EpetraExt_CrsSingletonFilter_LinearProblem.cpp

Go to the documentation of this file.

// @HEADER
// ***********************************************************************
// 
//     EpetraExt: Epetra Extended - Linear Algebra Services Package
//                 Copyright (2001) Sandia Corporation
// 
// Under terms of Contract DE-AC04-94AL85000, there is a non-exclusive
// license for use of this work by or on behalf of the U.S. Government.
// 
// This library is free software; you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as
// published by the Free Software Foundation; either version 2.1 of the
// License, or (at your option) any later version.
//  
// This library 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
// Lesser General Public License for more details.
//  
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
// USA
// Questions? Contact Michael A. Heroux (maherou@sandia.gov) 
// 
// ***********************************************************************
// @HEADER

#include "Epetra_Map.h"
#include "Epetra_Util.h"
#include "Epetra_Export.h"
#include "Epetra_Import.h"
#include "Epetra_MultiVector.h"
#include "Epetra_Vector.h"
#include "Epetra_IntVector.h"
#include "Epetra_Comm.h"
#include "Epetra_LinearProblem.h"
#include "Epetra_MapColoring.h"
#include "EpetraExt_CrsSingletonFilter_LinearProblem.h"

namespace EpetraExt {

//==============================================================================
LinearProblem_CrsSingletonFilter::
LinearProblem_CrsSingletonFilter( bool verbose )
: verbose_(verbose)
{
  InitializeDefaults();
}
//==============================================================================
LinearProblem_CrsSingletonFilter::
~LinearProblem_CrsSingletonFilter()
{
  if (ReducedProblem_!=0) delete ReducedProblem_;
  if (ReducedMatrix_!=0) delete ReducedMatrix_;
  if (ReducedLHS_!=0) delete ReducedLHS_;
  if (ReducedRHS_!=0) delete ReducedRHS_;
  if (ReducedMatrixDomainMap_!=ReducedMatrixColMap_) delete ReducedMatrixDomainMap_;
  if (OrigReducedMatrixDomainMap_!=ReducedMatrixColMap_ &&
      OrigReducedMatrixDomainMap_!=0) delete OrigReducedMatrixDomainMap_;
  if (ReducedMatrixRangeMap_!=ReducedMatrixRowMap_) delete ReducedMatrixRangeMap_;
  if (ReducedMatrixRowMap_!=0) delete ReducedMatrixRowMap_;
  if (ReducedMatrixColMap_!=0) delete ReducedMatrixColMap_;
  if (Full2ReducedRHSImporter_!=0) delete Full2ReducedRHSImporter_;
  if (Full2ReducedLHSImporter_!=0) delete Full2ReducedLHSImporter_;
  if (RedistributeDomainExporter_!=0) delete RedistributeDomainExporter_;
  if (RowMapColors_!=0) delete RowMapColors_;
  if (ColMapColors_!=0) delete ColMapColors_;

  if (ColSingletonRowLIDs_ != 0) delete [] ColSingletonRowLIDs_;
  if (ColSingletonColLIDs_ != 0) delete [] ColSingletonColLIDs_;
  if (ColSingletonPivotLIDs_ != 0) delete [] ColSingletonPivotLIDs_;
  if (ColSingletonPivots_ != 0) delete [] ColSingletonPivots_;
  if (tempExportX_ != 0) delete tempExportX_;
  if (Indices_ != 0) delete [] Indices_;
  if (tempX_ != 0) delete tempX_;
  if (tempB_ != 0) delete tempB_;

}

//==============================================================================
LinearProblem_CrsSingletonFilter::NewTypeRef
LinearProblem_CrsSingletonFilter::
operator()( LinearProblem_CrsSingletonFilter::OriginalTypeRef orig )
{
  analyze( orig );
  return construct();
}

//==============================================================================
bool
LinearProblem_CrsSingletonFilter::
analyze( LinearProblem_CrsSingletonFilter::OriginalTypeRef orig )
{
  origObj_ = &orig;

  FullMatrix_ = orig.GetMatrix();

  int flag = Analyze( FullMatrix_ );
  assert( flag >= 0 );

  if ( verbose_ && FullMatrix_->Comm().MyPID()==0 ) {
    cout << "\nAnalyzed Singleton Problem:\n";
    cout << "---------------------------\n";
  }
  if ( SingletonsDetected() ) {
    if ( verbose_ && FullMatrix_->Comm().MyPID()==0 ) {
      cout << "Singletons Detected!" << endl;;
      cout << "Num Singletons:      " << NumSingletons() << endl;
    }
  }
  else {
    if ( verbose_ && FullMatrix_->Comm().MyPID()==0 ) 
        cout << "No Singletons Detected!" << endl;
  }
/*
    cout << "List of Row Singletons:\n";
    int * slist = RowMapColors_->ColorLIDList(1);
    for( int i = 0; i < RowMapColors_->NumElementsWithColor(1); ++i )
      cout << slist[i] << " ";
    cout << "\n";
    cout << "List of Col Singletons:\n";
    slist = ColMapColors_->ColorLIDList(1);
    for( int i = 0; i < ColMapColors_->NumElementsWithColor(1); ++i )
      cout << slist[i] << " ";
    cout << "\n";
*/
  if ( verbose_ && FullMatrix_->Comm().MyPID()==0 )
    cout << "---------------------------\n\n";
 
  return true;
}

//==============================================================================
LinearProblem_CrsSingletonFilter::NewTypeRef
LinearProblem_CrsSingletonFilter::
construct()
{
  if( !origObj_ ) abort();

  int flag = ConstructReducedProblem( origObj_ );
  assert( flag >= 0 );

  newObj_ = ReducedProblem();

  if( verbose_ && FullMatrix_->Comm().MyPID()==0 )
  {
    cout << "\nConstructedSingleton Problem:\n";
    cout << "---------------------------\n";
    cout << "RatioOfDimensions:   " << RatioOfDimensions() << endl;
    cout << "RatioOfNonzeros:     " << RatioOfNonzeros() << endl;
    cout << "---------------------------\n\n";
  }

  return *newObj_;
}

//==============================================================================
bool LinearProblem_CrsSingletonFilter::fwd()
{
  int ierr = UpdateReducedProblem( FullProblem_ );
  if( ierr ) cout << "EDT_LinearProblem_CrsSingletonFilter::UpdateReducedProblem FAILED!\n";

  return (ierr==0);
}

//==============================================================================
bool LinearProblem_CrsSingletonFilter::rvs()
{
  int ierr = ComputeFullSolution();
  if( ierr ) cout << "EDT_LinearProblem_CrsSingletonFilter::ComputeFullSolution FAILED!\n";

  return (ierr==0);
}

//==============================================================================
void LinearProblem_CrsSingletonFilter::InitializeDefaults() { 

// Initialize all attributes that have trivial default values

  FullProblem_ = 0;
  ReducedProblem_ = 0;
  FullMatrix_ = 0;
  ReducedMatrix_ = 0;
  ReducedRHS_ = 0;
  ReducedLHS_ = 0;
  ReducedMatrixRowMap_ = 0;
  ReducedMatrixColMap_ = 0;
  ReducedMatrixDomainMap_ = 0;
  ReducedMatrixRangeMap_ = 0;
  OrigReducedMatrixDomainMap_ = 0;
  Full2ReducedRHSImporter_ = 0;
  Full2ReducedLHSImporter_ = 0;
  RedistributeDomainExporter_ = 0;

  ColSingletonRowLIDs_ = 0;
  ColSingletonColLIDs_ = 0;
  ColSingletonPivotLIDs_ = 0;
  ColSingletonPivots_ = 0;

  AbsoluteThreshold_ = 0;
  RelativeThreshold_ = 0;

  NumMyRowSingletons_ = -1;
  NumMyColSingletons_ = -1;
  NumGlobalRowSingletons_ = -1;
  NumGlobalColSingletons_ = -1;
  RatioOfDimensions_ = -1.0;
  RatioOfNonzeros_ = -1.0;

  HaveReducedProblem_ = false;
  UserDefinedEliminateMaps_ = false;
  AnalysisDone_ = false;
  SymmetricElimination_ = true;

  tempExportX_ = 0;
  tempX_ = 0;
  tempB_ = 0;

  Indices_ = 0;

  RowMapColors_ = 0;
  ColMapColors_ = 0;

  FullMatrixIsCrsMatrix_ = false;
  MaxNumMyEntries_ = 0;
  return;
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::Analyze(Epetra_RowMatrix * FullMatrix) {

  int i, j, jj;

  FullMatrix_ = FullMatrix; 

  if (AnalysisDone_) EPETRA_CHK_ERR(-1); // Analysis already done once.  Cannot do it again
  if (FullMatrix==0) EPETRA_CHK_ERR(-2); // Input matrix pointer is zero
  if (FullMatrix->NumGlobalRows()==0) EPETRA_CHK_ERR(-3); // Full matrix has zero dimension.
  if (FullMatrix->NumGlobalNonzeros()==0) EPETRA_CHK_ERR(-4); // Full matrix has no nonzero terms.

//  cout << "RowMAP\n";
//  cout << FullMatrixRowMap();
//  cout << "ColMAP\n";
//  cout << FullMatrixColMap();

  // First check for columns with single entries and find columns with singleton rows
  Epetra_IntVector ColProfiles(FullMatrixColMap()); ColProfiles.PutValue(0);
  Epetra_IntVector ColHasRowWithSingleton(FullMatrixColMap()); ColHasRowWithSingleton.PutValue(0);

  // RowIDs[j] will contain the local row ID associated with the jth column, 
  // if the jth col has a single entry
  Epetra_IntVector RowIDs(FullMatrixColMap()); RowIDs.PutValue(-1);

  // Define MapColoring objects
  RowMapColors_ = new Epetra_MapColoring(FullMatrixRowMap());  // Initial colors are all 0
  ColMapColors_ = new Epetra_MapColoring(FullMatrixColMap());
  Epetra_MapColoring & RowMapColors = *RowMapColors_;
  Epetra_MapColoring & ColMapColors = *ColMapColors_;


  int NumMyRows = FullMatrix->NumMyRows();
  int NumMyCols = FullMatrix->NumMyCols();

  // Set up for accessing full matrix.  Will do so row-by-row.
  EPETRA_CHK_ERR(InitFullMatrixAccess()); 

  // Scan matrix for singleton rows, build up column profiles
  int NumIndices;
  int * Indices;
  NumMyRowSingletons_ = 0;
  for (i=0; i<NumMyRows; i++) {
    // Get ith row
    EPETRA_CHK_ERR(GetRow(i, NumIndices, Indices));
    for (j=0; j<NumIndices; j++) {
      int ColumnIndex = Indices[j];
      ColProfiles[ColumnIndex]++; // Increment column count
      // Record local row ID for current column
      // will use to identify row to eliminate if column is a singleton
      RowIDs[ColumnIndex] = i;
    }
    // If row has single entry, color it and associated column with color=1
    if (NumIndices==1) {
      int j = Indices[0];
      ColHasRowWithSingleton[j]++;
      RowMapColors[i] = 1;
      ColMapColors[j] = 1;
      NumMyRowSingletons_++;
    }
  }

  // 1) The vector ColProfiles has column nonzero counts for each processor's contribution
  // Combine these to get total column profile information and then redistribute to processors
  // so each can determine if it is the owner of the row associated with the singleton column
  // 2) The vector ColHasRowWithSingleton[i] contain count of singleton rows  that are associated with 
  // the ith column on this processor.  Must tell other processors that they should also eliminate 
  // these columns.

  // Make a copy of ColProfiles for later use when detecting columns that disappear locally

  Epetra_IntVector NewColProfiles(ColProfiles);

  // If RowMatrixImporter is non-trivial, we need to perform a gather/scatter to accumulate results

  if (FullMatrix->RowMatrixImporter()!=0) {
    Epetra_IntVector tmpVec(FullMatrixDomainMap()); // Use for gather/scatter of column vectors
    EPETRA_CHK_ERR(tmpVec.PutValue(0));
    EPETRA_CHK_ERR(tmpVec.Export(ColProfiles, *FullMatrix->RowMatrixImporter(), Add));
    EPETRA_CHK_ERR(ColProfiles.Import(tmpVec, *FullMatrix->RowMatrixImporter(), Insert));
    
    EPETRA_CHK_ERR(tmpVec.PutValue(0));
    EPETRA_CHK_ERR(tmpVec.Export(ColHasRowWithSingleton, *FullMatrix->RowMatrixImporter(), Add));
    EPETRA_CHK_ERR(ColHasRowWithSingleton.Import(tmpVec, *FullMatrix->RowMatrixImporter(), Insert));
  }
  // ColProfiles now contains the nonzero column entry count for all columns that have
  // an entry on this processor.
  // ColHasRowWithSingleton now contains a count of singleton rows associated with the corresponding
  // local column.  Next we check to make sure no column is associated with more than one singleton row.

  if (ColHasRowWithSingleton.MaxValue()>1) {
    EPETRA_CHK_ERR(-2); // At least one col is associated with two singleton rows, can't handle it.
  }

  Epetra_IntVector RowHasColWithSingleton(FullMatrix->RowMatrixRowMap()); // Use to check for errors
  RowHasColWithSingleton.PutValue(0);
 
  NumMyColSingletons_ = 0;
  // Count singleton columns (that were not already counted as singleton rows)
  for (j=0; j<NumMyCols; j++) {
    int i = RowIDs[j];
    // Check if column is a singleton
    if (ColProfiles[j]==1 ) {
      // Check to make sure RowID is not invalid
//      assert(i!=-1);
      // Check to see if this column already eliminated by the row check above
      if (RowMapColors[i]!=1) {
  RowHasColWithSingleton[i]++; // Increment col singleton counter for ith row
  RowMapColors[i] = 2; // Use 2 for now, to distinguish between row eliminated directly or via column singletons
  ColMapColors[j] = 1;
  NumMyColSingletons_++;
  // If we delete a row, we need to keep track of associated column entries that were also deleted 
  // in case all entries in a column are eventually deleted, in which case the column should
  // also be deleted.
  EPETRA_CHK_ERR(GetRow(i, NumIndices, Indices));
  for (jj=0; jj<NumIndices; jj++) NewColProfiles[Indices[jj]]--;
  
      }
    }
    // Check if some other processor eliminated this column    
    else if (ColHasRowWithSingleton[j]==1 && RowMapColors[i]!=1) { 
  ColMapColors[j] = 1;
    }
  }
  if (RowHasColWithSingleton.MaxValue()>1) {
    EPETRA_CHK_ERR(-3); // At lease one row is associated with two singleton cols, can't handle it.
  }

 // Generate arrays that keep track of column singleton row, col and pivot info needed for post-solve phase
  EPETRA_CHK_ERR(CreatePostSolveArrays(RowIDs, RowMapColors, ColProfiles, NewColProfiles,
               ColHasRowWithSingleton));

  for (i=0; i<NumMyRows; i++) if (RowMapColors[i]==2) RowMapColors[i] = 1; // Convert all eliminated rows to same color

  FullMatrix->RowMatrixRowMap().Comm().SumAll(&NumMyRowSingletons_, &NumGlobalRowSingletons_, 1);
  FullMatrix->RowMatrixRowMap().Comm().SumAll(&NumMyColSingletons_, &NumGlobalColSingletons_, 1);
  AnalysisDone_ = true;
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::ConstructReducedProblem(Epetra_LinearProblem * Problem) {

  int i, j;
  if (HaveReducedProblem_) EPETRA_CHK_ERR(-1); // Setup already done once.  Cannot do it again
  if (Problem==0) EPETRA_CHK_ERR(-2); // Null problem pointer

  FullProblem_ = Problem;
  FullMatrix_ = dynamic_cast<Epetra_RowMatrix *>(Problem->GetMatrix());
  if (FullMatrix_==0) EPETRA_CHK_ERR(-3); // Need a RowMatrix
  if (Problem->GetRHS()==0) EPETRA_CHK_ERR(-4); // Need a RHS
  if (Problem->GetLHS()==0) EPETRA_CHK_ERR(-5); // Need a LHS
  // Generate reduced row and column maps

  Epetra_MapColoring & RowMapColors = *RowMapColors_;
  Epetra_MapColoring & ColMapColors = *ColMapColors_;

  ReducedMatrixRowMap_ = RowMapColors.GenerateMap(0);
  ReducedMatrixColMap_ = ColMapColors.GenerateMap(0);

  // Create domain and range map colorings by exporting map coloring of column and row maps

  if (FullMatrix()->RowMatrixImporter()!=0) {
    Epetra_MapColoring DomainMapColors(FullMatrixDomainMap());
    EPETRA_CHK_ERR(DomainMapColors.Export(*ColMapColors_, *FullMatrix()->RowMatrixImporter(), AbsMax));
    OrigReducedMatrixDomainMap_ = DomainMapColors.GenerateMap(0);
  }
  else
    OrigReducedMatrixDomainMap_ = ReducedMatrixColMap_;

  if (FullMatrixIsCrsMatrix_) {
    if (FullCrsMatrix()->Exporter()!=0) { // Non-trivial exporter
      Epetra_MapColoring RangeMapColors(FullMatrixRangeMap());
      EPETRA_CHK_ERR(RangeMapColors.Export(*RowMapColors_, *FullCrsMatrix()->Exporter(), 
             AbsMax));
      ReducedMatrixRangeMap_ = RangeMapColors.GenerateMap(0);
    }
    else
      ReducedMatrixRangeMap_ = ReducedMatrixRowMap_;
  }
  else
    ReducedMatrixRangeMap_ = ReducedMatrixRowMap_;

  // Check to see if the reduced system domain and range maps are the same.
  // If not, we need to remap entries of the LHS multivector so that they are distributed
  // conformally with the rows of the reduced matrix and the RHS multivector
  SymmetricElimination_ = ReducedMatrixRangeMap_->SameAs(*OrigReducedMatrixDomainMap_);
  if (!SymmetricElimination_) 
    ConstructRedistributeExporter(OrigReducedMatrixDomainMap_, ReducedMatrixRangeMap_, 
          RedistributeDomainExporter_, ReducedMatrixDomainMap_);
  else {
    ReducedMatrixDomainMap_ = OrigReducedMatrixDomainMap_;
    OrigReducedMatrixDomainMap_ = 0;
    RedistributeDomainExporter_ = 0;
  }
  
  // Create pointer to Full RHS, LHS
  Epetra_MultiVector * FullRHS = FullProblem()->GetRHS();
  Epetra_MultiVector * FullLHS = FullProblem()->GetLHS();
  int NumVectors = FullLHS->NumVectors();

  // Create importers
//  cout << "RedDomainMap\n";
//  cout << *ReducedMatrixDomainMap();
//  cout << "FullDomainMap\n";
//  cout << FullMatrixDomainMap();
  Full2ReducedLHSImporter_ = new Epetra_Import(*ReducedMatrixDomainMap(), FullMatrixDomainMap());
//  cout << "RedRowMap\n";
//  cout << *ReducedMatrixRowMap();
//  cout << "FullRHSMap\n";
//  cout << FullRHS->Map();
  Full2ReducedRHSImporter_ = new Epetra_Import(*ReducedMatrixRowMap(), FullRHS->Map());

  // Construct Reduced Matrix
  ReducedMatrix_ = new Epetra_CrsMatrix(Copy, *ReducedMatrixRowMap(), *ReducedMatrixColMap(), 0);

  // Create storage for temporary X values due to explicit elimination of rows
  tempExportX_ = new Epetra_MultiVector(FullMatrixColMap(), NumVectors);

  int NumEntries;
  int * Indices;
  double * Values;
  int NumMyRows = FullMatrix()->NumMyRows();
  int ColSingletonCounter = 0;
  for (i=0; i<NumMyRows; i++) {
    int curGRID = FullMatrixRowMap().GID(i);
    if (ReducedMatrixRowMap()->MyGID(curGRID)) { // Check if this row should go into reduced matrix

      EPETRA_CHK_ERR(GetRowGCIDs(i, NumEntries, Values, Indices)); // Get current row (Indices are global)
      
      int ierr = ReducedMatrix()->InsertGlobalValues(curGRID, NumEntries, 
                 Values, Indices); // Insert into reduce matrix
      // Positive errors will occur because we are submitting col entries that are not part of
      // reduced system.  However, because we specified a column map to the ReducedMatrix constructor
      // these extra column entries will be ignored and we will be politely reminded by a positive
      // error code
      if (ierr<0) EPETRA_CHK_ERR(ierr); 
    }
    else {
      EPETRA_CHK_ERR(GetRow(i, NumEntries, Values, Indices)); // Get current row
      if (NumEntries==1) {
  double pivot = Values[0];
  if (pivot==0.0) EPETRA_CHK_ERR(-1); // Encountered zero row, unable to continue
  int indX = Indices[0];
  for (j=0; j<NumVectors; j++)
    (*tempExportX_)[j][indX] = (*FullRHS)[j][i]/pivot;
      }
      // Otherwise, this is a singleton column and we will scan for the pivot element needed 
      // for post-solve equations
      else {
  int targetCol = ColSingletonColLIDs_[ColSingletonCounter];
  for (j=0; j<NumEntries; j++) {
    if (Indices[j]==targetCol) {
      double pivot = Values[j];
      if (pivot==0.0) EPETRA_CHK_ERR(-2); // Encountered zero column, unable to continue
      ColSingletonPivotLIDs_[ColSingletonCounter] = j; // Save for later use
      ColSingletonPivots_[ColSingletonCounter] = pivot;
      ColSingletonCounter++;
      break;
    }
  }
      }
    }
  }

  // Now convert to local indexing.  We have constructed things so that the domain and range of the
  // matrix will have the same map.  If the reduced matrix domain and range maps were not the same, the
  // differences were addressed in the ConstructRedistributeExporter() method
  EPETRA_CHK_ERR(ReducedMatrix()->FillComplete(*ReducedMatrixDomainMap(), *ReducedMatrixRangeMap()));

  // Construct Reduced LHS (Puts any initial guess values into reduced system)

  ReducedLHS_ = new Epetra_MultiVector(*ReducedMatrixDomainMap(), NumVectors);
  EPETRA_CHK_ERR(ReducedLHS_->Import(*FullLHS, *Full2ReducedLHSImporter_, Insert));
  FullLHS->PutScalar(0.0); // zero out Full LHS since we will inject values as we get them

  // Construct Reduced RHS

  // First compute influence of already-known values of X on RHS
  tempX_ = new Epetra_MultiVector(FullMatrixDomainMap(), NumVectors);
  tempB_ = new Epetra_MultiVector(FullRHS->Map(), NumVectors);
  
  //Inject known X values into tempX for purpose of computing tempB = FullMatrix*tempX
  // Also inject into full X since we already know the solution

  if (FullMatrix()->RowMatrixImporter()!=0) {
    EPETRA_CHK_ERR(tempX_->Export(*tempExportX_, *FullMatrix()->RowMatrixImporter(), Add));
    EPETRA_CHK_ERR(FullLHS->Export(*tempExportX_, *FullMatrix()->RowMatrixImporter(), Add));
  }
  else {
    tempX_->Update(1.0, *tempExportX_, 0.0);
    FullLHS->Update(1.0, *tempExportX_, 0.0);
  }


  EPETRA_CHK_ERR(FullMatrix()->Multiply(false, *tempX_, *tempB_));

  EPETRA_CHK_ERR(tempB_->Update(1.0, *FullRHS, -1.0)); // tempB now has influence of already-known X values

  ReducedRHS_ = new Epetra_MultiVector(*ReducedMatrixRowMap(), FullRHS->NumVectors());
  EPETRA_CHK_ERR(ReducedRHS_->Import(*tempB_, *Full2ReducedRHSImporter_, Insert));

  // Finally construct Reduced Linear Problem
  ReducedProblem_ = new Epetra_LinearProblem(ReducedMatrix_, ReducedLHS_, ReducedRHS_);

  double fn = FullMatrix()->NumGlobalRows();
  double fnnz = FullMatrix()->NumGlobalNonzeros();
  double rn = ReducedMatrix()->NumGlobalRows();
  double rnnz = ReducedMatrix()->NumGlobalNonzeros();

  RatioOfDimensions_ = rn/fn;
  RatioOfNonzeros_ = rnnz/fnnz;
  HaveReducedProblem_ = true;
  
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::UpdateReducedProblem(Epetra_LinearProblem * Problem) {

  int i, j;

  if (Problem==0) EPETRA_CHK_ERR(-1); // Null problem pointer

  FullProblem_ = Problem;
  FullMatrix_ = dynamic_cast<Epetra_RowMatrix *>(Problem->GetMatrix());
  if (FullMatrix_==0) EPETRA_CHK_ERR(-2); // Need a RowMatrix
  if (Problem->GetRHS()==0) EPETRA_CHK_ERR(-3); // Need a RHS
  if (Problem->GetLHS()==0) EPETRA_CHK_ERR(-4); // Need a LHS
  if (!HaveReducedProblem_) EPETRA_CHK_ERR(-5); // Must have set up reduced problem

  // Create pointer to Full RHS, LHS
  Epetra_MultiVector * FullRHS = FullProblem()->GetRHS();
  Epetra_MultiVector * FullLHS = FullProblem()->GetLHS();
  int NumVectors = FullLHS->NumVectors();

  int NumEntries;
  int * Indices;
  double * Values;
  int NumMyRows = FullMatrix()->NumMyRows();
  int ColSingletonCounter = 0;
  for (i=0; i<NumMyRows; i++) {
    int curGRID = FullMatrixRowMap().GID(i);
    if (ReducedMatrixRowMap()->MyGID(curGRID)) { // Check if this row should go into reduced matrix
      EPETRA_CHK_ERR(GetRowGCIDs(i, NumEntries, Values, Indices)); // Get current row (indices global)
      int ierr = ReducedMatrix()->ReplaceGlobalValues(curGRID, NumEntries, 
                  Values, Indices);
      // Positive errors will occur because we are submitting col entries that are not part of
      // reduced system.  However, because we specified a column map to the ReducedMatrix constructor
      // these extra column entries will be ignored and we will be politely reminded by a positive
      // error code
      if (ierr<0) EPETRA_CHK_ERR(ierr); 
    }
    // Otherwise if singleton row we explicitly eliminate this row and solve for corresponding X value
    else {
      EPETRA_CHK_ERR(GetRow(i, NumEntries, Values, Indices)); // Get current row
      if (NumEntries==1) {
  double pivot = Values[0];
  if (pivot==0.0) EPETRA_CHK_ERR(-1); // Encountered zero row, unable to continue
  int indX = Indices[0];
  for (j=0; j<NumVectors; j++)
    (*tempExportX_)[j][indX] = (*FullRHS)[j][i]/pivot;
      }
      // Otherwise, this is a singleton column and we will scan for the pivot element needed 
      // for post-solve equations
      else {
  j = ColSingletonPivotLIDs_[ColSingletonCounter];
  double pivot = Values[j];
  if (pivot==0.0) EPETRA_CHK_ERR(-2); // Encountered zero column, unable to continue
  ColSingletonPivots_[ColSingletonCounter] = pivot;
  ColSingletonCounter++;
      }
    }
  }

  assert(ColSingletonCounter==NumMyColSingletons_); // Sanity test

  // Update Reduced LHS (Puts any initial guess values into reduced system)

  ReducedLHS_->PutScalar(0.0); // zero out Reduced LHS
  EPETRA_CHK_ERR(ReducedLHS_->Import(*FullLHS, *Full2ReducedLHSImporter_, Insert));
  FullLHS->PutScalar(0.0); // zero out Full LHS since we will inject values as we get them

  // Construct Reduced RHS

  // Zero out temp space
  tempX_->PutScalar(0.0);
  tempB_->PutScalar(0.0);
  
  //Inject known X values into tempX for purpose of computing tempB = FullMatrix*tempX
  // Also inject into full X since we already know the solution

  if (FullMatrix()->RowMatrixImporter()!=0) {
    EPETRA_CHK_ERR(tempX_->Export(*tempExportX_, *FullMatrix()->RowMatrixImporter(), Add));
    EPETRA_CHK_ERR(FullLHS->Export(*tempExportX_, *FullMatrix()->RowMatrixImporter(), Add));
  }
  else {
    tempX_->Update(1.0, *tempExportX_, 0.0);
    FullLHS->Update(1.0, *tempExportX_, 0.0);
  }


  EPETRA_CHK_ERR(FullMatrix()->Multiply(false, *tempX_, *tempB_));

  EPETRA_CHK_ERR(tempB_->Update(1.0, *FullRHS, -1.0)); // tempB now has influence of already-known X values

  ReducedRHS_->PutScalar(0.0);
  EPETRA_CHK_ERR(ReducedRHS_->Import(*tempB_, *Full2ReducedRHSImporter_, Insert));
    return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::ConstructRedistributeExporter(Epetra_Map * SourceMap, Epetra_Map * TargetMap,
                   Epetra_Export * & RedistributeExporter,
                   Epetra_Map * & RedistributeMap) {

  int IndexBase = SourceMap->IndexBase();
  if (IndexBase!=TargetMap->IndexBase()) EPETRA_CHK_ERR(-1);

  const Epetra_Comm & Comm = TargetMap->Comm();

  int TargetNumMyElements = TargetMap->NumMyElements();
  int SourceNumMyElements = SourceMap->NumMyElements();

  // ContiguousTargetMap has same number of elements per PE as TargetMap, but uses contigious indexing 
  Epetra_Map ContiguousTargetMap(-1, TargetNumMyElements, IndexBase,Comm);

  // Same for ContiguousSourceMap
  Epetra_Map ContiguousSourceMap(-1, SourceNumMyElements, IndexBase, Comm);

  assert(ContiguousSourceMap.NumGlobalElements()==ContiguousTargetMap.NumGlobalElements());

  // Now create a vector that contains the global indices of the Source Epetra_MultiVector
  Epetra_IntVector SourceIndices(View, ContiguousSourceMap, SourceMap->MyGlobalElements());

  // Create an exporter to send the SourceMap global IDs to the target distribution
  Epetra_Export Exporter(ContiguousSourceMap, ContiguousTargetMap);
  
  // Create a vector to catch the global IDs in the target distribution
  Epetra_IntVector TargetIndices(ContiguousTargetMap);
  TargetIndices.Export(SourceIndices, Exporter, Insert);

  // Create a new map that describes how the Source MultiVector should be laid out so that it has
  // the same number of elements on each processor as the TargetMap
  RedistributeMap = new Epetra_Map(-1, TargetNumMyElements, TargetIndices.Values(), IndexBase, Comm);

  // This exporter will finally redistribute the Source MultiVector to the same layout as the TargetMap
  RedistributeExporter = new Epetra_Export(*SourceMap, *RedistributeMap);
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::ComputeFullSolution() {

  int jj, k;

  Epetra_MultiVector * FullLHS = FullProblem()->GetLHS(); 
  Epetra_MultiVector * FullRHS = FullProblem()->GetRHS(); 

  tempX_->PutScalar(0.0); tempExportX_->PutScalar(0.0);
  // Inject values that the user computed for the reduced problem into the full solution vector
  EPETRA_CHK_ERR(tempX_->Export(*ReducedLHS_, *Full2ReducedLHSImporter_, Add));
  FullLHS->Update(1.0, *tempX_, 1.0);

  // Next we will use our full solution vector which is populated with pre-filter solution
  // values and reduced system solution values to compute the sum of the row contributions
  // that must be subtracted to get the post-filter solution values

  EPETRA_CHK_ERR(FullMatrix()->Multiply(false, *FullLHS, *tempB_));



  // Finally we loop through the local rows that were associated with column singletons and compute the
  // solution for these equations.

  int NumVectors = tempB_->NumVectors();
  for (k=0; k<NumMyColSingletons_; k++) {
    int i = ColSingletonRowLIDs_[k];
    int j = ColSingletonColLIDs_[k];
    double pivot = ColSingletonPivots_[k];
    for (jj=0; jj<NumVectors; jj++)
      (*tempExportX_)[jj][j]= ((*FullRHS)[jj][i] - (*tempB_)[jj][i])/pivot;
  }

  // Finally, insert values from post-solve step and we are done!!!!

  
  if (FullMatrix()->RowMatrixImporter()!=0) {
    EPETRA_CHK_ERR(tempX_->Export(*tempExportX_, *FullMatrix()->RowMatrixImporter(), Add));
  }
  else {
    tempX_->Update(1.0, *tempExportX_, 0.0);
  }

  FullLHS->Update(1.0, *tempX_, 1.0);
    
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::InitFullMatrixAccess() {

  MaxNumMyEntries_ = FullMatrix()->MaxNumEntries(); 

  // Cast to CrsMatrix, if possible.  Can save some work.
  FullCrsMatrix_ = dynamic_cast<Epetra_CrsMatrix *>(FullMatrix());
  FullMatrixIsCrsMatrix_ = (FullCrsMatrix_!=0); // Pointer is non-zero if cast worked
  Indices_ = new int[MaxNumMyEntries_];
  Values_.Size(MaxNumMyEntries_);

  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::GetRow(int Row, int & NumIndices, int * & Indices) {

  if (FullMatrixIsCrsMatrix_) { // View of current row
    EPETRA_CHK_ERR(FullCrsMatrix()->Graph().ExtractMyRowView(Row, NumIndices, Indices)); 
  }
  else { // Copy of current row (we must get the values, but we ignore them)
    EPETRA_CHK_ERR(FullMatrix()->ExtractMyRowCopy(Row, MaxNumMyEntries_, NumIndices, 
              Values_.Values(), Indices_));
    Indices = Indices_;
  } 
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::GetRow(int Row, int & NumIndices, 
              double * & Values, int * & Indices) {

  if (FullMatrixIsCrsMatrix_) { // View of current row
    EPETRA_CHK_ERR(FullCrsMatrix_->ExtractMyRowView(Row, NumIndices, Values, Indices)); 
  }
  else { // Copy of current row (we must get the values, but we ignore them)
    EPETRA_CHK_ERR(FullMatrix()->ExtractMyRowCopy(Row, MaxNumMyEntries_, NumIndices, 
              Values_.Values(), Indices_));
    Values = Values_.Values();
    Indices = Indices_;
  } 
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::GetRowGCIDs(int Row, int & NumIndices, 
             double * & Values, int * & GlobalIndices) {

    EPETRA_CHK_ERR(FullMatrix()->ExtractMyRowCopy(Row, MaxNumMyEntries_, NumIndices, 
              Values_.Values(), Indices_));
    for (int j=0; j<NumIndices; j++) Indices_[j] = FullMatrixColMap().GID(Indices_[j]);
    Values = Values_.Values();
    GlobalIndices = Indices_;
  return(0);
}
//==============================================================================
int LinearProblem_CrsSingletonFilter::CreatePostSolveArrays(const Epetra_IntVector & RowIDs,
                 const Epetra_MapColoring & RowMapColors,
                 const Epetra_IntVector & ColProfiles,
                 const Epetra_IntVector & NewColProfiles,
                 const Epetra_IntVector & ColHasRowWithSingleton) {

  int j;

  if (NumMyColSingletons_==0) return(0); // Nothing to do

  Epetra_MapColoring & ColMapColors = *ColMapColors_;

  int NumMyCols = FullMatrix()->NumMyCols();

  // We will need these arrays for the post-solve phase
  ColSingletonRowLIDs_ = new int[NumMyColSingletons_];
  ColSingletonColLIDs_ = new int[NumMyColSingletons_];
  ColSingletonPivotLIDs_ = new int[NumMyColSingletons_];
  ColSingletonPivots_ = new double[NumMyColSingletons_];
  
  // Register singleton columns (that were not already counted as singleton rows)
  // Check to see if any columns disappeared because all associated rows were eliminated
  int NumMyColSingletonstmp = 0;
  for (j=0; j<NumMyCols; j++) {
    int i = RowIDs[j];
    if ( ColProfiles[j]==1 && RowMapColors[i]!=1 ) {
      ColSingletonRowLIDs_[NumMyColSingletonstmp] = i;
      ColSingletonColLIDs_[NumMyColSingletonstmp] = j;
      NumMyColSingletonstmp++;
    }
    // Also check for columns that were eliminated implicitly by 
    // having all associated row eliminated
    else if (NewColProfiles[j]==0 && ColHasRowWithSingleton[j]!=1 && RowMapColors[i]==0) {
    ColMapColors[j] = 1;
    }
  }

  assert(NumMyColSingletonstmp==NumMyColSingletons_); //Sanity check
  Epetra_Util sorter;
  sorter.Sort(true, NumMyColSingletons_, ColSingletonRowLIDs_, 0, 0, 1, &ColSingletonColLIDs_);
    
  return(0);
}

} //namespace EpetraExt

---