HI everyone, I'm trying to a very large scale linear equations like this :
B*X=L
the Matrix B could be as big as 200,000 *200,000

I can use SimplicialLLT to get the solution with little solution time
But now I need to get some specific values in the Inverse matrix of B

I set the Matrix L as a identity matrix, and tried to solute it with SimplicialLLT

Code: Select all

   
   SimplicialLLT < SparseMatrix < double > > llt;
   llt.compute(B);
   SparseMatrix < double > L;
   L.resize(roNum, roNum);
   L.setIdentity();
   X = llt.solve(L);


But the program crashed, can't get any result
It's because the matrix is too big?
Does Eigen have any good solution for the problem?
Thanks a lot~

Heyho,

in general the inverse of a sparse matrix is not sparse. Consequently do not try to calculate the inverse.
If you need specific values, I think you are stuck with calculating it row by row.

You might consider using iterative solvers as well to speed up your problem.

Cheers Jens