Home

Context Navigation

source: palm/trunk/SOURCE/init_pegrid.f90 @ 4337

Last change on this file since 4337 was 4301, checked in by oliver.maas, 5 years ago
Deleted parameter recycling_yshift. y-shift in case of non-cyclic boundary conditions and turbulent_inflow = .TRUE. is now steered by parameter y_shift, that is also used in case of cyclic boundary conditions.
Property svn:keywords set to `Id`
File size: 51.0 KB

Line
1	!> @file init_pegrid.f90
2	!------------------------------------------------------------------------------!
3	! This file is part of the PALM model system.
4	!
5	! PALM is free software: you can redistribute it and/or modify it under the
6	! terms of the GNU General Public License as published by the Free Software
7	! Foundation, either version 3 of the License, or (at your option) any later
8	! version.
9	!
10	! PALM is distributed in the hope that it will be useful, but WITHOUT ANY
11	! WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
12	! A PARTICULAR PURPOSE. See the GNU General Public License for more details.
13	!
14	! You should have received a copy of the GNU General Public License along with
15	! PALM. If not, see <http://www.gnu.org/licenses/>.
16	!
17	! Copyright 1997-2019 Leibniz Universitaet Hannover
18	!------------------------------------------------------------------------------!
19	!
20	! Current revisions:
21	! ------------------
22	!
23	!
24	! Former revisions:
25	! -----------------
26	! $Id: init_pegrid.f90 4301 2019-11-22 12:09:09Z Giersch $
27	! changed message PA0467
28	!
29	! 4264 2019-10-15 16:00:23Z scharf
30	! corrected error message string
31	!
32	! 4241 2019-09-27 06:32:47Z raasch
33	! Check added to ensure that subdomain grid has at least the size as given by the number
34	! of ghost points
35	!
36	! 4182 2019-08-22 15:20:23Z scharf
37	! Corrected "Former revisions" section
38	!
39	! 4045 2019-06-21 10:58:47Z raasch
40	! bugfix: kind attribute added to nint function to allow for large integers which may appear in
41	! case of default recycling width and small grid spacings
42	!
43	! 3999 2019-05-23 16:09:37Z suehring
44	! Spend 3 ghost points also in case of pw-scheme when nesting is applied
45	!
46	! 3897 2019-04-15 11:51:14Z suehring
47	! Minor revision of multigrid check; give warning instead of an abort.
48	!
49	! 3890 2019-04-12 15:59:20Z suehring
50	! Check if grid coarsening is possible on subdomain, in order to avoid that
51	! multigrid approach effectively reduces to a Gauss-Seidel scheme.
52	!
53	! 3885 2019-04-11 11:29:34Z kanani
54	! Changes related to global restructuring of location messages and introduction
55	! of additional debug messages
56	!
57	! 3884 2019-04-10 13:31:55Z Giersch
58	! id_recycling is only calculated in case of tubulent inflow
59	!
60	! 3761 2019-02-25 15:31:42Z raasch
61	! unused variable removed
62	!
63	! 3655 2019-01-07 16:51:22Z knoop
64	! variables documented
65	!
66	! Revision 1.1 1997/07/24 11:15:09 raasch
67	! Initial revision
68	!
69	!
70	! Description:
71	! ------------
72	!> Determination of the virtual processor topology (if not prescribed by the
73	!> user)and computation of the grid point number and array bounds of the local
74	!> domains.
75	!> @todo: remove MPI-data types for 2D exchange on coarse multigrid level (not
76	!> used any more)
77	!------------------------------------------------------------------------------!
78	SUBROUTINE init_pegrid
79
80
81	USE control_parameters, &
82	ONLY: bc_dirichlet_l, bc_dirichlet_n, bc_dirichlet_r, bc_dirichlet_s, &
83	bc_lr, bc_ns, bc_radiation_l, bc_radiation_n, bc_radiation_r, &
84	bc_radiation_s, coupling_mode, coupling_topology, gathered_size,&
85	grid_level, grid_level_count, maximum_grid_level, &
86	message_string, mg_switch_to_pe0_level, momentum_advec, &
87	psolver, outflow_source_plane, recycling_width, scalar_advec, &
88	subdomain_size, turbulent_inflow, turbulent_outflow, y_shift
89
90	USE grid_variables, &
91	ONLY: dx
92
93	USE indices, &
94	ONLY: mg_loc_ind, nbgp, nnx, nny, nnz, nx, nx_a, nx_o, nxl, nxl_mg, &
95	nxlu, nxr, nxr_mg, ny, ny_a, ny_o, nyn, nyn_mg, nys, nys_mg, &
96	nysv, nz, nzb, nzt, nzt_mg, wall_flags_1, wall_flags_2, &
97	wall_flags_3, wall_flags_4, wall_flags_5, wall_flags_6, &
98	wall_flags_7, wall_flags_8, wall_flags_9, wall_flags_10
99
100	USE kinds
101
102	USE pegrid
103
104	USE pmc_interface, &
105	ONLY: nested_run
106
107	USE spectra_mod, &
108	ONLY: calculate_spectra, dt_dosp
109
110	USE synthetic_turbulence_generator_mod, &
111	ONLY: id_stg_left, id_stg_north, id_stg_right, id_stg_south, &
112	use_syn_turb_gen
113
114	USE transpose_indices, &
115	ONLY: nxl_y, nxl_yd, nxl_z, nxr_y, nxr_yd, nxr_z, nyn_x, nyn_z, nys_x,&
116	nys_z, nzb_x, nzb_y, nzb_yd, nzt_x, nzt_yd, nzt_y
117
118	USE vertical_nesting_mod, &
119	ONLY: vnested, vnest_init_pegrid_domain, vnest_init_pegrid_rank
120
121	IMPLICIT NONE
122
123	INTEGER(iwp) :: i !< running index over number of processors or number of multigrid level
124	INTEGER(iwp) :: id_inflow_l !< ID indicating processors located at the left inflow boundary
125	INTEGER(iwp) :: id_outflow_l !< local value of id_outflow
126	INTEGER(iwp) :: id_outflow_source_l !< local value of id_outflow_source
127	INTEGER(iwp) :: id_recycling_l !< ID indicating processors located at the recycling plane
128	INTEGER(iwp) :: id_stg_left_l !< left lateral boundary local core id in case of turbulence generator
129	INTEGER(iwp) :: id_stg_north_l !< north lateral boundary local core id in case of turbulence generator
130	INTEGER(iwp) :: id_stg_right_l !< right lateral boundary local core id in case of turbulence generator
131	INTEGER(iwp) :: id_stg_south_l !< south lateral boundary local core id in case of turbulence generator
132	INTEGER(iwp) :: ind(5) !< array containing the subdomain bounds
133	INTEGER(iwp) :: j !< running index, used for various loops
134	INTEGER(iwp) :: k !< number of vertical grid points in different multigrid level
135	INTEGER(iwp) :: maximum_grid_level_l !< maximum number of grid level without switching to PE 0
136	INTEGER(iwp) :: mg_levels_x !< maximum number of grid level allowed along x-direction
137	INTEGER(iwp) :: mg_levels_y !< maximum number of grid level allowed along y-direction
138	INTEGER(iwp) :: mg_levels_z !< maximum number of grid level allowed along z-direction
139	INTEGER(iwp) :: mg_switch_to_pe0_level_l !< maximum number of grid level with switching to PE 0
140	INTEGER(iwp) :: nnx_y !< quotient of number of grid points along x-direction and number of PEs used along y-direction
141	INTEGER(iwp) :: nny_x !< quotient of number of grid points along y-direction and number of PEs used along x-direction
142	INTEGER(iwp) :: nny_z !< quotient of number of grid points along y-direction and number of PEs used along x-direction
143	INTEGER(iwp) :: nnz_x !< quotient of number of grid points along z-direction and number of PEs used along x-direction
144	INTEGER(iwp) :: nnz_y !< quotient of number of grid points along z-direction and number of PEs used along x-direction
145	INTEGER(iwp) :: numproc_sqr !< square root of the number of processors
146	INTEGER(iwp) :: nxl_l !< lower index bound along x-direction on subdomain and different multigrid level
147	INTEGER(iwp) :: nxr_l !< upper index bound along x-direction on subdomain and different multigrid level
148	INTEGER(iwp) :: nyn_l !< lower index bound along y-direction on subdomain and different multigrid level
149	INTEGER(iwp) :: nys_l !< upper index bound along y-direction on subdomain and different multigrid level
150	INTEGER(iwp) :: nzb_l !< lower index bound along z-direction on subdomain and different multigrid level
151	INTEGER(iwp) :: nzt_l !< upper index bound along z-direction on subdomain and different multigrid level
152	!$ INTEGER(iwp) :: omp_get_num_threads !< number of OpenMP threads
153
154	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: ind_all !< dummy array containing index bounds on subdomain, used for gathering
155	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nxlf !< lower index bound allong x-direction for every PE
156	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nxrf !< upper index bound allong x-direction for every PE
157	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nynf !< lower index bound allong y-direction for every PE
158	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nysf !< lower index bound allong y-direction for every PE
159
160	INTEGER(iwp), DIMENSION(2) :: pdims_remote !< number of PEs used for coupled model (only in atmospher-ocean coupling)
161	INTEGER(iwp) :: lcoord(2) !< PE coordinates of left neighbor along x and y
162	INTEGER(iwp) :: rcoord(2) !< PE coordinates of right neighbor along x and y
163
164	!
165	!-- Get the number of OpenMP threads
166	!$OMP PARALLEL
167	!$ threads_per_task = omp_get_num_threads()
168	!$OMP END PARALLEL
169
170
171	#if defined( __parallel )
172
173	CALL location_message( 'creating virtual PE grids + MPI derived data types', 'start' )
174
175	!
176	!-- Determine the processor topology or check it, if prescribed by the user
177	IF ( npex == -1 .AND. npey == -1 ) THEN
178
179	!
180	!-- Automatic determination of the topology
181	numproc_sqr = SQRT( REAL( numprocs, KIND=wp ) )
182	pdims(1) = MAX( numproc_sqr , 1 )
183	DO WHILE ( MOD( numprocs , pdims(1) ) /= 0 )
184	pdims(1) = pdims(1) - 1
185	ENDDO
186	pdims(2) = numprocs / pdims(1)
187
188	ELSEIF ( npex /= -1 .AND. npey /= -1 ) THEN
189
190	!
191	!-- Prescribed by user. Number of processors on the prescribed topology
192	!-- must be equal to the number of PEs available to the job
193	IF ( ( npex * npey ) /= numprocs ) THEN
194	WRITE( message_string, * ) 'number of PEs of the prescribed ', &
195	'topology (', npex*npey,') does not match & the number of ', &
196	'PEs available to the job (', numprocs, ')'
197	CALL message( 'init_pegrid', 'PA0221', 1, 2, 0, 6, 0 )
198	ENDIF
199	pdims(1) = npex
200	pdims(2) = npey
201
202	ELSE
203	!
204	!-- If the processor topology is prescribed by the user, the number of
205	!-- PEs must be given in both directions
206	message_string = 'if the processor topology is prescribed by th' // &
207	'e user & both values of "npex" and "npey" must be given' // &
208	' in the &NAMELIST-parameter file'
209	CALL message( 'init_pegrid', 'PA0222', 1, 2, 0, 6, 0 )
210
211	ENDIF
212
213	!
214	!-- If necessary, set horizontal boundary conditions to non-cyclic
215	IF ( bc_lr /= 'cyclic' ) cyclic(1) = .FALSE.
216	IF ( bc_ns /= 'cyclic' ) cyclic(2) = .FALSE.
217
218
219	!
220	!-- Create the virtual processor grid
221	CALL MPI_CART_CREATE( comm_palm, ndim, pdims, cyclic, reorder, &
222	comm2d, ierr )
223	CALL MPI_COMM_RANK( comm2d, myid, ierr )
224	WRITE (myid_char,'(''_'',I6.6)') myid
225
226	CALL MPI_CART_COORDS( comm2d, myid, ndim, pcoord, ierr )
227	CALL MPI_CART_SHIFT( comm2d, 0, 1, pleft, pright, ierr )
228	CALL MPI_CART_SHIFT( comm2d, 1, 1, psouth, pnorth, ierr )
229	!
230	!-- In case of cyclic boundary conditions, a y-shift at the boundaries in
231	!-- x-direction can be introduced via parameter y_shift. The shift is done
232	!-- by modifying the processor grid in such a way that processors located
233	!-- at the x-boundary communicate across it to processors with y-coordinate
234	!-- shifted by y_shift relative to their own. This feature can not be used
235	!-- in combination with an fft pressure solver. It has been implemented to
236	!-- counter the effect of streak structures in case of cyclic boundary
237	!-- conditions. For a description of these see Munters
238	!-- (2016; dx.doi.org/10.1063/1.4941912)
239	!--
240	!-- Get coordinates of left and right neighbor on PE grid
241	IF ( y_shift /= 0 ) THEN
242	IF ( bc_lr == 'cyclic' ) THEN
243	IF ( TRIM( psolver ) /= 'multigrid' .AND. &
244	TRIM( psolver ) /= 'multigrid_noopt') &
245	THEN
246	message_string = 'y_shift /= 0 requires a multigrid pressure solver '
247	CALL message( 'check_parameters', 'PA0468', 1, 2, 0, 6, 0 )
248	ENDIF
249
250	CALL MPI_CART_COORDS( comm2d, pright, ndim, rcoord, ierr )
251	CALL MPI_CART_COORDS( comm2d, pleft, ndim, lcoord, ierr )
252
253	!
254	!-- If the x(y)-coordinate of the right (left) neighbor is smaller (greater)
255	!-- than that of the calling process, then the calling process is located on
256	!-- the right (left) boundary of the processor grid. In that case,
257	!-- the y-coordinate of that neighbor is increased (decreased) by y_shift.
258	!-- The rank of the process with that coordinate is then inquired and the
259	!-- neighbor rank for MPI_SENDRECV, pright (pleft) is set to it.
260	!-- In this way, the calling process receives a new right (left) neighbor
261	!-- for all future MPI_SENDRECV calls. That neighbor has a y-coordinate
262	!-- of y+(-)y_shift, where y is the original right (left) neighbor's
263	!-- y-coordinate. The modulo-operation ensures that if the neighbor's
264	!-- y-coordinate exceeds the grid-boundary, it will be relocated to
265	!-- the opposite part of the grid cyclicly.
266	IF ( rcoord(1) < pcoord(1) ) THEN
267	rcoord(2) = MODULO( rcoord(2) + y_shift, pdims(2) )
268	CALL MPI_CART_RANK( comm2d, rcoord, pright, ierr )
269	ENDIF
270
271	IF ( lcoord(1) > pcoord(1) ) THEN
272	lcoord(2) = MODULO( lcoord(2) - y_shift, pdims(2) )
273	CALL MPI_CART_RANK( comm2d, lcoord, pleft, ierr )
274	ENDIF
275
276	ELSE
277	!
278	!-- y-shift for non-cyclic boundary conditions is only implemented
279	!-- for the turbulence recycling method in inflow_turbulence.f90
280	IF ( .NOT. turbulent_inflow ) THEN
281	message_string = 'y_shift /= 0 is only allowed for cyclic ' // &
282	'boundary conditions in both directions ' // &
283	'or with turbulent_inflow == .TRUE.'
284	CALL message( 'check_parameters', 'PA0467', 1, 2, 0, 6, 0 )
285	ENDIF
286	ENDIF
287	ENDIF
288	!
289	!-- Vertical nesting: store four lists that identify partner ranks to exchange
290	!-- data
291	IF ( vnested ) CALL vnest_init_pegrid_rank
292
293	!
294	!-- Determine sub-topologies for transpositions
295	!-- Transposition from z to x:
296	remain_dims(1) = .TRUE.
297	remain_dims(2) = .FALSE.
298	CALL MPI_CART_SUB( comm2d, remain_dims, comm1dx, ierr )
299	CALL MPI_COMM_RANK( comm1dx, myidx, ierr )
300	!
301	!-- Transposition from x to y
302	remain_dims(1) = .FALSE.
303	remain_dims(2) = .TRUE.
304	CALL MPI_CART_SUB( comm2d, remain_dims, comm1dy, ierr )
305	CALL MPI_COMM_RANK( comm1dy, myidy, ierr )
306
307
308	!
309	!-- Calculate array bounds along x-direction for every PE.
310	ALLOCATE( nxlf(0:pdims(1)-1), nxrf(0:pdims(1)-1), nynf(0:pdims(2)-1), &
311	nysf(0:pdims(2)-1) )
312
313	IF ( MOD( nx+1 , pdims(1) ) /= 0 ) THEN
314	WRITE( message_string, * ) 'x-direction: gridpoint number (',nx+1,') ', &
315	'is not an& integral multiple of the number',&
316	' of processors (',pdims(1),')'
317	CALL message( 'init_pegrid', 'PA0225', 1, 2, 0, 6, 0 )
318	ELSE
319	nnx = ( nx + 1 ) / pdims(1)
320	ENDIF
321
322	!
323	!-- Left and right array bounds, number of gridpoints
324	DO i = 0, pdims(1)-1
325	nxlf(i) = i * nnx
326	nxrf(i) = ( i + 1 ) * nnx - 1
327	ENDDO
328
329	!
330	!-- Calculate array bounds in y-direction for every PE.
331	IF ( MOD( ny+1 , pdims(2) ) /= 0 ) THEN
332	WRITE( message_string, * ) 'y-direction: gridpoint number (',ny+1,') ', &
333	'is not an& integral multiple of the number',&
334	' of processors (',pdims(2),')'
335	CALL message( 'init_pegrid', 'PA0227', 1, 2, 0, 6, 0 )
336	ELSE
337	nny = ( ny + 1 ) / pdims(2)
338	ENDIF
339
340	!
341	!-- South and north array bounds
342	DO j = 0, pdims(2)-1
343	nysf(j) = j * nny
344	nynf(j) = ( j + 1 ) * nny - 1
345	ENDDO
346
347	!
348	!-- Local array bounds of the respective PEs
349	nxl = nxlf(pcoord(1))
350	nxr = nxrf(pcoord(1))
351	nys = nysf(pcoord(2))
352	nyn = nynf(pcoord(2))
353	nzb = 0
354	nzt = nz
355	nnz = nz
356
357	!
358	!-- Set switches to define if the PE is situated at the border of the virtual
359	!-- processor grid
360	IF ( nxl == 0 ) left_border_pe = .TRUE.
361	IF ( nxr == nx ) right_border_pe = .TRUE.
362	IF ( nys == 0 ) south_border_pe = .TRUE.
363	IF ( nyn == ny ) north_border_pe = .TRUE.
364
365	!
366	!-- Calculate array bounds and gridpoint numbers for the transposed arrays
367	!-- (needed in the pressure solver)
368	!-- For the transposed arrays, cyclic boundaries as well as top and bottom
369	!-- boundaries are omitted, because they are obstructive to the transposition
370
371	!
372	!-- 1. transposition z --> x
373	!-- This transposition is not neccessary in case of a 1d-decomposition along x
374	IF ( psolver == 'poisfft' .OR. calculate_spectra ) THEN
375
376	IF ( pdims(2) /= 1 ) THEN
377	IF ( MOD( nz , pdims(1) ) /= 0 ) THEN
378	WRITE( message_string, * ) 'transposition z --> x:& ', &
379	'nz=',nz,' is not an integral multiple ',&
380	'of pdims(1)=',pdims(1)
381	CALL message( 'init_pegrid', 'PA0230', 1, 2, 0, 6, 0 )
382	ENDIF
383	ENDIF
384
385	nys_x = nys
386	nyn_x = nyn
387	nny_x = nny
388	nnz_x = nz / pdims(1)
389	nzb_x = 1 + myidx * nnz_x
390	nzt_x = ( myidx + 1 ) * nnz_x
391	sendrecvcount_zx = nnx * nny * nnz_x
392
393	ENDIF
394
395
396	IF ( psolver == 'poisfft' ) THEN
397	!
398	!-- 2. transposition x --> y
399	IF ( MOD( nx+1 , pdims(2) ) /= 0 ) THEN
400	WRITE( message_string, * ) 'transposition x --> y:& ', &
401	'nx+1=',nx+1,' is not an integral ', &
402	'multiple of pdims(2)=',pdims(2)
403	CALL message( 'init_pegrid', 'PA0231', 1, 2, 0, 6, 0 )
404	ENDIF
405
406	nnz_y = nnz_x
407	nzb_y = nzb_x
408	nzt_y = nzt_x
409	nnx_y = (nx+1) / pdims(2)
410	nxl_y = myidy * nnx_y
411	nxr_y = ( myidy + 1 ) * nnx_y - 1
412	sendrecvcount_xy = nnx_y * nny_x * nnz_y
413	!
414	!-- 3. transposition y --> z
415	!-- (ELSE: x --> y in case of 1D-decomposition along x)
416	nxl_z = nxl_y
417	nxr_z = nxr_y
418	nny_z = (ny+1) / pdims(1)
419	nys_z = myidx * nny_z
420	nyn_z = ( myidx + 1 ) * nny_z - 1
421	sendrecvcount_yz = nnx_y * nny_z * nnz_y
422
423	IF ( pdims(2) /= 1 ) THEN
424	!
425	!-- y --> z
426	!-- This transposition is not neccessary in case of a 1d-decomposition
427	!-- along x, except that the uptream-spline method is switched on
428	IF ( MOD( ny+1 , pdims(1) ) /= 0 ) THEN
429	WRITE( message_string, * ) 'transposition y --> z:& ', &
430	'ny+1=',ny+1,' is not an integral ', &
431	'multiple of pdims(1)=',pdims(1)
432	CALL message( 'init_pegrid', 'PA0232', 1, 2, 0, 6, 0 )
433	ENDIF
434
435	ELSE
436	!
437	!-- x --> y
438	!-- This condition must be fulfilled for a 1D-decomposition along x
439	IF ( MOD( ny+1 , pdims(1) ) /= 0 ) THEN
440	WRITE( message_string, * ) 'transposition x --> y:& ', &
441	'ny+1=',ny+1,' is not an integral ', &
442	'multiple of pdims(1)=',pdims(1)
443	CALL message( 'init_pegrid', 'PA0233', 1, 2, 0, 6, 0 )
444	ENDIF
445
446	ENDIF
447
448	ENDIF
449
450	!
451	!-- Indices for direct transpositions z --> y (used for calculating spectra)
452	IF ( calculate_spectra ) THEN
453	IF ( MOD( nz, pdims(2) ) /= 0 ) THEN
454	WRITE( message_string, * ) 'direct transposition z --> y (needed ', &
455	'for spectra):& nz=',nz,' is not an ', &
456	'integral multiple of pdims(2)=',pdims(2)
457	CALL message( 'init_pegrid', 'PA0234', 1, 2, 0, 6, 0 )
458	ELSE
459	nxl_yd = nxl
460	nxr_yd = nxr
461	nzb_yd = 1 + myidy * ( nz / pdims(2) )
462	nzt_yd = ( myidy + 1 ) * ( nz / pdims(2) )
463	sendrecvcount_zyd = nnx * nny * ( nz / pdims(2) )
464	ENDIF
465	ENDIF
466
467	IF ( psolver == 'poisfft' .OR. calculate_spectra ) THEN
468	!
469	!-- Indices for direct transpositions y --> x
470	!-- (they are only possible in case of a 1d-decomposition along x)
471	IF ( pdims(2) == 1 ) THEN
472	nny_x = nny / pdims(1)
473	nys_x = myid * nny_x
474	nyn_x = ( myid + 1 ) * nny_x - 1
475	nzb_x = 1
476	nzt_x = nz
477	sendrecvcount_xy = nnx * nny_x * nz
478	ENDIF
479
480	ENDIF
481
482	IF ( psolver == 'poisfft' ) THEN
483	!
484	!-- Indices for direct transpositions x --> y
485	!-- (they are only possible in case of a 1d-decomposition along y)
486	IF ( pdims(1) == 1 ) THEN
487	nnx_y = nnx / pdims(2)
488	nxl_y = myid * nnx_y
489	nxr_y = ( myid + 1 ) * nnx_y - 1
490	nzb_y = 1
491	nzt_y = nz
492	sendrecvcount_xy = nnx_y * nny * nz
493	ENDIF
494
495	ENDIF
496
497	!
498	!-- Arrays for storing the array bounds are needed any more
499	DEALLOCATE( nxlf , nxrf , nynf , nysf )
500
501
502	!
503	!-- Collect index bounds from other PEs (to be written to restart file later)
504	ALLOCATE( hor_index_bounds(4,0:numprocs-1) )
505
506	IF ( myid == 0 ) THEN
507
508	hor_index_bounds(1,0) = nxl
509	hor_index_bounds(2,0) = nxr
510	hor_index_bounds(3,0) = nys
511	hor_index_bounds(4,0) = nyn
512
513	!
514	!-- Receive data from all other PEs
515	DO i = 1, numprocs-1
516	CALL MPI_RECV( ibuf, 4, MPI_INTEGER, i, MPI_ANY_TAG, comm2d, status, &
517	ierr )
518	hor_index_bounds(:,i) = ibuf(1:4)
519	ENDDO
520
521	ELSE
522	!
523	!-- Send index bounds to PE0
524	ibuf(1) = nxl
525	ibuf(2) = nxr
526	ibuf(3) = nys
527	ibuf(4) = nyn
528	CALL MPI_SEND( ibuf, 4, MPI_INTEGER, 0, myid, comm2d, ierr )
529
530	ENDIF
531
532
533	#if defined( __print )
534	!
535	!-- Control output
536	IF ( myid == 0 ) THEN
537	PRINT, ' processor topology *'
538	PRINT*, ' '
539	PRINT*, 'myid pcoord left right south north idx idy nxl: nxr',&
540	&' nys: nyn'
541	PRINT*, '------------------------------------------------------------',&
542	&'-----------'
543	WRITE (*,1000) 0, pcoord(1), pcoord(2), pleft, pright, psouth, pnorth, &
544	myidx, myidy, nxl, nxr, nys, nyn
545	1000 FORMAT (I4,2X,'(',I3,',',I3,')',3X,I4,2X,I4,3X,I4,2X,I4,2X,I3,1X,I3, &
546	2(2X,I4,':',I4))
547
548	!
549	!-- Receive data from the other PEs
550	DO i = 1,numprocs-1
551	CALL MPI_RECV( ibuf, 12, MPI_INTEGER, i, MPI_ANY_TAG, comm2d, status, &
552	ierr )
553	WRITE (*,1000) i, ( ibuf(j) , j = 1,12 )
554	ENDDO
555	ELSE
556
557	!
558	!-- Send data to PE0
559	ibuf(1) = pcoord(1); ibuf(2) = pcoord(2); ibuf(3) = pleft
560	ibuf(4) = pright; ibuf(5) = psouth; ibuf(6) = pnorth; ibuf(7) = myidx
561	ibuf(8) = myidy; ibuf(9) = nxl; ibuf(10) = nxr; ibuf(11) = nys
562	ibuf(12) = nyn
563	CALL MPI_SEND( ibuf, 12, MPI_INTEGER, 0, myid, comm2d, ierr )
564	ENDIF
565	#endif
566
567	!
568	!-- Determine the number of ghost point layers
569	IF ( scalar_advec == 'ws-scheme' .OR. &
570	momentum_advec == 'ws-scheme' .OR. nested_run ) THEN
571	nbgp = 3
572	ELSE
573	nbgp = 1
574	ENDIF
575
576	!
577	!-- Check that the number of computational grid points is not smaller than the number of
578	!-- ghost points.
579	IF ( nnx < nbgp ) THEN
580	WRITE( message_string, * ) 'number of subdomain grid points along x (', nnx, ') is smaller',&
581	'than the number of ghost points (', nbgp, ')'
582	CALL message( 'init_pegrid', 'PA0682', 1, 2, 0, 6, 0 )
583	ENDIF
584	IF ( nny < nbgp ) THEN
585	WRITE( message_string, * ) 'number of subdomain grid points along y (', nny, ') is smaller',&
586	'than the number of ghost points (', nbgp, ')'
587	CALL message( 'init_pegrid', 'PA0683', 1, 2, 0, 6, 0 )
588	ENDIF
589
590	!
591	!-- Create a new MPI derived datatype for the exchange of surface (xy) data,
592	!-- which is needed for coupled atmosphere-ocean runs.
593	!-- First, calculate number of grid points of an xy-plane.
594	ngp_xy = ( nxr - nxl + 1 + 2 * nbgp ) * ( nyn - nys + 1 + 2 * nbgp )
595	CALL MPI_TYPE_VECTOR( ngp_xy, 1, nzt-nzb+2, MPI_REAL, type_xy, ierr )
596	CALL MPI_TYPE_COMMIT( type_xy, ierr )
597
598	IF ( TRIM( coupling_mode ) /= 'uncoupled' .AND. .NOT. vnested ) THEN
599
600	!
601	!-- Pass the number of grid points of the atmosphere model to
602	!-- the ocean model and vice versa
603	IF ( coupling_mode == 'atmosphere_to_ocean' ) THEN
604
605	nx_a = nx
606	ny_a = ny
607
608	IF ( myid == 0 ) THEN
609
610	CALL MPI_SEND( nx_a, 1, MPI_INTEGER, numprocs, 1, comm_inter, &
611	ierr )
612	CALL MPI_SEND( ny_a, 1, MPI_INTEGER, numprocs, 2, comm_inter, &
613	ierr )
614	CALL MPI_SEND( pdims, 2, MPI_INTEGER, numprocs, 3, comm_inter, &
615	ierr )
616	CALL MPI_RECV( nx_o, 1, MPI_INTEGER, numprocs, 4, comm_inter, &
617	status, ierr )
618	CALL MPI_RECV( ny_o, 1, MPI_INTEGER, numprocs, 5, comm_inter, &
619	status, ierr )
620	CALL MPI_RECV( pdims_remote, 2, MPI_INTEGER, numprocs, 6, &
621	comm_inter, status, ierr )
622	ENDIF
623
624	CALL MPI_BCAST( nx_o, 1, MPI_INTEGER, 0, comm2d, ierr )
625	CALL MPI_BCAST( ny_o, 1, MPI_INTEGER, 0, comm2d, ierr )
626	CALL MPI_BCAST( pdims_remote, 2, MPI_INTEGER, 0, comm2d, ierr )
627
628	ELSEIF ( coupling_mode == 'ocean_to_atmosphere' ) THEN
629
630	nx_o = nx
631	ny_o = ny
632
633	IF ( myid == 0 ) THEN
634
635	CALL MPI_RECV( nx_a, 1, MPI_INTEGER, 0, 1, comm_inter, status, &
636	ierr )
637	CALL MPI_RECV( ny_a, 1, MPI_INTEGER, 0, 2, comm_inter, status, &
638	ierr )
639	CALL MPI_RECV( pdims_remote, 2, MPI_INTEGER, 0, 3, comm_inter, &
640	status, ierr )
641	CALL MPI_SEND( nx_o, 1, MPI_INTEGER, 0, 4, comm_inter, ierr )
642	CALL MPI_SEND( ny_o, 1, MPI_INTEGER, 0, 5, comm_inter, ierr )
643	CALL MPI_SEND( pdims, 2, MPI_INTEGER, 0, 6, comm_inter, ierr )
644	ENDIF
645
646	CALL MPI_BCAST( nx_a, 1, MPI_INTEGER, 0, comm2d, ierr)
647	CALL MPI_BCAST( ny_a, 1, MPI_INTEGER, 0, comm2d, ierr)
648	CALL MPI_BCAST( pdims_remote, 2, MPI_INTEGER, 0, comm2d, ierr)
649
650	ENDIF
651
652	ngp_a = ( nx_a+1 + 2 * nbgp ) * ( ny_a+1 + 2 * nbgp )
653	ngp_o = ( nx_o+1 + 2 * nbgp ) * ( ny_o+1 + 2 * nbgp )
654
655	!
656	!-- Determine if the horizontal grid and the number of PEs in ocean and
657	!-- atmosphere is same or not
658	IF ( nx_o == nx_a .AND. ny_o == ny_a .AND. &
659	pdims(1) == pdims_remote(1) .AND. pdims(2) == pdims_remote(2) ) &
660	THEN
661	coupling_topology = 0
662	ELSE
663	coupling_topology = 1
664	ENDIF
665
666	!
667	!-- Determine the target PEs for the exchange between ocean and
668	!-- atmosphere (comm2d)
669	IF ( coupling_topology == 0 ) THEN
670	!
671	!-- In case of identical topologies, every atmosphere PE has exactly one
672	!-- ocean PE counterpart and vice versa
673	IF ( TRIM( coupling_mode ) == 'atmosphere_to_ocean' ) THEN
674	target_id = myid + numprocs
675	ELSE
676	target_id = myid
677	ENDIF
678
679	ELSE
680	!
681	!-- In case of nonequivalent topology in ocean and atmosphere only for
682	!-- PE0 in ocean and PE0 in atmosphere a target_id is needed, since
683	!-- data echxchange between ocean and atmosphere will be done only
684	!-- between these PEs.
685	IF ( myid == 0 ) THEN
686
687	IF ( TRIM( coupling_mode ) == 'atmosphere_to_ocean' ) THEN
688	target_id = numprocs
689	ELSE
690	target_id = 0
691	ENDIF
692
693	ENDIF
694
695	ENDIF
696
697	ENDIF
698
699	!
700	!-- Store partner grid point co-ordinates as lists.
701	!-- Create custom MPI vector datatypes for contiguous data transfer
702	IF ( vnested ) CALL vnest_init_pegrid_domain
703
704	#else
705
706	!
707	!-- Array bounds when running on a single PE (respectively a non-parallel
708	!-- machine)
709	nxl = 0
710	nxr = nx
711	nnx = nxr - nxl + 1
712	nys = 0
713	nyn = ny
714	nny = nyn - nys + 1
715	nzb = 0
716	nzt = nz
717	nnz = nz
718
719	ALLOCATE( hor_index_bounds(4,0:0) )
720	hor_index_bounds(1,0) = nxl
721	hor_index_bounds(2,0) = nxr
722	hor_index_bounds(3,0) = nys
723	hor_index_bounds(4,0) = nyn
724
725	!
726	!-- Array bounds for the pressure solver (in the parallel code, these bounds
727	!-- are the ones for the transposed arrays)
728	nys_x = nys
729	nyn_x = nyn
730	nzb_x = nzb + 1
731	nzt_x = nzt
732
733	nxl_y = nxl
734	nxr_y = nxr
735	nzb_y = nzb + 1
736	nzt_y = nzt
737
738	nxl_z = nxl
739	nxr_z = nxr
740	nys_z = nys
741	nyn_z = nyn
742
743	#endif
744
745	!
746	!-- Calculate number of grid levels necessary for the multigrid poisson solver
747	!-- as well as the gridpoint indices on each level
748	IF ( psolver(1:9) == 'multigrid' ) THEN
749
750	!
751	!-- First calculate number of possible grid levels for the subdomains
752	mg_levels_x = 1
753	mg_levels_y = 1
754	mg_levels_z = 1
755
756	i = nnx
757	DO WHILE ( MOD( i, 2 ) == 0 .AND. i /= 2 )
758	i = i / 2
759	mg_levels_x = mg_levels_x + 1
760	ENDDO
761
762	j = nny
763	DO WHILE ( MOD( j, 2 ) == 0 .AND. j /= 2 )
764	j = j / 2
765	mg_levels_y = mg_levels_y + 1
766	ENDDO
767
768	k = nz ! do not use nnz because it might be > nz due to transposition
769	! requirements
770	DO WHILE ( MOD( k, 2 ) == 0 .AND. k /= 2 )
771	k = k / 2
772	mg_levels_z = mg_levels_z + 1
773	ENDDO
774	!
775	!-- The optimized MG-solver does not allow odd values for nz at the coarsest
776	!-- grid level
777	IF ( TRIM( psolver ) /= 'multigrid_noopt' ) THEN
778	IF ( MOD( k, 2 ) /= 0 ) mg_levels_z = mg_levels_z - 1
779	!
780	!-- An odd value of nz does not work. The finest level must have an even
781	!-- value.
782	IF ( mg_levels_z == 0 ) THEN
783	message_string = 'optimized multigrid method requires nz to be even'
784	CALL message( 'init_pegrid', 'PA0495', 1, 2, 0, 6, 0 )
785	ENDIF
786	ENDIF
787
788	maximum_grid_level = MIN( mg_levels_x, mg_levels_y, mg_levels_z )
789	!
790	!-- Check if subdomain sizes prevents any coarsening.
791	!-- This case, the maximum number of grid levels is 1, i.e. effectively
792	!-- a Gauss-Seidel scheme is applied rather than a multigrid approach.
793	!-- Give a warning in this case.
794	IF ( maximum_grid_level == 1 .AND. mg_switch_to_pe0_level == -1 ) THEN
795	message_string = 'No grid coarsening possible, multigrid ' // &
796	'approach effectively reduces to a Gauss-Seidel ' //&
797	'scheme.'
798
799	CALL message( 'poismg', 'PA0648', 0, 1, 0, 6, 0 )
800	ENDIF
801
802	!
803	!-- Find out, if the total domain allows more levels. These additional
804	!-- levels are identically processed on all PEs.
805	IF ( numprocs > 1 .AND. mg_switch_to_pe0_level /= -1 ) THEN
806
807	IF ( mg_levels_z > MIN( mg_levels_x, mg_levels_y ) ) THEN
808
809	mg_switch_to_pe0_level_l = maximum_grid_level
810
811	mg_levels_x = 1
812	mg_levels_y = 1
813
814	i = nx+1
815	DO WHILE ( MOD( i, 2 ) == 0 .AND. i /= 2 )
816	i = i / 2
817	mg_levels_x = mg_levels_x + 1
818	ENDDO
819
820	j = ny+1
821	DO WHILE ( MOD( j, 2 ) == 0 .AND. j /= 2 )
822	j = j / 2
823	mg_levels_y = mg_levels_y + 1
824	ENDDO
825
826	maximum_grid_level_l = MIN( mg_levels_x, mg_levels_y, mg_levels_z )
827
828	IF ( maximum_grid_level_l > mg_switch_to_pe0_level_l ) THEN
829	mg_switch_to_pe0_level_l = maximum_grid_level_l - &
830	mg_switch_to_pe0_level_l + 1
831	ELSE
832	mg_switch_to_pe0_level_l = 0
833	ENDIF
834
835	ELSE
836
837	mg_switch_to_pe0_level_l = 0
838	maximum_grid_level_l = maximum_grid_level
839
840	ENDIF
841
842	!
843	!-- Use switch level calculated above only if it is not pre-defined
844	!-- by user
845	IF ( mg_switch_to_pe0_level == 0 ) THEN
846	IF ( mg_switch_to_pe0_level_l /= 0 ) THEN
847	mg_switch_to_pe0_level = mg_switch_to_pe0_level_l
848	maximum_grid_level = maximum_grid_level_l
849	ENDIF
850
851	ELSE
852	!
853	!-- Check pre-defined value and reset to default, if neccessary
854	IF ( mg_switch_to_pe0_level < mg_switch_to_pe0_level_l .OR. &
855	mg_switch_to_pe0_level >= maximum_grid_level_l ) THEN
856	message_string = 'mg_switch_to_pe0_level ' // &
857	'out of range and reset to 0'
858	CALL message( 'init_pegrid', 'PA0235', 0, 1, 0, 6, 0 )
859	mg_switch_to_pe0_level = 0
860	ELSE
861	!
862	!-- Use the largest number of possible levels anyway and recalculate
863	!-- the switch level to this largest number of possible values
864	maximum_grid_level = maximum_grid_level_l
865
866	ENDIF
867
868	ENDIF
869
870	ENDIF
871
872	ALLOCATE( grid_level_count(maximum_grid_level), &
873	nxl_mg(0:maximum_grid_level), nxr_mg(0:maximum_grid_level), &
874	nyn_mg(0:maximum_grid_level), nys_mg(0:maximum_grid_level), &
875	nzt_mg(0:maximum_grid_level) )
876
877	grid_level_count = 0
878	!
879	!-- Index zero required as dummy due to definition of arrays f2 and p2 in
880	!-- recursive subroutine next_mg_level
881	nxl_mg(0) = 0; nxr_mg(0) = 0; nyn_mg(0) = 0; nys_mg(0) = 0; nzt_mg(0) = 0
882
883	nxl_l = nxl; nxr_l = nxr; nys_l = nys; nyn_l = nyn; nzt_l = nzt
884
885	DO i = maximum_grid_level, 1 , -1
886
887	IF ( i == mg_switch_to_pe0_level ) THEN
888	#if defined( __parallel )
889	!
890	!-- Save the grid size of the subdomain at the switch level, because
891	!-- it is needed in poismg.
892	ind(1) = nxl_l; ind(2) = nxr_l
893	ind(3) = nys_l; ind(4) = nyn_l
894	ind(5) = nzt_l
895	ALLOCATE( ind_all(5*numprocs), mg_loc_ind(5,0:numprocs-1) )
896	CALL MPI_ALLGATHER( ind, 5, MPI_INTEGER, ind_all, 5, &
897	MPI_INTEGER, comm2d, ierr )
898	DO j = 0, numprocs-1
899	DO k = 1, 5
900	mg_loc_ind(k,j) = ind_all(k+j*5)
901	ENDDO
902	ENDDO
903	DEALLOCATE( ind_all )
904	!
905	!-- Calculate the grid size of the total domain
906	nxr_l = ( nxr_l-nxl_l+1 ) * pdims(1) - 1
907	nxl_l = 0
908	nyn_l = ( nyn_l-nys_l+1 ) * pdims(2) - 1
909	nys_l = 0
910	!
911	!-- The size of this gathered array must not be larger than the
912	!-- array tend, which is used in the multigrid scheme as a temporary
913	!-- array. Therefore the subdomain size of an PE is calculated and
914	!-- the size of the gathered grid. These values are used in
915	!-- routines pres and poismg
916	subdomain_size = ( nxr - nxl + 2 * nbgp + 1 ) * &
917	( nyn - nys + 2 * nbgp + 1 ) * ( nzt - nzb + 2 )
918	gathered_size = ( nxr_l - nxl_l + 3 ) * ( nyn_l - nys_l + 3 ) * &
919	( nzt_l - nzb + 2 )
920
921	#else
922	message_string = 'multigrid gather/scatter impossible ' // &
923	'in non parallel mode'
924	CALL message( 'init_pegrid', 'PA0237', 1, 2, 0, 6, 0 )
925	#endif
926	ENDIF
927
928	nxl_mg(i) = nxl_l
929	nxr_mg(i) = nxr_l
930	nys_mg(i) = nys_l
931	nyn_mg(i) = nyn_l
932	nzt_mg(i) = nzt_l
933
934	nxl_l = nxl_l / 2
935	nxr_l = nxr_l / 2
936	nys_l = nys_l / 2
937	nyn_l = nyn_l / 2
938	nzt_l = nzt_l / 2
939
940	ENDDO
941
942	!
943	!-- Temporary problem: Currently calculation of maxerror in routine poismg crashes
944	!-- if grid data are collected on PE0 already on the finest grid level.
945	!-- To be solved later.
946	IF ( maximum_grid_level == mg_switch_to_pe0_level ) THEN
947	message_string = 'grid coarsening on subdomain level cannot be performed'
948	CALL message( 'poismg', 'PA0236', 1, 2, 0, 6, 0 )
949	ENDIF
950
951	ELSE
952
953	maximum_grid_level = 0
954
955	ENDIF
956
957	!
958	!-- Default level 0 tells exchange_horiz that all ghost planes have to be
959	!-- exchanged. grid_level is adjusted in poismg, where only one ghost plane
960	!-- is required.
961	grid_level = 0
962
963	#if defined( __parallel )
964	!
965	!-- Gridpoint number for the exchange of ghost points (y-line for 2D-arrays)
966	ngp_y = nyn - nys + 1 + 2 * nbgp
967
968	!
969	!-- Define new MPI derived datatypes for the exchange of ghost points in
970	!-- x- and y-direction for 2D-arrays (line)
971	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2*nbgp, nbgp, ngp_y, MPI_REAL, type_x, &
972	ierr )
973	CALL MPI_TYPE_COMMIT( type_x, ierr )
974
975	CALL MPI_TYPE_VECTOR( nbgp, ngp_y, ngp_y, MPI_REAL, type_y, ierr )
976	CALL MPI_TYPE_COMMIT( type_y, ierr )
977	!
978	!-- Define new MPI derived datatypes for the exchange of ghost points in
979	!-- x- and y-direction for 2D-INTEGER arrays (line) - on normal grid.
980	!-- Define types for 32-bit and 8-bit Integer. The 8-bit Integer are only
981	!-- required on normal grid, while 32-bit Integer may be also required on
982	!-- coarser grid level in case of multigrid solver.
983	!
984	!-- 8-bit Integer
985	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2*nbgp, nbgp, ngp_y, MPI_BYTE, &
986	type_x_byte, ierr )
987	CALL MPI_TYPE_COMMIT( type_x_byte, ierr )
988
989	CALL MPI_TYPE_VECTOR( nbgp, ngp_y, ngp_y, MPI_BYTE, &
990	type_y_byte, ierr )
991	CALL MPI_TYPE_COMMIT( type_y_byte, ierr )
992	!
993	!-- 32-bit Integer
994	ALLOCATE( type_x_int(0:maximum_grid_level), &
995	type_y_int(0:maximum_grid_level) )
996
997	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2*nbgp, nbgp, ngp_y, MPI_INTEGER, &
998	type_x_int(0), ierr )
999	CALL MPI_TYPE_COMMIT( type_x_int(0), ierr )
1000
1001	CALL MPI_TYPE_VECTOR( nbgp, ngp_y, ngp_y, MPI_INTEGER, type_y_int(0), ierr )
1002	CALL MPI_TYPE_COMMIT( type_y_int(0), ierr )
1003	!
1004	!-- Calculate gridpoint numbers for the exchange of ghost points along x
1005	!-- (yz-plane for 3D-arrays) and define MPI derived data type(s) for the
1006	!-- exchange of ghost points in y-direction (xz-plane).
1007	!-- Do these calculations for the model grid and (if necessary) also
1008	!-- for the coarser grid levels used in the multigrid method
1009	ALLOCATE ( ngp_xz(0:maximum_grid_level), &
1010	ngp_xz_int(0:maximum_grid_level), &
1011	ngp_yz(0:maximum_grid_level), &
1012	ngp_yz_int(0:maximum_grid_level), &
1013	type_xz(0:maximum_grid_level), &
1014	type_xz_int(0:maximum_grid_level), &
1015	type_yz(0:maximum_grid_level), &
1016	type_yz_int(0:maximum_grid_level) )
1017
1018	nxl_l = nxl; nxr_l = nxr; nys_l = nys; nyn_l = nyn; nzb_l = nzb; nzt_l = nzt
1019
1020	!
1021	!-- Discern between the model grid, which needs nbgp ghost points and
1022	!-- grid levels for the multigrid scheme. In the latter case only one
1023	!-- ghost point is necessary.
1024	!-- First definition of MPI-datatypes for exchange of ghost layers on normal
1025	!-- grid. The following loop is needed for data exchange in poismg.f90.
1026	!
1027	!-- Determine number of grid points of yz-layer for exchange
1028	ngp_yz(0) = (nzt - nzb + 2) * (nyn - nys + 1 + 2 * nbgp)
1029
1030	!
1031	!-- Define an MPI-datatype for the exchange of left/right boundaries.
1032	!-- Although data are contiguous in physical memory (which does not
1033	!-- necessarily require an MPI-derived datatype), the data exchange between
1034	!-- left and right PE's using the MPI-derived type is 10% faster than without.
1035	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2nbgp, nbgp(nzt-nzb+2), ngp_yz(0), &
1036	MPI_REAL, type_xz(0), ierr )
1037	CALL MPI_TYPE_COMMIT( type_xz(0), ierr )
1038
1039	CALL MPI_TYPE_VECTOR( nbgp, ngp_yz(0), ngp_yz(0), MPI_REAL, type_yz(0), &
1040	ierr )
1041	CALL MPI_TYPE_COMMIT( type_yz(0), ierr )
1042
1043	!
1044	!-- Define data types for exchange of 3D Integer arrays.
1045	ngp_yz_int(0) = (nzt - nzb + 2) * (nyn - nys + 1 + 2 * nbgp)
1046
1047	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2nbgp, nbgp(nzt-nzb+2), ngp_yz_int(0), &
1048	MPI_INTEGER, type_xz_int(0), ierr )
1049	CALL MPI_TYPE_COMMIT( type_xz_int(0), ierr )
1050
1051	CALL MPI_TYPE_VECTOR( nbgp, ngp_yz_int(0), ngp_yz_int(0), MPI_INTEGER, &
1052	type_yz_int(0), ierr )
1053	CALL MPI_TYPE_COMMIT( type_yz_int(0), ierr )
1054
1055	!
1056	!-- Definition of MPI-datatypes for multigrid method (coarser level grids)
1057	IF ( psolver(1:9) == 'multigrid' ) THEN
1058	!
1059	!-- Definition of MPI-datatyoe as above, but only 1 ghost level is used
1060	DO i = maximum_grid_level, 1 , -1
1061	!
1062	!-- For 3D-exchange on different multigrid level, one ghost point for
1063	!-- REAL arrays, two ghost points for INTEGER arrays
1064	ngp_xz(i) = (nzt_l - nzb_l + 2) * (nxr_l - nxl_l + 3)
1065	ngp_yz(i) = (nzt_l - nzb_l + 2) * (nyn_l - nys_l + 3)
1066
1067	ngp_xz_int(i) = (nzt_l - nzb_l + 2) * (nxr_l - nxl_l + 3)
1068	ngp_yz_int(i) = (nzt_l - nzb_l + 2) * (nyn_l - nys_l + 3)
1069	!
1070	!-- MPI data type for REAL arrays, for xz-layers
1071	CALL MPI_TYPE_VECTOR( nxr_l-nxl_l+3, nzt_l-nzb_l+2, ngp_yz(i), &
1072	MPI_REAL, type_xz(i), ierr )
1073	CALL MPI_TYPE_COMMIT( type_xz(i), ierr )
1074
1075	!
1076	!-- MPI data type for INTEGER arrays, for xz-layers
1077	CALL MPI_TYPE_VECTOR( nxr_l-nxl_l+3, nzt_l-nzb_l+2, ngp_yz_int(i), &
1078	MPI_INTEGER, type_xz_int(i), ierr )
1079	CALL MPI_TYPE_COMMIT( type_xz_int(i), ierr )
1080
1081	!
1082	!-- MPI data type for REAL arrays, for yz-layers
1083	CALL MPI_TYPE_VECTOR( 1, ngp_yz(i), ngp_yz(i), MPI_REAL, type_yz(i), &
1084	ierr )
1085	CALL MPI_TYPE_COMMIT( type_yz(i), ierr )
1086	!
1087	!-- MPI data type for INTEGER arrays, for yz-layers
1088	CALL MPI_TYPE_VECTOR( 1, ngp_yz_int(i), ngp_yz_int(i), MPI_INTEGER, &
1089	type_yz_int(i), ierr )
1090	CALL MPI_TYPE_COMMIT( type_yz_int(i), ierr )
1091
1092
1093	!-- For 2D-exchange of INTEGER arrays on coarser grid level, where 2 ghost
1094	!-- points need to be exchanged. Only required for 32-bit Integer arrays.
1095	CALL MPI_TYPE_VECTOR( nxr_l-nxl_l+5, 2, nyn_l-nys_l+5, MPI_INTEGER, &
1096	type_x_int(i), ierr )
1097	CALL MPI_TYPE_COMMIT( type_x_int(i), ierr )
1098
1099
1100	CALL MPI_TYPE_VECTOR( 2, nyn_l-nys_l+5, nyn_l-nys_l+5, MPI_INTEGER, &
1101	type_y_int(i), ierr )
1102	CALL MPI_TYPE_COMMIT( type_y_int(i), ierr )
1103
1104	nxl_l = nxl_l / 2
1105	nxr_l = nxr_l / 2
1106	nys_l = nys_l / 2
1107	nyn_l = nyn_l / 2
1108	nzt_l = nzt_l / 2
1109
1110	ENDDO
1111
1112	ENDIF
1113
1114	#endif
1115
1116	#if defined( __parallel )
1117	!
1118	!-- Setting of flags for inflow/outflow/nesting conditions.
1119	IF ( pleft == MPI_PROC_NULL ) THEN
1120	IF ( bc_lr == 'dirichlet/radiation' .OR. bc_lr == 'nested' .OR. &
1121	bc_lr == 'nesting_offline' ) THEN
1122	bc_dirichlet_l = .TRUE.
1123	ELSEIF ( bc_lr == 'radiation/dirichlet' ) THEN
1124	bc_radiation_l = .TRUE.
1125	ENDIF
1126	ENDIF
1127
1128	IF ( pright == MPI_PROC_NULL ) THEN
1129	IF ( bc_lr == 'dirichlet/radiation' ) THEN
1130	bc_radiation_r = .TRUE.
1131	ELSEIF ( bc_lr == 'radiation/dirichlet' .OR. bc_lr == 'nested' .OR. &
1132	bc_lr == 'nesting_offline' ) THEN
1133	bc_dirichlet_r = .TRUE.
1134	ENDIF
1135	ENDIF
1136
1137	IF ( psouth == MPI_PROC_NULL ) THEN
1138	IF ( bc_ns == 'dirichlet/radiation' ) THEN
1139	bc_radiation_s = .TRUE.
1140	ELSEIF ( bc_ns == 'radiation/dirichlet' .OR. bc_ns == 'nested' .OR. &
1141	bc_ns == 'nesting_offline' ) THEN
1142	bc_dirichlet_s = .TRUE.
1143	ENDIF
1144	ENDIF
1145
1146	IF ( pnorth == MPI_PROC_NULL ) THEN
1147	IF ( bc_ns == 'dirichlet/radiation' .OR. bc_ns == 'nested' .OR. &
1148	bc_ns == 'nesting_offline' ) THEN
1149	bc_dirichlet_n = .TRUE.
1150	ELSEIF ( bc_ns == 'radiation/dirichlet' ) THEN
1151	bc_radiation_n = .TRUE.
1152	ENDIF
1153	ENDIF
1154	!
1155	!-- In case of synthetic turbulence geneartor determine ids.
1156	!-- Please note, if no forcing or nesting is applied, the generator is applied
1157	!-- only at the left lateral boundary.
1158	IF ( use_syn_turb_gen ) THEN
1159	IF ( bc_dirichlet_l ) THEN
1160	id_stg_left_l = myidx
1161	ELSE
1162	id_stg_left_l = 0
1163	ENDIF
1164	IF ( bc_dirichlet_r ) THEN
1165	id_stg_right_l = myidx
1166	ELSE
1167	id_stg_right_l = 0
1168	ENDIF
1169	IF ( bc_dirichlet_s ) THEN
1170	id_stg_south_l = myidy
1171	ELSE
1172	id_stg_south_l = 0
1173	ENDIF
1174	IF ( bc_dirichlet_n ) THEN
1175	id_stg_north_l = myidy
1176	ELSE
1177	id_stg_north_l = 0
1178	ENDIF
1179
1180	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1181	CALL MPI_ALLREDUCE( id_stg_left_l, id_stg_left, 1, MPI_INTEGER, &
1182	MPI_SUM, comm1dx, ierr )
1183
1184	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1185	CALL MPI_ALLREDUCE( id_stg_right_l, id_stg_right, 1, MPI_INTEGER, &
1186	MPI_SUM, comm1dx, ierr )
1187
1188	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1189	CALL MPI_ALLREDUCE( id_stg_south_l, id_stg_south, 1, MPI_INTEGER, &
1190	MPI_SUM, comm1dy, ierr )
1191
1192	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1193	CALL MPI_ALLREDUCE( id_stg_north_l, id_stg_north, 1, MPI_INTEGER, &
1194	MPI_SUM, comm1dy, ierr )
1195
1196	ENDIF
1197
1198	!
1199	!-- Broadcast the id of the inflow PE
1200	IF ( bc_dirichlet_l ) THEN
1201	id_inflow_l = myidx
1202	ELSE
1203	id_inflow_l = 0
1204	ENDIF
1205	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1206	CALL MPI_ALLREDUCE( id_inflow_l, id_inflow, 1, MPI_INTEGER, MPI_SUM, &
1207	comm1dx, ierr )
1208
1209	!
1210	!-- Broadcast the id of the recycling plane
1211	!-- WARNING: needs to be adjusted in case of inflows other than from left side!
1212	IF ( turbulent_inflow ) THEN
1213
1214	IF ( NINT( recycling_width / dx, KIND=idp ) >= nxl .AND. &
1215	NINT( recycling_width / dx, KIND=idp ) <= nxr ) THEN
1216	id_recycling_l = myidx
1217	ELSE
1218	id_recycling_l = 0
1219	ENDIF
1220	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1221	CALL MPI_ALLREDUCE( id_recycling_l, id_recycling, 1, MPI_INTEGER, MPI_SUM, &
1222	comm1dx, ierr )
1223
1224	ENDIF
1225
1226	!
1227	!-- Broadcast the id of the outflow PE and outflow-source plane
1228	IF ( turbulent_outflow ) THEN
1229
1230	IF ( bc_radiation_r ) THEN
1231	id_outflow_l = myidx
1232	ELSE
1233	id_outflow_l = 0
1234	ENDIF
1235	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1236	CALL MPI_ALLREDUCE( id_outflow_l, id_outflow, 1, MPI_INTEGER, MPI_SUM, &
1237	comm1dx, ierr )
1238
1239	IF ( NINT( outflow_source_plane / dx ) >= nxl .AND. &
1240	NINT( outflow_source_plane / dx ) <= nxr ) THEN
1241	id_outflow_source_l = myidx
1242	ELSE
1243	id_outflow_source_l = 0
1244	ENDIF
1245	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1246	CALL MPI_ALLREDUCE( id_outflow_source_l, id_outflow_source, 1, &
1247	MPI_INTEGER, MPI_SUM, comm1dx, ierr )
1248
1249	ENDIF
1250
1251	CALL location_message( 'creating virtual PE grids + MPI derived data types', 'finished' )
1252
1253	#else
1254	IF ( bc_lr == 'dirichlet/radiation' ) THEN
1255	bc_dirichlet_l = .TRUE.
1256	bc_radiation_r = .TRUE.
1257	ELSEIF ( bc_lr == 'radiation/dirichlet' ) THEN
1258	bc_radiation_l = .TRUE.
1259	bc_dirichlet_r = .TRUE.
1260	ENDIF
1261
1262	IF ( bc_ns == 'dirichlet/radiation' ) THEN
1263	bc_dirichlet_n = .TRUE.
1264	bc_radiation_s = .TRUE.
1265	ELSEIF ( bc_ns == 'radiation/dirichlet' ) THEN
1266	bc_radiation_n = .TRUE.
1267	bc_dirichlet_s = .TRUE.
1268	ENDIF
1269	#endif
1270
1271	!
1272	!-- At the inflow or outflow, u or v, respectively, have to be calculated for
1273	!-- one more grid point.
1274	IF ( bc_dirichlet_l .OR. bc_radiation_l ) THEN
1275	nxlu = nxl + 1
1276	ELSE
1277	nxlu = nxl
1278	ENDIF
1279	IF ( bc_dirichlet_s .OR. bc_radiation_s ) THEN
1280	nysv = nys + 1
1281	ELSE
1282	nysv = nys
1283	ENDIF
1284
1285	!
1286	!-- Allocate wall flag arrays used in the multigrid solver
1287	IF ( psolver(1:9) == 'multigrid' ) THEN
1288
1289	DO i = maximum_grid_level, 1, -1
1290
1291	SELECT CASE ( i )
1292
1293	CASE ( 1 )
1294	ALLOCATE( wall_flags_1(nzb:nzt_mg(i)+1, &
1295	nys_mg(i)-1:nyn_mg(i)+1, &
1296	nxl_mg(i)-1:nxr_mg(i)+1) )
1297
1298	CASE ( 2 )
1299	ALLOCATE( wall_flags_2(nzb:nzt_mg(i)+1, &
1300	nys_mg(i)-1:nyn_mg(i)+1, &
1301	nxl_mg(i)-1:nxr_mg(i)+1) )
1302
1303	CASE ( 3 )
1304	ALLOCATE( wall_flags_3(nzb:nzt_mg(i)+1, &
1305	nys_mg(i)-1:nyn_mg(i)+1, &
1306	nxl_mg(i)-1:nxr_mg(i)+1) )
1307
1308	CASE ( 4 )
1309	ALLOCATE( wall_flags_4(nzb:nzt_mg(i)+1, &
1310	nys_mg(i)-1:nyn_mg(i)+1, &
1311	nxl_mg(i)-1:nxr_mg(i)+1) )
1312
1313	CASE ( 5 )
1314	ALLOCATE( wall_flags_5(nzb:nzt_mg(i)+1, &
1315	nys_mg(i)-1:nyn_mg(i)+1, &
1316	nxl_mg(i)-1:nxr_mg(i)+1) )
1317
1318	CASE ( 6 )
1319	ALLOCATE( wall_flags_6(nzb:nzt_mg(i)+1, &
1320	nys_mg(i)-1:nyn_mg(i)+1, &
1321	nxl_mg(i)-1:nxr_mg(i)+1) )
1322
1323	CASE ( 7 )
1324	ALLOCATE( wall_flags_7(nzb:nzt_mg(i)+1, &
1325	nys_mg(i)-1:nyn_mg(i)+1, &
1326	nxl_mg(i)-1:nxr_mg(i)+1) )
1327
1328	CASE ( 8 )
1329	ALLOCATE( wall_flags_8(nzb:nzt_mg(i)+1, &
1330	nys_mg(i)-1:nyn_mg(i)+1, &
1331	nxl_mg(i)-1:nxr_mg(i)+1) )
1332
1333	CASE ( 9 )
1334	ALLOCATE( wall_flags_9(nzb:nzt_mg(i)+1, &
1335	nys_mg(i)-1:nyn_mg(i)+1, &
1336	nxl_mg(i)-1:nxr_mg(i)+1) )
1337
1338	CASE ( 10 )
1339	ALLOCATE( wall_flags_10(nzb:nzt_mg(i)+1, &
1340	nys_mg(i)-1:nyn_mg(i)+1, &
1341	nxl_mg(i)-1:nxr_mg(i)+1) )
1342
1343	CASE DEFAULT
1344	message_string = 'more than 10 multigrid levels'
1345	CALL message( 'init_pegrid', 'PA0238', 1, 2, 0, 6, 0 )
1346
1347	END SELECT
1348
1349	ENDDO
1350
1351	ENDIF
1352
1353	END SUBROUTINE init_pegrid

Note: See TracBrowser for help on using the repository browser.

Download in other formats:

| Impressum | ©Leibniz Universität Hannover |