sparskit::pgmres Interface Reference

An ILUT preconditioned GMRES algorithm. This routine utilizes the L and U matrices generated by the ILUT routine to precondition the GMRES algorithm. The stopping criteria utilized is based simply on reducing the residual norm to the requested tolerance. More...

Public Member Functions
subroutine	pgmres (n, im, rhs, sol, vv, eps, maxits, iout, aa, ja, ia, alu, jlu, ju, ierr)

Detailed Description

An ILUT preconditioned GMRES algorithm. This routine utilizes the L and U matrices generated by the ILUT routine to precondition the GMRES algorithm. The stopping criteria utilized is based simply on reducing the residual norm to the requested tolerance.

Parameters

[in]	n	The row dimension of the matrix.
[in]	im	The size of the Krylov subspace. This value should not exceed 50.
[in,out]	rhs	The N-element right-hand-side vector. On output, the contents of this array are overwritten.
[in,out]	sol	On input, the N-element solution estimate. On output, the computed solution.
[out]	vv	An N-by-IM+1 workspace matrix.
[in]	eps	The convergence tolerance against which the norm of the residual is checked.
[in]	maxits	The maximum number of iterations to allow.
[in]	iout	The device output number for printing intermediate results. Set to a value less than or equal to zero to suppress printing.
[in]	aa	The non-zero elements of matrix A.
[in]	ja	The column indices of matrix A.
[in]	ia	The index in A where the requested row starts.
[in]	alu	The LU-factored matrix from ILUT.
[in]	jlu	The LU-factored matrix from ILUT.
[in]	ju	The LU-factored matrix from ILUT.
[out]	ierr	Error flag: 0: Successful return 1: Convergence not achieved. -1: The initial guess seems to be the exact solution.

Definition at line 540 of file sparskit.f90.

Constructor & Destructor Documentation

pgmres()

subroutine sparskit::pgmres::pgmres	(	integer(int32), intent(in)	n,
		integer(int32), intent(in)	im,
		real(real64), dimension(n), intent(inout)	rhs,
		real(real64), dimension(n), intent(inout)	sol,
		real(real64), dimension(n,*), intent(out)	vv,
		real(real64), intent(in)	eps,
		integer(int32), intent(in)	maxits,
		integer(int32), intent(in)	iout,
		real(real64), dimension(*), intent(in)	aa,
		integer(int32), dimension(*), intent(in)	ja,
		integer(int32), dimension(n+1), intent(in)	ia,
		real(real64), dimension(*), intent(in)	alu,
		integer(int32), dimension(*), intent(in)	jlu,
		integer(int32), dimension(n), intent(in)	ju,
		integer(int32), intent(out)	ierr )

Definition at line 540 of file sparskit.f90.

---

The documentation for this interface was generated from the following file:

• D:/Code/linalg/src/sparskit.f90