helmholtz Module Reference

Data Types
type	approx_space
	Type to hold the approximation space. Should not be modified outside this module, so more of a handle. More...

Public Member Functions
subroutine, public	init_approx_space (apx, n_max, ntot)
	Initialize approximation space object. More...
subroutine, public	hsolve (name, u, r, h1, h2, vmk, vml, imsh, tol, maxit, isd, apx, bi)
	Either std. Helmholtz solve, or a projection + Helmholtz solve. More...

Private Member Functions
subroutine, private	projh (r, h1, h2, bi, vml, vmk, apx, wl, ws, name4)
	Project out the part of the residual in the approx space. More...
subroutine, private	gensh (v1, h1, h2, vml, vmk, apx, ws)
	Reconstruct the solution to the original problem by adding back the approximate solution, add solution to approximation space. More...
subroutine, private	hconj (apx, k, h1, h2, vml, vmk, ws)
	Update the k-th row/column of H_red. More...
subroutine, private	updrhsh (apx, h1, h2, vml, vmk, ws)
	Recompute H_red if dt has changed. More...
subroutine, private	hmhzpf (name, u, r, h1, h2, mask, mult, imesh, tli, maxit, isd, bi)

Detailed Description

Definition at line 15 of file navier4.F90.

Member Function/Subroutine Documentation

subroutine, private helmholtz::gensh ( real(dp), dimension (*), intent(inout)  v1, real(dp), dimension (*), intent(in)  h1, real(dp), dimension (*), intent(in)  h2, real(dp), dimension(*), intent(in)  vml, real(dp), dimension(*), intent(in)  vmk, type(approx_space), intent(inout)  apx, real(dp), dimension(:), intent(out)  ws  )

private

Reconstruct the solution to the original problem by adding back the approximate solution, add solution to approximation space.

Parameters

[in,out]	v1	Full solution
[in]	h1	coefficient of A (stiffness)
[in]	h2	coefficient of M (mass)
[in]	vmk	multiplicity array
[in]	vml	mask array
[out]	ws	small workspace to pass-through
[in,out]	apx	current approximation space

Definition at line 215 of file navier4.F90.

References copy(), and hconj().

Referenced by hsolve().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine, private helmholtz::hconj ( type(approx_space), intent(inout)  apx, integer, intent(in)  k, real(dp), dimension(*), intent(in)  h1, real(dp), dimension(*), intent(in)  h2, real(dp), dimension(*), intent(in)  vml, real(dp), dimension(*), intent(in)  vmk, real(dp), dimension(*), intent(out)  ws  )

private

Update the k-th row/column of H_red.

Parameters

[in,out]	apx	Current approximation space
[in]	k	Index of new projector
[in]	h1	coefficient of A (stiffness)
[in]	h2	coefficient of M (mass)
[in]	vml	multiplicity array
[in]	vmk	mask array
[out]	ws	small workspace

Definition at line 250 of file navier4.F90.

References axhelm(), ctimer::dnekclock(), dssum(), and gop().

Referenced by gensh(), and updrhsh().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine, private helmholtz::hmhzpf ( character(4)  name, real(dp), dimension (lx1,ly1,lz1,1), intent(out)  u, real(dp), dimension (lx1,ly1,lz1,1), intent(in)  r, real(dp), dimension (lx1,ly1,lz1,1), intent(in)  h1, real(dp), dimension (lx1,ly1,lz1,1), intent(in)  h2, real(dp), dimension (lx1,ly1,lz1,1), intent(in)  mask, real(dp), dimension (lx1,ly1,lz1,1), intent(in)  mult, integer  imesh, real(dp)  tli, integer  maxit, integer  isd, real(dp), dimension (lx1,ly1,lz1,1), intent(in)  bi  )

private

Parameters

[out]	u	solution vector
[in]	r	right hand side
[in]	h1	coefficient of A (stiffness)
[in]	h2	coefficient of M (mass)
[in]	mask	mask array
[in]	mult	multiplicity array
[in]	bi	inverse of mass matrix

Definition at line 344 of file navier4.F90.

References cggo(), chktcg1(), and ctimer::dnekclock().

Referenced by hsolve().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine, public helmholtz::hsolve	(	character(4), intent(in)	name,
		real(dp), dimension (lx1,ly1,lz1,lelv), intent(out)	u,
		real(dp), dimension (lx1,ly1,lz1,lelv), intent(inout)	r,
		real(dp), dimension (lx1,ly1,lz1,lelv), intent(in)	h1,
		real(dp), dimension (lx1,ly1,lz1,lelv), intent(in)	h2,
		real(dp), dimension (lx1,ly1,lz1,lelv), intent(in)	vmk,
		real(dp), dimension (lx1,ly1,lz1,lelv), intent(in)	vml,
		integer, intent(in)	imsh,
		real(dp), intent(in)	tol,
		integer, intent(in)	maxit,
		integer, intent(in)	isd,
		type(approx_space), intent(inout)	apx,
		real(dp), dimension (lx1,ly1,lz1,*), intent(in)	bi
	)

Either std. Helmholtz solve, or a projection + Helmholtz solve.

Parameters

[in]	name	name of field we're solving for
[out]	u	solution vector
[in,out]	r	right hand side
[in]	h1	coefficient of A (stiffness)
[in]	h2	coefficient of M (mass)
[in]	vmk	mask array
[in]	vml	multiplicity array
[in]	imsh	imesh?
[in]	tol	residual tolerance
[in]	maxit	maximum number of iterations
[in]	isd	something to do with axi-symmetric
[in,out]	apx	current approximation space
[in]	bi	inverse of mass matrix

Definition at line 396 of file navier4.F90.

References string::capit(), chcopy(), dssum(), gensh(), hmholtz(), hmhzpf(), and projh().

Referenced by cdscal(), and plan4().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine, public helmholtz::init_approx_space	(	type(approx_space), intent(out)	apx,
		integer, intent(in)	n_max,
		integer, intent(in)	ntot
	)

Initialize approximation space object.

Simple assigns and allocations

Definition at line 41 of file navier4.F90.

Referenced by plan4().

Here is the caller graph for this function:

subroutine, private helmholtz::projh ( real(dp), dimension(*), intent(inout)  r, real(dp), dimension(*), intent(in)  h1, real(dp), dimension(*), intent(in)  h2, real(dp), dimension(*), intent(in)  bi, real(dp), dimension(*), intent(in)  vml, real(dp), dimension(*), intent(in)  vmk, type(approx_space), intent(inout)  apx, real(dp), dimension(*), intent(out)  wl, real(dp), dimension(*), intent(out)  ws, character(4), intent(in)  name4  )

private

Project out the part of the residual in the approx space.

Starts by finding the EVD H_red to get orthogonal projectors. Then, computes the overlap of the residual with each projector and mixes them based on the EVD to get an orthogonal projection. Next, multiplies by 1/ to take the inverse and expands back into the full space to populate the approximate solution: projectors(:,0)

Parameters

[in,out]	r	residual
[in]	h1	coefficient of A (stiffness)
[in]	h2	coefficient of M (mass)
[in]	vml	multiplicity array
[in]	vmk	mask array
[in]	bi	inverse mass matrix
[out]	wl	large work array (size lx1*ly1*lz1*nelv)
[out]	ws	small work array (size 2*max vecs)
[in,out]	apx	Current approx space
[in]	name4	Name of field for debug printing

Note

This dsyev call and the following dgemv are task-parallel!

Definition at line 62 of file navier4.F90.

References axhelm(), ctimer::dnekclock(), dssum(), glsc2(), gop(), and updrhsh().

Referenced by hsolve().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine, private helmholtz::updrhsh ( type(approx_space), intent(inout)  apx, real(dp), dimension(*), intent(in)  h1, real(dp), dimension(*), intent(in)  h2, real(dp), dimension(*), intent(in)  vml, real(dp), dimension(*), intent(in)  vmk, real(dp), dimension(*), intent(out)  ws  )

private

Recompute H_red if dt has changed.

Parameters

[in,out]	apx	current approximation space
[in]	h1	coefficient of A (stiffness)
[in]	h2	coefficient of M (mass)
[in]	vml	multiplicity array
[in]	vmk	mask array
[out]	ws	small workspace

Definition at line 301 of file navier4.F90.

References hconj().

Referenced by projh().

Here is the call graph for this function:

Here is the caller graph for this function:

---

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

• navier4.F90