Home | History | Annotate | Download | only in SparseLU
      1 // This file is part of Eigen, a lightweight C++ template library
      2 // for linear algebra.
      3 //
      4 // Copyright (C) 2012 Dsir Nuentsa-Wakam <desire.nuentsa_wakam (at) inria.fr>
      5 // Copyright (C) 2012 Gael Guennebaud <gael.guennebaud (at) inria.fr>
      6 //
      7 // This Source Code Form is subject to the terms of the Mozilla
      8 // Public License v. 2.0. If a copy of the MPL was not distributed
      9 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
     10 
     11 #ifndef EIGEN_SPARSELU_SUPERNODAL_MATRIX_H
     12 #define EIGEN_SPARSELU_SUPERNODAL_MATRIX_H
     13 
     14 namespace Eigen {
     15 namespace internal {
     16 
     17 /** \ingroup SparseLU_Module
     18  * \brief a class to manipulate the L supernodal factor from the SparseLU factorization
     19  *
     20  * This class  contain the data to easily store
     21  * and manipulate the supernodes during the factorization and solution phase of Sparse LU.
     22  * Only the lower triangular matrix has supernodes.
     23  *
     24  * NOTE : This class corresponds to the SCformat structure in SuperLU
     25  *
     26  */
     27 /* TODO
     28  * InnerIterator as for sparsematrix
     29  * SuperInnerIterator to iterate through all supernodes
     30  * Function for triangular solve
     31  */
     32 template <typename _Scalar, typename _StorageIndex>
     33 class MappedSuperNodalMatrix
     34 {
     35   public:
     36     typedef _Scalar Scalar;
     37     typedef _StorageIndex StorageIndex;
     38     typedef Matrix<StorageIndex,Dynamic,1> IndexVector;
     39     typedef Matrix<Scalar,Dynamic,1> ScalarVector;
     40   public:
     41     MappedSuperNodalMatrix()
     42     {
     43 
     44     }
     45     MappedSuperNodalMatrix(Index m, Index n,  ScalarVector& nzval, IndexVector& nzval_colptr, IndexVector& rowind,
     46              IndexVector& rowind_colptr, IndexVector& col_to_sup, IndexVector& sup_to_col )
     47     {
     48       setInfos(m, n, nzval, nzval_colptr, rowind, rowind_colptr, col_to_sup, sup_to_col);
     49     }
     50 
     51     ~MappedSuperNodalMatrix()
     52     {
     53 
     54     }
     55     /**
     56      * Set appropriate pointers for the lower triangular supernodal matrix
     57      * These infos are available at the end of the numerical factorization
     58      * FIXME This class will be modified such that it can be use in the course
     59      * of the factorization.
     60      */
     61     void setInfos(Index m, Index n, ScalarVector& nzval, IndexVector& nzval_colptr, IndexVector& rowind,
     62              IndexVector& rowind_colptr, IndexVector& col_to_sup, IndexVector& sup_to_col )
     63     {
     64       m_row = m;
     65       m_col = n;
     66       m_nzval = nzval.data();
     67       m_nzval_colptr = nzval_colptr.data();
     68       m_rowind = rowind.data();
     69       m_rowind_colptr = rowind_colptr.data();
     70       m_nsuper = col_to_sup(n);
     71       m_col_to_sup = col_to_sup.data();
     72       m_sup_to_col = sup_to_col.data();
     73     }
     74 
     75     /**
     76      * Number of rows
     77      */
     78     Index rows() { return m_row; }
     79 
     80     /**
     81      * Number of columns
     82      */
     83     Index cols() { return m_col; }
     84 
     85     /**
     86      * Return the array of nonzero values packed by column
     87      *
     88      * The size is nnz
     89      */
     90     Scalar* valuePtr() {  return m_nzval; }
     91 
     92     const Scalar* valuePtr() const
     93     {
     94       return m_nzval;
     95     }
     96     /**
     97      * Return the pointers to the beginning of each column in \ref valuePtr()
     98      */
     99     StorageIndex* colIndexPtr()
    100     {
    101       return m_nzval_colptr;
    102     }
    103 
    104     const StorageIndex* colIndexPtr() const
    105     {
    106       return m_nzval_colptr;
    107     }
    108 
    109     /**
    110      * Return the array of compressed row indices of all supernodes
    111      */
    112     StorageIndex* rowIndex()  { return m_rowind; }
    113 
    114     const StorageIndex* rowIndex() const
    115     {
    116       return m_rowind;
    117     }
    118 
    119     /**
    120      * Return the location in \em rowvaluePtr() which starts each column
    121      */
    122     StorageIndex* rowIndexPtr() { return m_rowind_colptr; }
    123 
    124     const StorageIndex* rowIndexPtr() const
    125     {
    126       return m_rowind_colptr;
    127     }
    128 
    129     /**
    130      * Return the array of column-to-supernode mapping
    131      */
    132     StorageIndex* colToSup()  { return m_col_to_sup; }
    133 
    134     const StorageIndex* colToSup() const
    135     {
    136       return m_col_to_sup;
    137     }
    138     /**
    139      * Return the array of supernode-to-column mapping
    140      */
    141     StorageIndex* supToCol() { return m_sup_to_col; }
    142 
    143     const StorageIndex* supToCol() const
    144     {
    145       return m_sup_to_col;
    146     }
    147 
    148     /**
    149      * Return the number of supernodes
    150      */
    151     Index nsuper() const
    152     {
    153       return m_nsuper;
    154     }
    155 
    156     class InnerIterator;
    157     template<typename Dest>
    158     void solveInPlace( MatrixBase<Dest>&X) const;
    159 
    160 
    161 
    162 
    163   protected:
    164     Index m_row; // Number of rows
    165     Index m_col; // Number of columns
    166     Index m_nsuper; // Number of supernodes
    167     Scalar* m_nzval; //array of nonzero values packed by column
    168     StorageIndex* m_nzval_colptr; //nzval_colptr[j] Stores the location in nzval[] which starts column j
    169     StorageIndex* m_rowind; // Array of compressed row indices of rectangular supernodes
    170     StorageIndex* m_rowind_colptr; //rowind_colptr[j] stores the location in rowind[] which starts column j
    171     StorageIndex* m_col_to_sup; // col_to_sup[j] is the supernode number to which column j belongs
    172     StorageIndex* m_sup_to_col; //sup_to_col[s] points to the starting column of the s-th supernode
    173 
    174   private :
    175 };
    176 
    177 /**
    178   * \brief InnerIterator class to iterate over nonzero values of the current column in the supernodal matrix L
    179   *
    180   */
    181 template<typename Scalar, typename StorageIndex>
    182 class MappedSuperNodalMatrix<Scalar,StorageIndex>::InnerIterator
    183 {
    184   public:
    185      InnerIterator(const MappedSuperNodalMatrix& mat, Index outer)
    186       : m_matrix(mat),
    187         m_outer(outer),
    188         m_supno(mat.colToSup()[outer]),
    189         m_idval(mat.colIndexPtr()[outer]),
    190         m_startidval(m_idval),
    191         m_endidval(mat.colIndexPtr()[outer+1]),
    192         m_idrow(mat.rowIndexPtr()[mat.supToCol()[mat.colToSup()[outer]]]),
    193         m_endidrow(mat.rowIndexPtr()[mat.supToCol()[mat.colToSup()[outer]]+1])
    194     {}
    195     inline InnerIterator& operator++()
    196     {
    197       m_idval++;
    198       m_idrow++;
    199       return *this;
    200     }
    201     inline Scalar value() const { return m_matrix.valuePtr()[m_idval]; }
    202 
    203     inline Scalar& valueRef() { return const_cast<Scalar&>(m_matrix.valuePtr()[m_idval]); }
    204 
    205     inline Index index() const { return m_matrix.rowIndex()[m_idrow]; }
    206     inline Index row() const { return index(); }
    207     inline Index col() const { return m_outer; }
    208 
    209     inline Index supIndex() const { return m_supno; }
    210 
    211     inline operator bool() const
    212     {
    213       return ( (m_idval < m_endidval) && (m_idval >= m_startidval)
    214                 && (m_idrow < m_endidrow) );
    215     }
    216 
    217   protected:
    218     const MappedSuperNodalMatrix& m_matrix; // Supernodal lower triangular matrix
    219     const Index m_outer;                    // Current column
    220     const Index m_supno;                    // Current SuperNode number
    221     Index m_idval;                          // Index to browse the values in the current column
    222     const Index m_startidval;               // Start of the column value
    223     const Index m_endidval;                 // End of the column value
    224     Index m_idrow;                          // Index to browse the row indices
    225     Index m_endidrow;                       // End index of row indices of the current column
    226 };
    227 
    228 /**
    229  * \brief Solve with the supernode triangular matrix
    230  *
    231  */
    232 template<typename Scalar, typename Index_>
    233 template<typename Dest>
    234 void MappedSuperNodalMatrix<Scalar,Index_>::solveInPlace( MatrixBase<Dest>&X) const
    235 {
    236     /* Explicit type conversion as the Index type of MatrixBase<Dest> may be wider than Index */
    237 //    eigen_assert(X.rows() <= NumTraits<Index>::highest());
    238 //    eigen_assert(X.cols() <= NumTraits<Index>::highest());
    239     Index n    = int(X.rows());
    240     Index nrhs = Index(X.cols());
    241     const Scalar * Lval = valuePtr();                 // Nonzero values
    242     Matrix<Scalar,Dynamic,Dest::ColsAtCompileTime, ColMajor> work(n, nrhs);     // working vector
    243     work.setZero();
    244     for (Index k = 0; k <= nsuper(); k ++)
    245     {
    246       Index fsupc = supToCol()[k];                    // First column of the current supernode
    247       Index istart = rowIndexPtr()[fsupc];            // Pointer index to the subscript of the current column
    248       Index nsupr = rowIndexPtr()[fsupc+1] - istart;  // Number of rows in the current supernode
    249       Index nsupc = supToCol()[k+1] - fsupc;          // Number of columns in the current supernode
    250       Index nrow = nsupr - nsupc;                     // Number of rows in the non-diagonal part of the supernode
    251       Index irow;                                     //Current index row
    252 
    253       if (nsupc == 1 )
    254       {
    255         for (Index j = 0; j < nrhs; j++)
    256         {
    257           InnerIterator it(*this, fsupc);
    258           ++it; // Skip the diagonal element
    259           for (; it; ++it)
    260           {
    261             irow = it.row();
    262             X(irow, j) -= X(fsupc, j) * it.value();
    263           }
    264         }
    265       }
    266       else
    267       {
    268         // The supernode has more than one column
    269         Index luptr = colIndexPtr()[fsupc];
    270         Index lda = colIndexPtr()[fsupc+1] - luptr;
    271 
    272         // Triangular solve
    273         Map<const Matrix<Scalar,Dynamic,Dynamic, ColMajor>, 0, OuterStride<> > A( &(Lval[luptr]), nsupc, nsupc, OuterStride<>(lda) );
    274         Map< Matrix<Scalar,Dynamic,Dest::ColsAtCompileTime, ColMajor>, 0, OuterStride<> > U (&(X(fsupc,0)), nsupc, nrhs, OuterStride<>(n) );
    275         U = A.template triangularView<UnitLower>().solve(U);
    276 
    277         // Matrix-vector product
    278         new (&A) Map<const Matrix<Scalar,Dynamic,Dynamic, ColMajor>, 0, OuterStride<> > ( &(Lval[luptr+nsupc]), nrow, nsupc, OuterStride<>(lda) );
    279         work.topRows(nrow).noalias() = A * U;
    280 
    281         //Begin Scatter
    282         for (Index j = 0; j < nrhs; j++)
    283         {
    284           Index iptr = istart + nsupc;
    285           for (Index i = 0; i < nrow; i++)
    286           {
    287             irow = rowIndex()[iptr];
    288             X(irow, j) -= work(i, j); // Scatter operation
    289             work(i, j) = Scalar(0);
    290             iptr++;
    291           }
    292         }
    293       }
    294     }
    295 }
    296 
    297 } // end namespace internal
    298 
    299 } // end namespace Eigen
    300 
    301 #endif // EIGEN_SPARSELU_MATRIX_H
    302