed FillComplete() was called).

Parameters

Row	- (In) Local row number to get indices.
NumIndices	- (Out) Number of Indices.
Indices	- (Out) Column indices corresponding to values.

Returns: Integer error code, set to 0 if successful. Returns -1 if invalid row. Returns -2 if graph is not completed.

Precondition: IndicesAreLocal()==true

◆ FillComplete() [1/2]

int Epetra_CrsGraph::FillComplete ( )

Tranform to local index space. Perform other operations to allow optimal matrix operations.

This overloading of the FillComplete method assumes that the domain-map and range-map both equal the row-map, and simply calls FillComplete(RowMap(), RowMap()).

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared (i.e., if the underlying graph-data object has a reference-count greater than 1).

Postcondition: IndicesAreLocal()==true, Filled()==true

◆ FillComplete() [2/2]

int Epetra_CrsGraph::FillComplete	(	const Epetra_BlockMap &	DomainMap,
		const Epetra_BlockMap &	RangeMap )

Transform to local index space using specified Domain/Range maps. Perform other operations to allow optimal matrix operations.

Performs this sequence of operations:

Transform indices to local index space
Sort column-indices within each row
Compress out any redundant indices within rows
Compute global data such as num-nonzeros, maximum row-lengths, etc.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared (i.e., if the underlying graph-data object has a reference-count greater than 1).

Postcondition: IndicesAreLocal()==true, Filled()==true

References DomainMap(), and RangeMap().

◆ GCID()

int Epetra_CrsGraph::GCID ( int LCID_in ) const

inline

Returns the global column index for give local column index, returns IndexBase-1 if we don't have this local column.

Precondition: HaveColMap()==true (If HaveColMap()==false, returns -1)

References RowMap().

◆ GlobalMaxColDim()

int Epetra_CrsGraph::GlobalMaxColDim ( ) const

inline

Returns the max column dimension of block entries across all processors.

Precondition: Filled()==true

◆ GlobalMaxNumIndices()

int Epetra_CrsGraph::GlobalMaxNumIndices ( ) const

inline

Returns the maximun number of nonzero entries across all rows across all processors.

Precondition: Filled()==true

◆ GlobalMaxNumNonzeros()

int Epetra_CrsGraph::GlobalMaxNumNonzeros ( ) const

inline

Returns the maximun number of nonzero points across all rows across all processors.

This function returns the max over all processor of MaxNumNonzeros().

Precondition: Filled()==true

◆ GlobalMaxRowDim()

int Epetra_CrsGraph::GlobalMaxRowDim ( ) const

inline

Returns the max row dimension of block entries across all processors.

Precondition: Filled()==true

◆ HaveColMap()

bool Epetra_CrsGraph::HaveColMap ( ) const

inline

Returns true if we have a well-defined ColMap, and returns false otherwise.

Precondition: We have a well-defined ColMap if a) a ColMap was passed in at construction, or b) the MakeColMap function has been called. (Calling either of the FillComplete functions will result in MakeColMap being called.)

◆ IndexBase()

int Epetra_CrsGraph::IndexBase ( ) const

inline

Returns the index base for row and column indices for this graph.

Index base for this map.

References RowMap().

◆ InsertGlobalIndices()

int Epetra_CrsGraph::InsertGlobalIndices	(	int	GlobalRow,
		int	NumIndices,
		int *	Indices )

Enter a list of elements in a specified global row of the graph.

Parameters

Row	- (In) Global row number of indices.
NumIndices	- (In) Number of Indices.
Indices	- (In) Global column indices to insert.

Returns: Integer error code, set to 0 if successful. If the insertion requires that additional memory be allocated for the row, a positive error code of 1 is returned. If the graph is a 'View' mode graph, then a positive warning code of 2 will be returned if the specified row already exists. Returns 1 if underlying graph data is shared by multiple graph instances.

Precondition: IndicesAreGlobal()==true, StorageOptimized()==false

References InsertGlobalIndices().

Referenced by InsertGlobalIndices(), and Epetra_FECrsGraph::~Epetra_FECrsGraph().

◆ InsertMyIndices()

int Epetra_CrsGraph::InsertMyIndices	(	int	LocalRow,
		int	NumIndices,
		int *	Indices )

Enter a list of elements in a specified local row of the graph.

Parameters

Row	- (In) Local row number of indices.
NumIndices	- (In) Number of Indices.
Indices	- (In) Local column indices to insert.

Returns: Integer error code, set to 0 if successful. If the insertion requires that additional memory be allocated for the row, a positive error code of 1 is returned. If one or more of the indices is ignored (due to not being contained in the column-map), then a positive warning code of 2 is returned. If the graph is a 'View' mode graph, then a positive warning code of 3 will be returned if the specified row already exists. Returns 1 if underlying graph data is shared by multiple graph instances.

Precondition: IndicesAreLocal()==true, StorageOptimized()==false

◆ LCID()

int Epetra_CrsGraph::LCID ( int GCID_in ) const

inline

Returns the local column index for given global column index, returns -1 if no local column for this global column.

Precondition: HaveColMap()==true (If HaveColMap()==false, returns -1)

References ColMap().

Referenced by MyGCID().

◆ LowerTriangular()

bool Epetra_CrsGraph::LowerTriangular ( ) const

inline

If graph is lower triangular in local index space, this query returns true, otherwise it returns false.

Precondition: Filled()==true

◆ MaxColDim()

int Epetra_CrsGraph::MaxColDim ( ) const

inline

Returns the max column dimension of block entries on the processor.

Precondition: Filled()==true

◆ MaxNumIndices()

int Epetra_CrsGraph::MaxNumIndices ( ) const

inline

Returns the maximum number of nonzero entries across all rows on this processor.

Precondition: Filled()==true

◆ MaxNumNonzeros()

int Epetra_CrsGraph::MaxNumNonzeros ( ) const

inline

Returns the maximum number of nonzero points across all rows on this processor.

For each entry in the graph, let i = the GRID of the entry and j = the CGID of the entry. Then the entry size is the product of the rowmap elementsize of i and the colmap elementsize of i. Let ki = sum of all entry sizes for the entries in the ith row. For example, if the ith block row had 5 block entries and the element size of each entry was 4-by-4, ki would be 80. Then this function returns the max over all ki for all row on this processor.

Precondition: Filled()==true

◆ MaxRowDim()

int Epetra_CrsGraph::MaxRowDim ( ) const

inline

Returns the max row dimension of block entries on the processor.

Precondition: Filled()==true

◆ MyGCID()

bool Epetra_CrsGraph::MyGCID ( int GCID_in ) const

inline

Returns true if the GCID passed in belongs to the calling processor in this map, otherwise returns false.

Precondition: HaveColMap()==true (If HaveColMap()==false, returns -1)

References LCID().

◆ MyLCID()

bool Epetra_CrsGraph::MyLCID ( int LCID_in ) const

inline

Returns true if the LRID passed in belongs to the calling processor in this map, otherwise returns false.

Precondition: HaveColMap()==true (If HaveColMap()==false, returns -1)

◆ NoDiagonal()

bool Epetra_CrsGraph::NoDiagonal ( ) const

inline

If graph has no diagonal entries in global index space, this query returns true, otherwise it returns false.

Precondition: Filled()==true

◆ NumGlobalBlockCols()

int Epetra_CrsGraph::NumGlobalBlockCols ( ) const

inline

Returns the number of Block matrix columns in global matrix.

Precondition: Filled()==true

References RowMap().

◆ NumGlobalBlockDiagonals()

int Epetra_CrsGraph::NumGlobalBlockDiagonals ( ) const

inline

Returns the number of Block diagonal entries in the global graph, based on global row/column index comparisons.

Precondition: Filled()==true

References RowMap().

◆ NumGlobalCols()

int Epetra_CrsGraph::NumGlobalCols ( ) const

inline

Returns the number of matrix columns in global matrix.

Precondition: Filled()==true

References RowMap().

◆ NumGlobalDiagonals()

int Epetra_CrsGraph::NumGlobalDiagonals ( ) const

inline

Returns the number of diagonal entries in the global graph, based on global row/column index comparisons.

Precondition: Filled()==true

References RowMap().

◆ NumGlobalEntries()

int Epetra_CrsGraph::NumGlobalEntries ( ) const

inline

Returns the number of entries in the global graph.

Precondition: Filled()==true

References RowMap().

◆ NumGlobalNonzeros()

int Epetra_CrsGraph::NumGlobalNonzeros ( ) const

inline

Returns the number of indices in the global graph.

Note that if the graph's maps are defined such that some nonzeros appear on more than one processor, then those nonzeros will be counted more than once. If the user wishes to assemble a graph from overlapping data, they can use Epetra_FECrsGraph.

Precondition: Filled()==true

References RowMap().

◆ NumMyBlockCols()

int Epetra_CrsGraph::NumMyBlockCols ( ) const

inline

Returns the number of Block matrix columns on this processor.

Precondition: Filled()==true

◆ NumMyBlockDiagonals()

int Epetra_CrsGraph::NumMyBlockDiagonals ( ) const

inline

Returns the number of Block diagonal entries in the local graph, based on global row/column index comparisons.

Precondition: Filled()==true

◆ NumMyCols()

int Epetra_CrsGraph::NumMyCols ( ) const

inline

Returns the number of entries in the set of column-indices that appear on this processor.

The set of column-indices that appear on this processor is the union of column-indices that appear in all local rows. The size of this set isn't available until FillComplete() has been called.

Precondition: Filled()==true

◆ NumMyDiagonals()

int Epetra_CrsGraph::NumMyDiagonals ( ) const

inline

Returns the number of diagonal entries in the local graph, based on global row/column index comparisons.

Precondition: Filled()==true

◆ NumMyEntries()

int Epetra_CrsGraph::NumMyEntries ( ) const

inline

Returns the number of entries on this processor.

Precondition: Filled()==true

◆ NumMyNonzeros()

int Epetra_CrsGraph::NumMyNonzeros ( ) const

inline

Returns the number of indices in the local graph.

Precondition: Filled()==true

◆ operator=()

Epetra_CrsGraph & Epetra_CrsGraph::operator= ( const Epetra_CrsGraph & Source )

Assignment operator.

This will do a Level 1 deep copy. It will share ownership of the CrsGraphData with the right hand side Graph.

References Epetra_CrsGraph().

◆ operator[]()

int * Epetra_CrsGraph::operator[] ( int Loc )

inline

Inlined bracket operator for fast access to data. (Const and Non-const versions).

No error checking and dangerous for optimization purposes.

Parameters

Loc	(In) - Local row.

Returns: reference to pointer to locally indexed Loc row in matrix.

References StorageOptimized().

◆ OptimizeStorage()

int Epetra_CrsGraph::OptimizeStorage ( )

Make consecutive row index sections contiguous, minimize internal storage used for constructing graph.

After construction and during initialization (when indices are being added via InsertGlobalIndices() etc.), the column- indices for each row are held in a separate piece of allocated memory. This method moves the column-indices for all rows into one large contiguous array and eliminates internal storage that is not needed after graph construction. Calling this method can have a significant impact on memory costs and machine performance.

If this object was constructed in View mode then this method can't make non-contiguous indices contiguous and will return a warning code of 1 if the viewed data isn't already contiguous.

Returns: Integer error code, set to 0 if successful.

Precondition: Filled()==true.; If CV=View when the graph was constructed, then this method will be effective only if the indices of the graph were already contiguous. In this case, the indices are left untouched and internal storage for the graph is minimized.

Postcondition: StorageOptimized()==true, if successful

◆ Print()

virtual void Epetra_CrsGraph::Print ( std::ostream & os ) const

virtual

Print method.

Reimplemented from Epetra_DistObject.

◆ RangeMap()

const Epetra_BlockMap & Epetra_CrsGraph::RangeMap ( ) const

inline

Returns the RangeMap associated with this graph.

Precondition: Filled()==true

Referenced by FillComplete(), and TransformToLocal().

◆ ReferenceCount()

int Epetra_CrsGraph::ReferenceCount ( ) const

inline

Returns the reference count of CrsGraphData.

(Intended for testing purposes.)

◆ RemoveEmptyProcessesInPlace()

int Epetra_CrsGraph::RemoveEmptyProcessesInPlace ( const Epetra_BlockMap * NewMap )

Remove processes owning zero rows from the Maps and their communicator.

Warning: This method is ONLY for use by experts.; We make NO promises of backwards compatibility. This method may change or disappear at any time.

Parameters

newMap [in] This must be the result of calling the removeEmptyProcesses() method on the row BlockMap. If it is not, this method's behavior is undefined. This pointer will be null on excluded processes.

◆ RemoveGlobalIndices() [1/2]

int Epetra_CrsGraph::RemoveGlobalIndices	(	int	GlobalRow,
		int	NumIndices,
		int *	Indices )

Remove a list of elements from a specified global row of the graph.

Parameters

Row	- (In) Global row number of indices.
NumIndices	- (In) Number of Indices.
Indices	- (In) Global column indices to remove.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared.

Precondition: IndicesAreGlobal()==true, StorageOptimized()==false

References RemoveGlobalIndices().

Referenced by RemoveGlobalIndices().

◆ RemoveGlobalIndices() [2/2]

int Epetra_CrsGraph::RemoveGlobalIndices ( long long Row )

Remove all indices from a specified global row of the graph.

Parameters

Row	- (In) Global row number of indices.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared.

Precondition: IndicesAreGlobal()==true, StorageOptimized()==false

◆ RemoveMyIndices() [1/2]

int Epetra_CrsGraph::RemoveMyIndices	(	int	LocalRow,
		int	NumIndices,
		int *	Indices )

Remove a list of elements from a specified local row of the graph.

Parameters

Row	- (In) Local row number of indices.
NumIndices	- (In) Number of Indices.
Indices	- (In) Local column indices to remove.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared.

Precondition: IndicesAreLocal()==true, StorageOptimized()==false

◆ RemoveMyIndices() [2/2]

int Epetra_CrsGraph::RemoveMyIndices ( int Row )

Remove all indices from a specified local row of the graph.

Parameters

Row	- (In) Local row number of indices.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared.

Precondition: IndicesAreLocal()==true, StorageOptimized()==false

◆ RemoveRedundantIndices()

int Epetra_CrsGraph::RemoveRedundantIndices ( )

protected

Removes any redundant column indices in the rows of the graph.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared.

◆ ReplaceColMap()

int Epetra_CrsGraph::ReplaceColMap ( const Epetra_BlockMap & newmap )

Replaces the current ColMap with the user-specified map object, but only if no entries have been inserted into the graph yet (both IndicesAreLocal() and IndicesAreGlobal() are false) or currentmap->PointSameAs(newmap) is true. This is a collective function. Returns 0 if map is replaced, -1 if not.

Precondition: (IndicesAreLocal()==false && IndicesAreGlobal()==false) || ColMap().PointSameAs(newmap)==true

◆ ReplaceDomainMapAndImporter()

int Epetra_CrsGraph::ReplaceDomainMapAndImporter	(	const Epetra_BlockMap &	NewDomainMap,
		const Epetra_Import *	NewImporter )

Replaces the current DomainMap & Importer with the user-specified map object.

Replaces the current DomainMap and Importer with the user-specified map object, but only if the matrix has been FillCompleted, Importer's TargetMap matches the ColMap and Importer's SourceMap matches the DomainMap (assuming the importer isn't null). If an Importer is passed in, Epetra_CrsMatrix will copy it. Returns 0 if map/importer is replaced, -1 if not.

Precondition: (!NewImporter && ColMap().PointSameAs(NewDomainMap)) || (NewImporter && ColMap().PointSameAs(NewImporter->TargetMap()) && NewDomainMap.PointSameAs(NewImporter->SourceMap()))

◆ ReplaceRowMap()

int Epetra_CrsGraph::ReplaceRowMap ( const Epetra_BlockMap & newmap )

Replaces the current RowMap with the user-specified map object, but only if currentmap->PointSameAs(newmap) is true. This is a collective function. Returns 0 if map is replaced, -1 if not.

Precondition: RowMap().PointSameAs(newmap)==true

◆ SortGhostsAssociatedWithEachProcessor()

void Epetra_CrsGraph::SortGhostsAssociatedWithEachProcessor ( bool Flag )

inline

Forces FillComplete() to locally order ghostnodes associated with each remote processor in ascending order.

To be compliant with AztecOO, FillComplete() already locally orders ghostnodes such that information received from processor k has a lower local numbering than information received from processor j if k is less than j. SortGhostsAssociatedWithEachProcessor(True) further forces FillComplete() to locally number all ghostnodes received from processor k in ascending order. That is, the local numbering of b is less than c if the global numbering of b is less than c and if both b and c are owned by the same processor. This is done to be compliant with some limited block features within ML. In particular, some ML features require that a block structure of the matrix be maintained even within the ghost variables.

◆ SortIndices()

int Epetra_CrsGraph::SortIndices ( )

protected

Sort column indices, row-by-row, in ascending order.

Returns: Integer error code, set to 0 if successful. Returns 1 if data is shared.

◆ UpperTriangular()

bool Epetra_CrsGraph::UpperTriangular ( ) const

inline

If graph is upper triangular in local index space, this query returns true, otherwise it returns false.

Precondition: Filled()==true

The documentation for this class was generated from the following file:

Epetra_CrsGraph.h

Public Member Functions

Expert Users and Developers Only

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ Epetra_CrsGraph() [1/5]

◆ Epetra_CrsGraph() [2/5]

◆ Epetra_CrsGraph() [3/5]

◆ Epetra_CrsGraph() [4/5]

◆ Epetra_CrsGraph() [5/5]

Member Function Documentation

◆ ColMap()

◆ DataPtr()

◆ DomainMap()

◆ ExpertExtractIndexOffset()

◆ ExpertExtractIndices()

◆ ExtractGlobalRowCopy()

◆ ExtractGlobalRowView()

◆ ExtractMyRowCopy()

◆ ExtractMyRowView()

◆ FillComplete() [1/2]

◆ FillComplete() [2/2]

◆ GCID()

◆ GlobalMaxColDim()

◆ GlobalMaxNumIndices()

◆ GlobalMaxNumNonzeros()

◆ GlobalMaxRowDim()

◆ HaveColMap()

◆ IndexBase()

◆ InsertGlobalIndices()

◆ InsertMyIndices()

◆ LCID()

◆ LowerTriangular()

◆ MaxColDim()

◆ MaxNumIndices()

◆ MaxNumNonzeros()

◆ MaxRowDim()

◆ MyGCID()

◆ MyLCID()

◆ NoDiagonal()

◆ NumGlobalBlockCols()

◆ NumGlobalBlockDiagonals()

◆ NumGlobalCols()

◆ NumGlobalDiagonals()

◆ NumGlobalEntries()

◆ NumGlobalNonzeros()

◆ NumMyBlockCols()

◆ NumMyBlockDiagonals()

◆ NumMyCols()

◆ NumMyDiagonals()

◆ NumMyEntries()

◆ NumMyNonzeros()

◆ operator=()

◆ operator[]()

◆ OptimizeStorage()

◆ Print()

◆ RangeMap()

◆ ReferenceCount()

◆ RemoveEmptyProcessesInPlace()

◆ RemoveGlobalIndices() [1/2]

◆ RemoveGlobalIndices() [2/2]

◆ RemoveMyIndices() [1/2]

◆ RemoveMyIndices() [2/2]

◆ RemoveRedundantIndices()

◆ ReplaceColMap()

◆ ReplaceDomainMapAndImporter()

◆ ReplaceRowMap()

◆ SortGhostsAssociatedWithEachProcessor()

◆ SortIndices()

◆ UpperTriangular()