Functions/Subroutines
subroutine	gen_fast_spacing ()
	Generate fast diagonalization matrices for each element. More...

subroutine	plane_space (lr, ls, lt, i1, i2, w, x, y, z, nx, nxn, nz0, nzn)
	Here, spacing is based on harmonic mean. pff 2/10/07. More...

subroutine	get_fast_bc (lbr, rbr, lbs, rbs, lbt, rbt, e, bsym, ierr)

subroutine	load_semhat_weighted ()
	Note that this routine performs the following matrix multiplies after getting the matrices back from semhat: dgl = bgl dgl jgl = bgl jgl. More...

subroutine	do_semhat_weight (jgl, dgl, bgl, n)
	re-weight jgl and dgl with bgl More...

subroutine	generate_semhat (a, b, c, d, z, dgll, jgll, bgl, zgl, dgl, jgl, n, w)
	Generate matrices for single element, 1D operators: More...

subroutine	fd_weights_full (xx, x, n, m, c)
	This routine evaluates the derivative based on all points in the stencils. It is more memory efficient than "fd_weights". More...

subroutine	swap_lengths ()

subroutine	row_zero (a, m, n, e)
	zero the eth row of a More...

subroutine	swap (b, ind, n, temp)
	Reorder vector using temporary buffer. More...

Function/Subroutine Documentation

subroutine do_semhat_weight	(	real(dp), dimension(1:n-1,0:n)	jgl,
		real(dp), dimension(1:n-1,0:n)	dgl,
		real(dp), dimension(1:n-1)	bgl,
		integer	n
	)

re-weight jgl and dgl with bgl

Definition at line 289 of file fast3d.F90.

Referenced by load_semhat_weighted().

Here is the caller graph for this function:

subroutine fd_weights_full	(	real(dp)	xx,
		real(dp), dimension(0:n)	x,
		integer	n,
		integer	m,
		real(dp), dimension(0:n,0:m)	c
	)

This routine evaluates the derivative based on all points in the stencils. It is more memory efficient than "fd_weights".

This set of routines comes from the appendix of A Practical Guide to Pseudospectral Methods, B. Fornberg Cambridge Univ. Press, 1996. (pff)

Input parameters: xx – point at wich the approximations are to be accurate x – array of x-ordinates: x(0:n) n – polynomial degree of interpolant (# of points := n+1) m – highest order of derivative to be approxxmated at xi Output: c – set of coefficients c(0:n,0:m). c(j,k) is to be applied at x(j) when the kth derivative is approxxmated by a stencil extending over x(0),x(1),...x(n).

Definition at line 409 of file fast3d.F90.

Referenced by gen_dgl(), gen_int(), generate_semhat(), hsmg_routines::hsmg_setup_intpm(), and xyzlin().

Here is the caller graph for this function:

subroutine gen_fast_spacing ( )

Generate fast diagonalization matrices for each element.

Definition at line 3 of file fast3d.F90.

References load_semhat_weighted(), plane_space(), speclib::zwgl(), and speclib::zwgll().

Referenced by set_overlap().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine generate_semhat	(	real(dp), dimension(0:n,0:n)	a,
		real(dp), dimension(0:n)	b,
		real(dp), dimension(0:n,0:n)	c,
		real(dp), dimension(0:n,0:n)	d,
		real(dp), dimension(0:n)	z,
		real(dp), dimension(0:n,1:n-1)	dgll,
		real(dp), dimension(0:n,1:n-1)	jgll,
		real(dp), dimension(1:n-1)	bgl,
		real(dp), dimension(1:n-1)	zgl,
		real(dp), dimension(1:n-1,0:n)	dgl,
		real(dp), dimension(1:n-1,0:n)	jgl,
		integer	n,
		real(dp), dimension(0:(n+1)*2)	w
	)

Generate matrices for single element, 1D operators:

a = Laplacian b = diagonal mass matrix c = convection operator b*d d = derivative matrix dgll = derivative matrix, mapping from pressure nodes to velocity jgll = interpolation matrix, mapping from pressure nodes to velocity z = GLL points

zgl = GL points bgl = diagonal mass matrix on GL dgl = derivative matrix, mapping from velocity nodes to pressure jgl = interpolation matrix, mapping from velocity nodes to pressure

n = polynomial degree (velocity space) w = work array of size 2*n+2

Currently, this is set up for pressure nodes on the interior GLL pts.

Definition at line 328 of file fast3d.F90.

References fd_weights_full(), speclib::zwgl(), and speclib::zwgll().

Referenced by hsmg_routines::h1mg_setup_semhat(), hsmg_routines::hsmg_setup_semhat(), and load_semhat_weighted().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine get_fast_bc	(	integer	lbr,
		integer	rbr,
		integer	lbs,
		integer	rbs,
		integer	lbt,
		integer	rbt,
		integer	e,
		integer	bsym,
		integer	ierr
	)

Definition at line 192 of file fast3d.F90.

References exitti(), and parallel::lglel().

Referenced by hsmg_routines::h1mg_setup_mask(), and hsmg_routines::hsmg_setup_fast().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine load_semhat_weighted ( )

Note that this routine performs the following matrix multiplies after getting the matrices back from semhat: dgl = bgl dgl jgl = bgl jgl.

Definition at line 274 of file fast3d.F90.

References do_semhat_weight(), and generate_semhat().

Referenced by gen_fast_spacing().

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine plane_space	(	real(dp), dimension(*)	lr,
		real(dp), dimension(*)	ls,
		real(dp), dimension(*)	lt,
		integer	i1,
		integer	i2,
		real(dp), dimension(*)	w,
		real(dp), dimension(0:nxn,0:nxn,nz0:nzn,*)	x,
		real(dp), dimension(0:nxn,0:nxn,nz0:nzn,*)	y,
		real(dp), dimension(0:nxn,0:nxn,nz0:nzn,*)	z,
		integer	nx,
		integer	nxn,
		integer	nz0,
		integer	nzn
	)