使用Octave API for C ++的稀疏矩阵构造

Question

I'm trying to use the Octave C++ API for solving sparse linear system. The problem I have encountered is that I cannot find an efficient way to construct the sparse linear system.

SparseMatrix A;
int dim,nnz;
// dim = dimension
// nnz = number of non-zero entry in the matrix A

/*
    somehow assign the values for dim and nnz
*/

A = SparseMatrix ( dim, dim, nnz );

// row index array
int *pidx_r = new int[nnz];
// col index array
int *pidx_c = new int[nnz];
// value array
double *pval = new double[nnz];
// total number of nonzero elements
int tot; 

/*
    somehow assign values for pidx_r, pidx_c, pval and tot
*/

for ( int i = 0; i < tot; i++ )
    A ( pidx_r[i], pidx_c[i] ) = pval[i];

The above piece of code describes my naive implementation which is extremely inefficient, and I expect that there should be some member functions to massively insert values into the sparse matrix.

For example, A=SparseMatrix(pidx_r,pidx_c,pval);

However, I cannot find any member functions for doing so. At least, it seems that the naive implementation is the only approach.

Given I have prepared a matrix in some format, I would like to ask if there is any method to construct a sparse matrix efficiently using Octave API for C++?

Answer 1

I just found this tutorial, the method in the first paragraph looks like what you were looking for.

link

Answer 2

@wirew0rm The method mentioned in the link is trivial and inefficient.

In fact, I have already found a solution which basically modifies the content of a inherited member SparseMatrixRep.

Though the documentation stated that it is unsafe and not recommended, the performance is significantly improved.