lammps/lib/atc/SparseMatrix.h

300 lines
11 KiB
C
Raw Normal View History

#ifndef SPARSEMATRIX_H
#define SPARSEMATRIX_H
#include <exception>
#include "MatrixLibrary.h"
#include <algorithm>
namespace ATC_matrix {
/**
* @struct TRI_COORD
* @brief Triplet SparseMatrix entry
*/
template <typename T>
struct TRI_COORD
{
TRI_COORD<T>(INDEX row=0, INDEX col=0);
TRI_COORD<T>(INDEX row, INDEX col, T val, bool add_to=0);
INDEX i, j;
T v;
bool add;
};
template<typename T>
void ParMultAB(MPI_Comm comm, const SparseMatrix<T>& A, const Matrix<T>& B, DenseMatrix<T>& C);
/**
* @class SparseMatrix
* @brief Stores data in triplet format or CRS format
*/
template<typename T>
class SparseMatrix : public Matrix<T>
{
//* SparseMatrix-Vector multiplication (S * v)
friend DenseVector<T> operator*<T>(const SparseMatrix<T> &A, const Vector<T>& x);
//* SparseMatrix-DenseMatrix multiplication (S * F)
friend DenseMatrix<T> operator*<T>(const SparseMatrix<T> &A, const Matrix<T>& D);
//* SparseMatrix-DiagonalMatrix multiplication (S * D)
friend SparseMatrix<T> operator*<T>(const SparseMatrix<T> &A, const DiagonalMatrix<T>& D);
//* SparseMatrix-SparseMatrix multiplication (S * S)
friend SparseMatrix<T> operator*<T>(const SparseMatrix<T> &A, const SparseMatrix<T> &B);
//* computes the product of a SparseMatrix tranpose with a SparseVector (M'*v).
friend SparseVector<T> operator*<T>(const SparseMatrix<T> &M, const SparseVector<T> &v);
//* computes the product of a SparseMatrix tranpose with a SparseVector (M'*v).
friend SparseVector<T> operator*<T>(const SparseVector<T> &v, const SparseMatrix<T> &M);
template<typename U>
friend void ParMultAB(MPI_Comm comm, const SparseMatrix<U>& A, const Matrix<U>& B, DenseMatrix<U>& C);
public:
SparseMatrix(INDEX rows=0, INDEX cols=0);
SparseMatrix(const SparseMatrix<T>& c);
SparseMatrix(const DenseMatrix<T>& c);
SparseMatrix(INDEX* rows, INDEX* cols, T* vals, INDEX size,
INDEX nRows, INDEX nCols, INDEX nRowsCRS);
virtual ~SparseMatrix() { _delete(); }
//* General index by value (requires a binary search on the row)
T operator()(INDEX i, INDEX j) const;
//* General index by reference (requires a binary search on the row)
T& operator()(INDEX i, INDEX j);
//* General flat index by value operator (by nth nonzero)
T operator[](INDEX i) const;
//* General flat index by reference operator (by nth nonzero)
T& operator[](INDEX i);
//* sets a value to index i,j
void set(INDEX i, INDEX j, T v);
//* adds a value to index i,j
void add(INDEX i, INDEX j, T v);
//* return a triplet value of the ith nonzero
TRIPLET<T> triplet(INDEX i) const;
//* full reset - completely wipes out all SparseMatrix data
void reset(INDEX rows=0, INDEX cols=0, bool zero=true);
//* only changes the bounds of the matrix, no deletion
void resize(INDEX rows=0, INDEX cols=0, bool zero=true);
//* reset - from DenseMatrix - this will be SLOW
void reset(const DenseMatrix<T>& D, double TOL=-1.0);
//* copy data
void copy(const T * ptr, INDEX rows=0, INDEX cols=0);
void dense_copy(DenseMatrix<T>& D) const;
DenseMatrix<T> dense_copy(void) const;
//* returns true if the matrix has no nonzero elements
bool empty() const;
//* returns the user-specified number of rows
INDEX nRows() const;
INDEX nRowsCRS() const;
//* returns the user-specified number of cols
INDEX nCols() const;
//* returns the number of non-zero elements
INDEX size() const;
//* returns the number of non-zeros in a row
INDEX RowSize(INDEX r) const;
//* returns a pointer to the CRS list of rows
inline INDEX* rows() const;
//* returns a pointer to the CRS list of cols
inline INDEX* cols() const;
//* returns a pointer to the nonzero data
inline T* ptr() const;
//* checks if the index i,j falls in the user-specified range
bool in_range(INDEX i, INDEX j) const;
//* check if the total matrix has a value set for an index pair
bool has_entry(INDEX i, INDEX j) const;
//* check if the uncompressed part of the matrix has a value set for an index pair
bool has_entry_uncompressed(INDEX i, INDEX j) const;
//* check if the compressed part matrix has a value set for an index pair
bool has_entry_compressed(INDEX i, INDEX j) const;
//* check if the matrix has been compressed at least once
bool has_template(void) const;
/*
* \section assignment operators
*/
//* copies SparseMatrix R to this
SparseMatrix<T>& operator=(const SparseMatrix &R);
//* sets all nonzero values to a constant
SparseMatrix<T>& operator=(const T v);
//* scales all nonzero values by a constant
SparseMatrix<T>& operator*=(const T &a);
//* calls operator*= of base class
SparseMatrix<T>& operator*=(const SparseMatrix<T> &a);
// Adds two matrices together.
SparseMatrix<T>& operator+=(const SparseMatrix & R);
/*
* \section Multiplication operations
*/
//-----------------------------------------------------------------------------
// multiply sparse matrix by a vector
//-----------------------------------------------------------------------------
virtual void MultMv(const Vector<T>& v, DenseVector<T>& c) const
{
compress(*this);
GCK(*this, v, this->nCols() != v.size(), "SparseMatrix * Vector")
// resize c if necessary
if (c.size() != this->nRows()) {
c.resize(this->nRows());
c.zero();
}
INDEX i, j;
for (i = 0; i < this->_nRowsCRS; i++)
for (j = this->_ia[i]; j < this->_ia[i + 1]; j++)
c(i) += this->_val[j] * v(this->_ja[j]);
}
//-----------------------------------------------------------------------------
// multiply sparse matrix by dense matrix
//-----------------------------------------------------------------------------
virtual void MultAB(const Matrix<T>& B, DenseMatrix<T>& C) const
{
GCK(*this, B, this->nCols() != B.nRows(), "SparseMatrix * DenseMatrix")
const INDEX J = B.nCols();
INDEX i, ik, j;
for (i = 0; i < this->_nRowsCRS; i++)
for (ik = this->_ia[i]; ik < this->_ia[i + 1]; ik++)
for (j = 0; j < J; j++)
C(i, j) += this->_val[ik] * B(this->_ja[ik], j); // C(i,j) = S(i,k) * B(k, j)
}
//-----------------------------------------------------------------------------
// Multiplies this SparseMatrix transposed times a vector
//-----------------------------------------------------------------------------
virtual DenseVector<T> transMat(const Vector<T> &x) const
{
compress(*this);
DenseVector<T> y(nCols(), true);
GCK(*this, x, nRows()!=x.size(),"operator *: Sparse matrix incompatible with Vector.")
INDEX i, ij;
for(i=0; i<_nRowsCRS; i++)
for(ij=_ia[i]; ij<_ia[i+1]; ij++)
y(_ja[ij]) += _val[ij]*x(i);
return y;
}
//-----------------------------------------------------------------------------
// Matrix Transpose/DenseMatrix multiply
//-----------------------------------------------------------------------------
virtual DenseMatrix<T> transMat(const DenseMatrix<T> &D) const
{
compress(*this);
GCK(*this, D, nRows()!=D.nRows(),"transMat: Sparse matrix incompatible with DenseMatrix.")
DenseMatrix<T> C(nCols(), D.nCols(), true); // initialized to zero
INDEX j, k, ki;
for (k=0; k<_nRowsCRS; k++)
for (ki=_ia[k]; ki<_ia[k+1]; ki++)
for (j=0; j<D.nCols(); j++)
C(_ja[ki], j) += _val[ki]*D(k,j); // C(i,j) = S(k,i) * D(k, j)
return C;
}
//-----------------------------------------------------------------------------
// Matrix Transpose/SparseMatrix multiply - IS THIS REALLY NEEDED??
//-----------------------------------------------------------------------------
virtual DenseMatrix<T> transMat(const SparseMatrix<T> &D) const
{
compress(*this);
GCK(*this, D, nRows()!=D.nRows(),"transMat: Sparse matrix incompatible with DenseMatrix.")
DenseMatrix<T> C(nCols(), D.nCols(), true); // initialized to zero
INDEX k, ki, kj;
for (k=0; k<_nRowsCRS; k++)
for (kj=D._ia[k]; kj<D._ia[k+1]; kj++)
for (ki=_ia[k]; ki<_ia[k+1]; ki++)
C(_ja[ki], D._ja[kj]) += _val[ki]*D._val[kj]; // C(i,j) = S(k,i)*D(k,j)
return C;
}
SparseMatrix<T> transpose() const;
SparseMatrix<T>& row_scale(const Vector<T> &v);
SparseMatrix<T>& col_scale(const Vector<T> &v);
DenseVector<T> col_sum() const;
DenseVector<INDEX> column_count() const;
DiagonalMatrix<T> diag() const;
DiagonalMatrix<T> row_sum_lump() const;
void row(INDEX i, DenseVector<T>& row, DenseVector<INDEX>& indx) const;
void weighted_least_squares(const SparseMatrix<T> &N, const DiagonalMatrix<T> &D);
void set_all_elements_to(const T &v);
T row_max(INDEX row) const;
T row_min(INDEX row) const;
/*
* \section I/O functions
*/
//* outputs this SparseMatrix to a formatted string
string to_string() const;
using Matrix<T>::matlab;
//* writes a command to recreate this matrix in matlab to a stream
void matlab(ostream &o, const string &name="S") const;
//* prints a row histogram for each row
void print_row_histogram(const string &name, INDEX nbins = 10) const;
//* prints a histogram of the values in a row
void print_row_histogram(INDEX row, INDEX nbins) const;
//* prints the current triplets
void print_triplets() const;
//! Writes the matrix to a binary file (after a compress).
void binary_write(std::fstream& f) const;
//! Reads a SparseMatrix from a binary file. (wipes out any original data)
void binary_read(std::fstream& f);
//* Dump templated type to disk; operation not safe for all types
void write_restart(FILE *f) const;
/*
* \section Utility functions
*/
//* converts all triplets and merges with CRS
void compress();
//* converts T to CRS
static void compress(const SparseMatrix<T> &C);
//* sorts and returns the # of unique triplets
INDEX CountUniqueTriplets();
private:
//* creates a CRS structure
void _create(INDEX size, INDEX nrows);
//* clears all memory and nulls references
void _delete();
//* copies all data from another SparseMatrix
void _copy(const SparseMatrix<T> &C);
//* general sparse matrix assignment
void _set_equal(const Matrix<T> &r);
//* returns the first row with a nonzero in it (from the CRS structure only)
int _first_nonzero_row_crs() const;
/*
* \section CRS storage variables
*/
protected:
T * _val; // matrix non-zeros
INDEX *_ia, *_ja; // ptrs to rows, column indexes
INDEX _size, _nRowsCRS; // # of non-zeros, rows
bool hasTemplate_;
void copy(const SparseMatrix<T> &C);
//* new (unsorted triplet values - won't intersect CRS values)
mutable vector<TRI_COORD<T> > _tri;
/*
* \section User specified variables
*/
INDEX _nRows, _nCols;
static T _zero;
};
} // end namespace
#include "SparseMatrix-inl.h"
#endif