|
blitzdg
an open-source project aiming to implement parallel discontinuous Galerkin (dg) solvers for common partial differential equations systems using blitz++ for array and tensor manipulations and MPI for distributed parallelism.
|
Public Member Functions | |
| LUSolver (Array< double, 2 > *const &, SparseMatrixConverter const &) | |
| Constructor. More... | |
| Array< double, 2 > & | get_A () |
| Returns a reference to the matrix A. | |
| void | factorize () |
| Factorize the matrix A with UMFPACK. More... | |
| void | solve (Array< double, 1 > const &, Array< double, 1 > &) |
| Solve Ax=b using UMFPACK. More... | |
| LUSolver::LUSolver | ( | Array< double, 2 > *const & | Ain, |
| SparseMatrixConverter const & | _matrixConverter | ||
| ) |
Constructor.
Takes a pointer reference to a blitz 2D array (The matrix A to be used by the solver in Ax=b).
| void LUSolver::factorize | ( | ) |
Factorize the matrix A with UMFPACK.
Computes L,U factors and permutation matrices P,Q such that P*A*Q=LU.
| void LUSolver::solve | ( | Array< double, 1 > const & | rhs, |
| Array< double, 1 > & | soln | ||
| ) |
Solve Ax=b using UMFPACK.
Requires LUSolver.factorize() to be called first. 'x' is returned in 'soln' reference. 'b' is specified by 'rhs' reference.
1.8.11