adam_field_object

ADAM, field class definition.

A structured block is composed of hexahedron finite volumes with quadrilateral faces using the following internal numeration for nodes and faces:

 /|\Z
  |                            F(4)         _ F(6)
  |                            /|\          /!
  |                        7    |          /    8
  |                         *------------------*
  |                        /|   |        /    /|
  |                       / |   |       /    / |
  |                      /  |   |      /    /  |
  |                     /   |   |     /    /   |
  |                    /    |   |    +    /    |
  |                   /     |   |        /     |
  |                  /      |   +       /      |
  |                 /      3|          /       |4
  |                /        * --------/--------*
  |      F(1)<----/----+   /         /        /
  |              *------------------*    +-------->F(2)
  |             5|       /          |6      /
  |              |      /           |      /
  |              |     /        +   |     /
  |              |    /         |   |    /
  |              |   /      +   |   |   /
  |              |  /      /    |   |  /
  |              | /      /     |   | /
  |              |/      /      |   |/
  |              *------------------*
  |             1      /        |    2
  |                   /        \|/
  |   _ Y           |/_       F(3)
  |   /|         F(5)
  |  /
  | /
  |/                                                    X
  O----------------------------------------------------->

Each hexadron cells is faces-connected to its neighboring, thus the cells build a structured block with implicit connectivity, e.g. in 2D space a FriVolous block could be as the following:

                 _ J
                 /|                          _____
               5+ ...*----*----*----*----*...     |
               /    /    /    /    /    /         |
              /    /    /    /    /    /          |
            4+ ...*----*----*----*----*...        |
            /    /    /    /    /    /            |
           /    /    /    /    /    /             |
         3+ ...*----*----*----*----*...           |  Structured block of 4x4 Finite Volumes
         /    /    / FV /    /    /               |
        /    /    /    /    /    /                |
      2+ ...*----*----*----*----*...              |
      /    /    /    /    /    /                  |
     /    /    /    /    /    /                   |
   1+ ...*----*----*----*----*...                 |
   /     .    .    .    .    .                    |
  /      .    .    .    .    .               _____
 O-------+----+----+----+----+-------------> I
         1    2    3    4    5

Source: src/lib/common/adam_field_object.F90

Dependencies

Contents

• field_object
• adapt
• blocks_reorder
• compute_metrics
• compute_normL2_residuals
• initialize
• load_blocks
• load_from_ini_file
• mark_sphere
• mark_all_blocks
• mpi_gather_refinements_needed
• get_refinements_needed
• mpi_redistribute
• save_blocks
• update_ghost_local
• reduce_ghost_local
• reduce_ghost_mpi
• check_reduce_ghost_maps
• update_ghost_mpi
• derefine1D
• derefine2D
• derefine3D
• refine1D
• refine2D
• refine3D
• refine_conservative
• update_coordinates
• description
• do_caxis_intersect
• do_cplane_intersect
• do_ray_intersect
• interp_seam_ghost
• restrict_seam_high_order
• mc_slope

Variables

Name	Type	Attributes	Description
INI_SECTION_NAME	character(len=5)	parameter	INI (config) file section name containing configs.

Derived Types

field_object

Field class definition.

Components

Name	Type	Attributes	Description
nv	integer(kind=I4P)		Number of field variables.
nv_pic	integer(kind=I4P)		Number of PIC (Particle In Cell) variables.
block_weight	integer(kind=I4P)		Block weight, cells_number * variables_number.
block_weight_pic	integer(kind=I4P)		Block weight, particles_number * pic_variables_number.
nb	integer(kind=I4P)		Number of all blocks that can be stored.
blocks_number	integer(kind=I4P)		Number of blocks actually stored.
np	integer(kind=I4P)		Number of all particles that can be stored for each block.
code	integer(kind=I8P)	allocatable	Morton codes [nb].
coordinates	integer(kind=I4P)	allocatable	Coordinates IJKL for each block [4,nb].
particles_number	integer(kind=I4P)	allocatable	Number of partcles actually stored in each block [nb].
emin	real(kind=R8P)	allocatable	Coordinates of minimum abscissa of each block [3,nb].
emax	real(kind=R8P)	allocatable	Coordinates of maximum abscissa of each block [3,nb].
dxyz	real(kind=R8P)	allocatable	Space steps of each block [3,nb].
x_node	real(kind=R8P)	allocatable	X node coordinates [0-ngc:ni+ngc,nb].
y_node	real(kind=R8P)	allocatable	Y node coordinates [0-ngc:nj+ngc,nb].
z_node	real(kind=R8P)	allocatable	Z node coordinates [0-ngc:nk+ngc,nb].
x_cell	real(kind=R8P)	allocatable	X cell coordinates [1-ngc:ni+ngc,nb].
y_cell	real(kind=R8P)	allocatable	Y cell coordinates [1-ngc:nj+ngc,nb].
z_cell	real(kind=R8P)	allocatable	Z cell coordinates [1-ngc:nk+ngc,nb].
req_send_recv	integer(kind=I4P)	allocatable	MPI request send/receive flags.
blocks_numbers	integer(kind=I4P)	allocatable	Number of blocks actually stored in all processes.
refinements_needed	integer(kind=I4P)	allocatable	Refinements needed of my blocks.
refinements_needed_all	integer(kind=I4P)	allocatable	Refinements needed of all blocks.
disp_count	integer(kind=I4P)	allocatable	Displacement of blocks that are received from process.
inner_blocks_number	integer(kind=I4P)		Number of inner blocks where I need fecs.
q_work	real(kind=R8P)	allocatable	Field cell centered variables, working buffer memory.
residuals	real(kind=R8P)	allocatable	Field residuals, normalized.
ngc	integer(kind=I4P)	pointer	Number of ghost cells.
ni	integer(kind=I4P)	pointer	Number of cells in i direction.
nj	integer(kind=I4P)	pointer	Number of cells in j direction.
nk	integer(kind=I4P)	pointer	Number of cells in k direction.

Type-Bound Procedures

Name	Attributes	Description
adapt	pass(self)	Adapt field accordingly to refine/derefine necessity.
blocks_reorder	pass(self)	Reorder blocks indexes in field.
compute_metrics	pass(self)	Compute metrics of each block.
compute_normL2_residuals	pass(self)	Compute L2 norm of residuals.
description	pass(self)	Return pretty-printed object description.
do_caxis_intersect	pass(self)	Return true if a block is intersected by coordinate-axis.
do_cplane_intersect	pass(self)	Return true if a block is intersected by coordinate-plane.
do_ray_intersect	pass(self)	Return true if a block is intersected by ray.
initialize	pass(self)	Initialize the field.
load_blocks	pass(self)	Load blocks data, used for restarting.
load_from_ini_file	pass(self)	Load object data from INI file.
mark_all_blocks	pass(self)	Mark all blocks to be refined, derefined, ecc.
mark_sphere	pass(self)	Mark blocks to be refined/derefined by sphere distance.
get_refinements_needed	pass(self)	Return gathered refinement flags and displacements.
mpi_gather_refinements_needed	pass(self)	Gather blocks refinement needed status between MPI processes.
mpi_redistribute	pass(self)	Redistribute blocks to processes.
save_blocks	pass(self)	Save blocks data, used for restarting.
check_reduce_ghost_maps	pass(self)	Check that every ghost is folded by one map row at most.
reduce_ghost_local	pass(self)	Fold deposited ghosts back onto their owner cells locally.
reduce_ghost_mpi	pass(self)	Fold deposited ghosts back onto their owner cells of other ranks.
update_ghost_local	pass(self)	Update ghosts locally.
update_ghost_mpi	pass(self)	Update ghosts MPI.
derefine1D	pass(self)	Derefine blocks, 1D.
derefine2D	pass(self)	Derefine blocks, 2D.
derefine3D	pass(self)	Derefine blocks, 3D.
refine1D	pass(self)	Refine blocks, 1D.
refine2D	pass(self)	Refine blocks, 2D.
refine3D	pass(self)	Refine blocks, 3D.
refine_conservative	pass(self)	Refine blocks, conservative limited prolongation.
update_coordinates	pass(self)	Update coordinates using the updated data in maps.

Subroutines

adapt

Adapt field accordingly to refine/derefine necessity.

prolongation selects how a refined block's children are filled: AMR_PROLONGATION_LINEAR (default, tensor linear interpolation, not conservative) or AMR_PROLONGATION_CONSERVATIVE (limited linear slopes whose children average to the parent, issue #74 D-M5-2). Derefinement is the mean of the children in both cases.

subroutine adapt(self, grid, plan, q, prolongation)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
plan	type(refinement_plan_object)	in		Refinement plan produced by tree%adapt.
q	real(kind=R8P)	inout		Field cell centered variables.
prolongation	integer(kind=I4P)	in	optional	Prolongation kind, AMR_PROLONGATION_*.

Call graph

blocks_reorder

Reorder blocks indexes in field.

subroutine blocks_reorder(self, grid, maps, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

compute_metrics

Compute metrics of each block.

subroutine compute_metrics(self, grid)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).

Call graph

compute_normL2_residuals

Compute L2 norm of residuals.

subroutine compute_normL2_residuals(self, grid, dq, norm)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
dq	real(kind=R8P)	in		Residuals.
norm	real(kind=R8P)	inout		Residuals norm.

Call graph

initialize

Initialize field.

add_adam mirrors tree_object%initialize's optional dummy of the same name (same default .true.). When true, the tree gains an ADAM ancestor node (Morton code -1) attributed to MPI rank 0. The field reflects this by setting blocks_number = 1 ON RANK 0 ONLY (the one that owns the ADAM block); other ranks keep blocks_number = 0. This matches the blocks_numbers(0) = 1 below at line 511 and the code(1) = -1 ADAM-block convention. When add_adam = .false., the tree starts empty and the field follows suit (blocks_number = 0 on every rank). The two add_adam defaults must agree to keep tree/field state consistent at end-of-initialize; callers that pass add_adam to one MUST pass the same value to the other.

subroutine initialize(self, grid, maps, tree, nb, file_parameters, nv, verbose, add_adam)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in	target	Grid (sibling realm component, threaded in).
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).
tree	type(tree_object)	in		Tree (sibling realm component, threaded in).
nb	integer(kind=I4P)	in		Number of all blocks that can be stored.
file_parameters	type(file_ini)	inout		INI file handler.
nv	integer(kind=I4P)	in	optional	Number of field variables.
verbose	logical	in	optional	Trigger verbose output.
add_adam	logical	in	optional	Add ADAM ancestor block (mirrors tree initialize's add_adam,

Call graph

load_blocks

Load blocks data, used for restarting.

Note: blocks memory must be already initialized with enough memory (proper nv,ni,nj,nk,ngc and nb>=blocks number).

subroutine load_blocks(self, grid, basename, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
basename	character(len=*)	in		Output base name.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

load_from_ini_file

Load object data from INI file.

subroutine load_from_ini_file(self, file_parameters, nv, go_on_fail)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
file_parameters	type(file_ini)	inout		INI file handler.
nv	integer(kind=I4P)	in	optional	Number of field variables.
go_on_fail	logical	in	optional	Go on if load fails.

Call graph

mark_sphere

Mark blocks to be refined/derefined by sphere distance.

subroutine mark_sphere(self, grid, center, radius, threshold)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
center	real(kind=R8P)	in		Sphere center coordinates [x,y,z].
radius	real(kind=R8P)	in		Sphere radius.
threshold	real(kind=R8P)	in	optional	Threshold for sphere proximity.

Call graph

mark_all_blocks

Mark all blocks to be refined, derefined, ecc.

subroutine mark_all_blocks(self, mark)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The tree.
mark	integer(kind=I4P)	in		Mark to be imposed [TO_BE_REFINED,...].

mpi_gather_refinements_needed

Gather blocks refinement needed status between MPI processes.

subroutine mpi_gather_refinements_needed(self)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.

Call graph

get_refinements_needed

Return gathered refinement flags and process displacements. Provides an accessor so callers do not reach into field internals. Call mpi_gather_refinements_needed before invoking this procedure.

subroutine get_refinements_needed(self, flags, disp)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The field.
flags	integer(kind=I4P)	out	allocatable	Refinements needed of all blocks.
disp	integer(kind=I4P)	out	allocatable	Displacement of received blocks per process.

Call graph

mpi_redistribute

Redistribute blocks to processes. @TODO: Morton codes are not yet redistributed, must be fixed.

subroutine mpi_redistribute(self, grid, maps, tree, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).
tree	type(tree_object)	in		Tree (sibling realm component, threaded in).
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

save_blocks

Save blocks data, used for restarting.

Every rank writes its file, with zero blocks if it owns none (issue #42): load_blocks reads one file per rank, so a rank that skipped the save aborted the restart, and a file left by an earlier save of that rank (when it still owned blocks) would have been read as current. status='replace' truncates a longer old file.

subroutine save_blocks(self, grid, basename, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
basename	character(len=*)	in		Output base name.
q	real(kind=R8P)	in		Field cell centered variables.

Call graph

update_ghost_local

Update (local) ghost cells.

subroutine update_ghost_local(self, grid, maps, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The base backend.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).
q	real(kind=R8P)	inout		Field component to be updated.

Call graph

reduce_ghost_local

Fold (local) ghost cells back onto the cells they mirror: the transpose of update_ghost_local.

Meant for quantities deposited by sources owned by one block (e.g. PIC charge/current), whose stencil spills into the ghost cells of the owner block: those contributions belong to the interior cells of the neighbour block that the ghosts mirror. For every same-level map row the ghost value is added to its donor cell, then the ghost is refilled with the (now complete) donor value, so that the result equals a single-block deposit of the same sources. Ghosts without a map row (physical boundaries) are left untouched. Only variables v_first:v_last are processed. Rows of 2:1 seams (coarse<->fine) have no transpose implemented yet: their presence is an error.

subroutine reduce_ghost_local(self, grid, maps, q, v_first, v_last)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).
q	real(kind=R8P)	inout		Field component to be reduced.
v_first	integer(kind=I4P)	in		First variable to reduce.
v_last	integer(kind=I4P)	in		Last variable to reduce.

Call graph

reduce_ghost_mpi

Fold the ghost cells filled by other ranks back onto their donor cells: the transpose of update_ghost_mpi.

MPI twin of reduce_ghost_local, for deposited quantities (e.g. PIC charge/current). The forward maps and buffers are reused with the roles swapped: every received ghost is packed into its receive-buffer slot, the receive segments travel back to the ranks that sent them, and each send-map row adds its slot to the donor cell. Superseded slots of the deduplicated receive map stay zero. The ghosts are NOT refilled: the caller completes the reduction with update_ghost_local + update_ghost_mpi once every fold (local and MPI) is done. Collective: every rank must call it. Only same-level rows are supported (2:1 seams have no transpose yet); see check_reduce_ghost_maps for the disjointness of the local and MPI maps that makes the two folds add each ghost exactly once.

subroutine reduce_ghost_mpi(self, grid, maps, q, v_first, v_last)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	inout		Maps (sibling realm component, threaded in).
q	real(kind=R8P)	inout		Field component to be reduced.
v_first	integer(kind=I4P)	in		First variable to reduce.
v_last	integer(kind=I4P)	in		Last variable to reduce.

Call graph

check_reduce_ghost_maps

Check that the transposed ghost updates (reduce_ghost_local + reduce_ghost_mpi) fold every ghost cell once at most: a ghost listed by two map rows (two local rows, two MPI rows, or one of each) would be added twice. The forward update tolerates such duplicates (last writer wins), the transpose does not. Only same-level maps are meaningful here.

subroutine check_reduce_ghost_maps(self, grid, maps)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).

Call graph

update_ghost_mpi

Update ghost cells within other processes.

subroutine update_ghost_mpi(self, grid, maps, q, step)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The base backend.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
maps	type(maps_object)	inout		Maps (sibling realm component, threaded in).
q	real(kind=R8P)	inout		Field component to be updated.
step	integer(kind=I4P)	in	optional	Step to be perfordmed in asyncronous comp.

Call graph

derefine1D

Derefine blocks.

Note: blocks number is not updated: mpi redistribute does it. This is dangerous...

subroutine derefine1D(self, grid, ratio, block_to_derefine, block_derefined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio.
block_to_derefine	integer(kind=I8P)	in	allocatable	List of blocks to be derefined.
block_derefined	integer(kind=I8P)	in	allocatable	List of derefined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

derefine2D

Derefine blocks.

Note: blocks number is not updated: mpi redistribute does it. This is dangerous...

subroutine derefine2D(self, grid, ratio, block_to_derefine, block_derefined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio.
block_to_derefine	integer(kind=I8P)	in	allocatable	List of blocks to be derefined.
block_derefined	integer(kind=I8P)	in	allocatable	List of derefined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

derefine3D

Derefine blocks.

Note: blocks number is not updated: mpi redistribute does it. This is dangerous...

subroutine derefine3D(self, grid, ratio, block_to_derefine, block_derefined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio.
block_to_derefine	integer(kind=I8P)	in	allocatable	List of blocks to be derefined.
block_derefined	integer(kind=I8P)	in	allocatable	List of derefined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

refine1D

Refine blocks.

Note: blocks number is not updated: mpi redistribute does it. This is dangerous...

subroutine refine1D(self, grid, ratio, block_to_refine, block_refined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio.
block_to_refine	integer(kind=I8P)	in	allocatable	List of blocks to be refined.
block_refined	integer(kind=I8P)	in	allocatable	List of refined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

refine2D

Refine blocks.

Note: blocks number is not updated: mpi redistribute does it. This is dangerous...

subroutine refine2D(self, grid, ratio, block_to_refine, block_refined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio.
block_to_refine	integer(kind=I8P)	in	allocatable	List of blocks to be refined.
block_refined	integer(kind=I8P)	in	allocatable	List of refined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

refine3D

Refine blocks.

Note: blocks number is not updated: mpi redistribute does it. This is dangerous...

subroutine refine3D(self, grid, ratio, block_to_refine, block_refined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio.
block_to_refine	integer(kind=I8P)	in	allocatable	List of blocks to be refined.
block_refined	integer(kind=I8P)	in	allocatable	List of refined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

refine_conservative

Refine blocks by a conservative limited linear prolongation (issue #74, D-M5-2).

Each parent cell is split along the refined axes (ratio 2: i; 4: i, j; 8: i, j, k) into children set, variable by variable, to q + phi * sum_d sigma_d * s_d / 4: s_d is the monotonized-central (MC) slope of the parent along axis d, in parent-cell units, from its two face neighbours (the first ghost layer included), and sigma_d = -1, +1 the side of the child. The offsets cancel over the children, so their mean is the parent: the mean restriction of derefine*D undoes this prolongation, and the discrete integral sum(q dV) is unchanged, to round-off. phi in [0, 1] scales the slopes so that no child leaves the range of the parent and its face neighbours (Barth and Jespersen 1989): the MC limit alone bounds each axis, not the sum over the axes, which a corner child takes. A variable that is positive on the parent and its neighbours stays positive. Linear data are reproduced exactly: the MC slope of linear data is its centred difference, and a corner child moves by at most 3/4 of the largest neighbour difference, so phi = 1. Like the linear prolongation, it reads the parent's first ghost layer, which the caller must have refreshed.

Note: blocks number is not updated: mpi redistribute does it.

subroutine refine_conservative(self, grid, ratio, block_to_refine, block_refined, q)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
ratio	integer(kind=I4P)	in		Refinement ratio, 2, 4 or 8.
block_to_refine	integer(kind=I8P)	in	allocatable	List of blocks to be refined.
block_refined	integer(kind=I8P)	in	allocatable	List of refined blocks with Morton code.
q	real(kind=R8P)	inout		Field cell centered variables.

Call graph

update_coordinates

Update coordinates using the updated data in maps (that in turn is updated by tree).

subroutine update_coordinates(self, maps, tree)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
maps	type(maps_object)	in		Maps (sibling realm component, threaded in).
tree	type(tree_object)	in		Tree (sibling realm component, threaded in).

Call graph

Functions

description

Return a pretty-formatted object description.

Attributes: pure

Returns: character(len=:)

function description(self) result(desc)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	in		The field.

Call graph

do_caxis_intersect

Return true if a block is intersected by coordinate-axis.

Returns: logical

function do_caxis_intersect(self, b, caxis_origin, caxis_direction, caxis_block_indexes) result(do_intersect)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
b	integer(kind=I4P)	in		Block index.
caxis_origin	real(kind=R8P)	in		Coordinate-axis origin.
caxis_direction	real(kind=R8P)	in		Coordinate-axis direction.
caxis_block_indexes	integer(kind=I4P)	out	optional	Block-local indexes of caxis intersection.

do_cplane_intersect

Return true if a block is intersected by coordinate-plane.

Returns: logical

function do_cplane_intersect(self, grid, b, cplane_origin, cplane_normal, cplane_block_indexes) result(do_intersect)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
grid	type(grid_object)	in		Grid (sibling realm component, threaded in).
b	integer(kind=I4P)	in		Block index.
cplane_origin	real(kind=R8P)	in		Coordinate-plane origin.
cplane_normal	real(kind=R8P)	in		Coordinate-plane normal.
cplane_block_indexes	integer(kind=I4P)	out	optional	Block-local indexes of cplane intersection.

Call graph

do_ray_intersect

Return true if a block is intersected by ray from ray_origin and oriented as ray_direction vector.

Returns: logical

function do_ray_intersect(self, b, ray_origin, ray_direction) result(do_intersect)

Arguments

Name	Type	Intent	Attributes	Description
self	class(field_object)	inout		The field.
b	integer(kind=I4P)	in		Block index.
ray_origin	real(kind=R8P)	in		Ray origin.
ray_direction	real(kind=R8P)	in		Ray direction.

Call graph

interp_seam_ghost

Evaluate the coarse->fine seam ghost interpolant for one (variable, fine ghost) pair (issue #21 N2): gather the donor footprint around the anchor (i_send,j_send,k_send) of block b_send and compose the 1D weights of the active regime. The anchor and the shift-clamped positions in meta guarantee the footprint reads REAL donor cells only (G3).

Attributes: pure

Returns: real(kind=R8P)

function interp_seam_ghost(regime, meta, ngc, q, v, i_send, j_send, k_send, b_send) result(value_)

Arguments

Name	Type	Intent	Attributes	Description
regime	integer(kind=I4P)	in		Active fill regime (SEAM_FILL_*).
meta	integer(kind=I4P)	in		Packed interpolation metadata (see seam_meta_pack).
ngc	integer(kind=I4P)	in		Ghost cells number (bounds of q).
q	real(kind=R8P)	in		Field variables.
v	integer(kind=I4P)	in		Variable index.
i_send	integer(kind=I4P)	in		Anchor donor cell, i.
j_send	integer(kind=I4P)	in		Anchor donor cell, j.
k_send	integer(kind=I4P)	in		Anchor donor cell, k.
b_send	integer(kind=I4P)	in		Donor block index.

Call graph

restrict_seam_high_order

Evaluate the high-order restriction of one (variable, coarse ghost) pair to the coarse centre, which sits midway between the fine cells first and first + 1 of each refined direction ((i_send,j_send,k_send) is the first fine cell of block b_send); the footprint first - code .. first - code + 5 is kept inside the donor block by the codes of meta (seam_restriction_code), which also carries the regime (seam_restriction_meta_pack):

• 6-point Lagrange (issue #68 C1): the tensor product of the 1D weights SEAM_W_LAGRANGE6, summed direction by direction (x innermost), the same order as the FNL twin;
• WENO (issue #68 C3): seam_restriction_weno6 on the gathered footprint. An unrefined direction (code SEAM_RESTRICT_IDENTITY) reads the cell itself only.

Attributes: pure

Returns: real(kind=R8P)

function restrict_seam_high_order(meta, ngc, q, v, i_send, j_send, k_send, b_send) result(value_)

Arguments

Name	Type	Intent	Attributes	Description
meta	integer(kind=I4P)	in		Packed restriction codes and regime (seam_restriction_meta_pack).
ngc	integer(kind=I4P)	in		Ghost cells number (bounds of q).
q	real(kind=R8P)	in		Field variables.
v	integer(kind=I4P)	in		Variable index.
i_send	integer(kind=I4P)	in		First fine cell of the restricted pair, i.
j_send	integer(kind=I4P)	in		First fine cell of the restricted pair, j.
k_send	integer(kind=I4P)	in		First fine cell of the restricted pair, k.
b_send	integer(kind=I4P)	in		Donor block index.

Call graph

mc_slope

Monotonized-central limited slope (van Leer 1977) from the backward and forward differences.

Zero at an extremum (the differences disagree in sign), else the smallest of the centred difference and twice each one-sided difference, with their common sign.

Attributes: elemental

Returns: real(kind=R8P)

function mc_slope(dm, dp) result(slope)

Arguments

Name	Type	Intent	Attributes	Description
dm	real(kind=R8P)	in		Backward difference, q(i) - q(i-1).
dp	real(kind=R8P)	in		Forward difference, q(i+1) - q(i).

Call graph