globalMeshData.H

Go to the documentation of this file.

/*---------------------------------------------------------------------------*\
  =========                 |
  \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox
   \\    /   O peration     |
    \\  /    A nd           | Copyright (C) 1991-2010 OpenCFD Ltd.
     \\/     M anipulation  |
-------------------------------------------------------------------------------
License
    This file is part of OpenFOAM.

    OpenFOAM is free software: you can redistribute it and/or modify it
    under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.

    OpenFOAM is distributed in the hope that it will be useful, but WITHOUT
    ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
    FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
    for more details.

    You should have received a copy of the GNU General Public License
    along with OpenFOAM.  If not, see <http://www.gnu.org/licenses/>.

Class
    Foam::globalMeshData

Description
    Various mesh related information for a parallel run. Upon construction
    constructs all info by using parallel communication. 

    Requires: 
    - all processor patches to have correct ordering.
    - all processorPatches to have their transforms set ('makeTransforms')

    The shared point addressing is quite interesting. It gives on each processor
    the vertices that cannot be set using a normal swap on processor patches.
    These are the vertices that are shared between more than 2 processors.

    There is an issue with these shared vertices if they originate from
    cyclics (i.e. are now separated processor patches). They will all be
    mapped to the same global point (so even though the processor points are
    not on the same location) since topologically they are one and the same.

    So if you ask for sharedPoints() you get only one of the coordinates of
    the topologically shared points.

    All the hard work of these shared points is done by the globalPoints class.

    Shared edges: similar to shared points gives on all processors the edges
    that are shared between more than two patches (i.e. the edges on which
    data cannot be synchronized by a straightforward edge data swap). Note
    that shared edges will use shared points but not all edges between shared
    points need to be shared edges (e.g. there might be an edge connecting
    two disconnected regions of shared points).

    Currently an edge is considered shared
    if it uses two shared points and is used more than once. This is not
    correct on processor patches but it only slightly overestimates the number
    of shared edges. Doing full analysis of how many patches use the edge
    would be too complicated.

    Shared edge calculation is demand driven so always make sure to have
    your first call to one of the access functions synchronous amongst all
    processors!


SourceFiles
    globalMeshData.C
    globalMeshDataMorph.C

\*---------------------------------------------------------------------------*/

#ifndef globalMeshData_H
#define globalMeshData_H

#include <OpenFOAM/polyMesh.H>
#include <OpenFOAM/Switch.H>
#include "processorTopology.H"
#include <OpenFOAM/labelPair.H>

// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

namespace Foam
{

// Forward declaration of friend functions and operators

class globalMeshData;
Ostream& operator<<(Ostream&, const globalMeshData&);


/*---------------------------------------------------------------------------*\
                           Class globalMeshData Declaration
\*---------------------------------------------------------------------------*/

class globalMeshData
:
    public processorTopology
{

    // Private class

        // To combineReduce a pointField. Just appends all lists.
        template <class T>
        class plusEqOp
        {

        public:

            void operator()(T& x, const T& y) const
            {
                label n = x.size();

                x.setSize(x.size() + y.size());

                forAll(y, i)
                {
                    x[n++] = y[i];
                }
            }
        };


        typedef List<labelPair> labelPairList;


    // Private data

        //- Reference to mesh
        const polyMesh& mesh_;


        // Data related to the complete mesh

            //- Bounding box of complete mesh
            boundBox bb_;

            //- Total number of points in the complete mesh
            label nTotalPoints_;

            //- Total number of faces in the complete mesh
            label nTotalFaces_;

            //- Total number of cells in the complete mesh
            label nTotalCells_;


        // Processor patch addressing (be careful if not running in parallel!)

            //- List of processor patch labels 
            //  (size of list = number of processor patches)
            labelList processorPatches_;

            //- List of indices into processorPatches_ for each patch.
            //  Index = -1 for non-processor patches.
            //  (size of list = number of patches)
            labelList processorPatchIndices_;

            //- processorPatchIndices_ of the neighbours processor patches
            labelList processorPatchNeighbours_;


        // Globally shared point addressing

            //- Total number of global points
            label nGlobalPoints_;

            //- Indices of local points that are globally shared
            labelList sharedPointLabels_;

            //- Indices of globally shared points in the master list
            //  This list contains all the shared points in the mesh
            labelList sharedPointAddr_;

            //- Shared point global labels.
            //  Global point index for every local shared point.
            //  Only valid if constructed with this information or if
            //  pointProcAddressing read.
            mutable labelList* sharedPointGlobalLabelsPtr_;


        // Globally shared edge addressing. Derived from shared points.
        // All demand driven since don't want to construct edges always.

            //- Total number of global edges
            mutable label nGlobalEdges_;

            //- Indices of local edges that are globally shared
            mutable labelList* sharedEdgeLabelsPtr_;

            //- Indices of globally shared edge in the master list
            //  This list contains all the shared edges in the mesh
            mutable labelList* sharedEdgeAddrPtr_;


    // Private Member Functions

        //- Set up processor patch addressing
        void initProcAddr();

        //- Helper function for shared edge addressing
        static void countSharedEdges
        (
            const HashTable<labelList, edge, Hash<edge> >& procSharedEdges,
            HashTable<label, edge, Hash<edge> >&,
            label&
        );

        //- Calculate shared edge addressing
        void calcSharedEdges() const;

        //- Count coincident faces.
        static label countCoincidentFaces
        (
            const scalar tolDim,
            const vectorField& separationDist
        );

        //- Disallow default bitwise copy construct
        globalMeshData(const globalMeshData&);

        //- Disallow default bitwise assignment
        void operator=(const globalMeshData&);


public:

    //- Runtime type information
    ClassName("globalMeshData");


    // Static data members

        //- Geomtric tolerance (fraction of bounding box)
        static const Foam::scalar matchTol_;


    // Constructors

        //- Construct from mesh, derive rest (does parallel communication!)
        globalMeshData(const polyMesh& mesh);

        //- Old behaviour: read constructor given IOobject and a polyMesh
        //  reference. Only use this for testing!
        globalMeshData(const IOobject& io, const polyMesh& mesh);


    // Destructor

        ~globalMeshData();

        //- Remove all demand driven data
        void clearOut();


    // Member Functions

        // Access

            //- Return the mesh reference
            const polyMesh& mesh() const
            {
                return mesh_;
            }

            //- Does the mesh contain processor patches? (also valid when
            //  not running parallel)
            bool parallel() const
            {
                return processorPatches_.size() > 0;
            }

            const boundBox& bb() const
            {
                return bb_;
            }

            //- Return total number of points in decomposed mesh
            label nTotalPoints() const
            {
                return nTotalPoints_;
            }

            //- Return total number of faces in decomposed mesh
            label nTotalFaces() const
            {
                return nTotalFaces_;
            }

            //- Return total number of cells in decomposed mesh
            label nTotalCells() const
            {
                return nTotalCells_;
            }


        // Processor patch addressing (be careful when not running in parallel)

            //- Return list of processor patch labels 
            //  (size of list = number of processor patches)
            const labelList& processorPatches() const
            {
                return processorPatches_;
            }

            //- Return list of indices into processorPatches_ for each patch.
            //  Index = -1 for non-processor parches.
            //  (size of list = number of patches)
            const labelList& processorPatchIndices() const
            {
                return processorPatchIndices_;
            }

            //- Return processorPatchIndices of the neighbours
            //  processor patches. -1 if not running parallel.
            const labelList& processorPatchNeighbours() const
            {
                return processorPatchNeighbours_;
            }


        // Globally shared point addressing

            //- Return number of globally shared points
            label nGlobalPoints() const
            {
                return nGlobalPoints_;
            }

            //- Return indices of local points that are globally shared
            const labelList& sharedPointLabels() const
            {
                return sharedPointLabels_;
            }

            //- Return addressing into the complete globally shared points
            //  list
            //  Note: It is assumed that a (never constructed) complete
            //  list of globally shared points exists.  The set of shared
            //  points on the current processor is a subset of all shared
            //  points. Shared point addressing gives the index in the
            //  list of all globally shared points for each of the locally
            //  shared points.
            const labelList& sharedPointAddr() const
            {
                return sharedPointAddr_;
            }

            //- Return shared point global labels. Tries to read
            //  'pointProcAddressing' and returns list or -1 if none
            //  available.
            const labelList& sharedPointGlobalLabels() const;

            //- Collect coordinates of shared points on all processors.
            //  (does parallel communication!)
            //  Note: not valid for cyclicParallel since shared cyclic points
            //  are merged into single global point. (use geometricSharedPoints
            //  instead)
            pointField sharedPoints() const;

            //- Like sharedPoints but keeps cyclic points separate.
            //  (does geometric merging; uses matchTol_*bb as merging tolerance)
            //  Use sharedPoints() instead.
            pointField geometricSharedPoints() const;



        // Globally shared edge addressing

            //- Return number of globally shared edges. Demand-driven
            //  calculation so call needs to be synchronous among processors!
            label nGlobalEdges() const;

            //- Return indices of local edges that are globally shared.
            //  Demand-driven
            //  calculation so call needs to be synchronous among processors!
            const labelList& sharedEdgeLabels() const;

            //- Return addressing into the complete globally shared edge
            //  list. The set of shared
            //  edges on the current processor is a subset of all shared
            //  edges. Shared edge addressing gives the index in the
            //  list of all globally shared edges for each of the locally
            //  shared edges.
            //  Demand-driven
            //  calculation so call needs to be synchronous among processors!
            const labelList& sharedEdgeAddr() const;


        // Edit

            //- Update for moving points.
            void movePoints(const pointField& newPoints);

            //- Change global mesh data given a topological change. Does a
            //  full parallel analysis to determine shared points and
            //  boundaries.
            void updateMesh();


        // Write

            bool write() const;


    // Ostream Operator

        friend Ostream& operator<<(Ostream&, const globalMeshData&);
};


// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

} // End namespace Foam

// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

#endif

// ************************ vim: set sw=4 sts=4 et: ************************ //