Home

Context Navigation

source: palm/trunk/SOURCE/init_pegrid.f90 @ 4454

Last change on this file since 4454 was 4444, checked in by raasch, 5 years ago
bugfix: cpp-directives for serial mode added
Property svn:keywords set to `Id`
File size: 51.6 KB

Rev	Line
[1682]	1	!> @file init_pegrid.f90
[2000]	2	!------------------------------------------------------------------------------!
[2696]	3	! This file is part of the PALM model system.
[1036]	4	!
[2000]	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.
[1036]	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	!
[4360]	17	! Copyright 1997-2020 Leibniz Universitaet Hannover
[2000]	18	!------------------------------------------------------------------------------!
[1036]	19	!
[254]	20	! Current revisions:
[1322]	21	! ------------------
[2201]	22	!
[3553]	23	!
[1321]	24	! Former revisions:
	25	! -----------------
	26	! $Id: init_pegrid.f90 4444 2020-03-05 15:59:50Z forkel $
[4444]	27	! bugfix: cpp-directives for serial mode added
	28	!
	29	! 4360 2020-01-07 11:25:50Z suehring
[4301]	30	! changed message PA0467
	31	!
	32	! 4264 2019-10-15 16:00:23Z scharf
[4264]	33	! corrected error message string
	34	!
	35	! 4241 2019-09-27 06:32:47Z raasch
[4241]	36	! Check added to ensure that subdomain grid has at least the size as given by the number
	37	! of ghost points
	38	!
	39	! 4182 2019-08-22 15:20:23Z scharf
[4182]	40	! Corrected "Former revisions" section
	41	!
	42	! 4045 2019-06-21 10:58:47Z raasch
[4045]	43	! bugfix: kind attribute added to nint function to allow for large integers which may appear in
	44	! case of default recycling width and small grid spacings
	45	!
	46	! 3999 2019-05-23 16:09:37Z suehring
[3999]	47	! Spend 3 ghost points also in case of pw-scheme when nesting is applied
	48	!
	49	! 3897 2019-04-15 11:51:14Z suehring
[3897]	50	! Minor revision of multigrid check; give warning instead of an abort.
	51	!
	52	! 3890 2019-04-12 15:59:20Z suehring
[3890]	53	! Check if grid coarsening is possible on subdomain, in order to avoid that
	54	! multigrid approach effectively reduces to a Gauss-Seidel scheme.
	55	!
	56	! 3885 2019-04-11 11:29:34Z kanani
[3885]	57	! Changes related to global restructuring of location messages and introduction
	58	! of additional debug messages
	59	!
	60	! 3884 2019-04-10 13:31:55Z Giersch
[3884]	61	! id_recycling is only calculated in case of tubulent inflow
	62	!
	63	! 3761 2019-02-25 15:31:42Z raasch
[3761]	64	! unused variable removed
	65	!
	66	! 3655 2019-01-07 16:51:22Z knoop
[3553]	67	! variables documented
[2259]	68	!
[4182]	69	! Revision 1.1 1997/07/24 11:15:09 raasch
	70	! Initial revision
	71	!
	72	!
[1]	73	! Description:
	74	! ------------
[1682]	75	!> Determination of the virtual processor topology (if not prescribed by the
	76	!> user)and computation of the grid point number and array bounds of the local
	77	!> domains.
[2696]	78	!> @todo: remove MPI-data types for 2D exchange on coarse multigrid level (not
	79	!> used any more)
[1]	80	!------------------------------------------------------------------------------!
[1682]	81	SUBROUTINE init_pegrid
	82
[1]	83
[1320]	84	USE control_parameters, &
[3182]	85	ONLY: bc_dirichlet_l, bc_dirichlet_n, bc_dirichlet_r, bc_dirichlet_s, &
	86	bc_lr, bc_ns, bc_radiation_l, bc_radiation_n, bc_radiation_r, &
[4444]	87	bc_radiation_s, &
[3241]	88	grid_level, grid_level_count, maximum_grid_level, &
[4444]	89	message_string, mg_switch_to_pe0_level, &
	90	psolver
[1]	91
[4444]	92
	93	#if defined( __parallel )
	94	USE control_parameters, &
	95	ONLY: coupling_mode, coupling_topology, gathered_size, momentum_advec, &
	96	outflow_source_plane, recycling_width, scalar_advec, subdomain_size, &
	97	turbulent_inflow, turbulent_outflow, y_shift
	98
[1320]	99	USE grid_variables, &
	100	ONLY: dx
[4444]	101	#endif
[1320]	102
	103	USE indices, &
[4444]	104	ONLY: nnx, nny, nnz, nx, nxl, nxl_mg, &
	105	nxlu, nxr, nxr_mg, ny, nyn, nyn_mg, nys, nys_mg, &
[1320]	106	nysv, nz, nzb, nzt, nzt_mg, wall_flags_1, wall_flags_2, &
	107	wall_flags_3, wall_flags_4, wall_flags_5, wall_flags_6, &
	108	wall_flags_7, wall_flags_8, wall_flags_9, wall_flags_10
[1]	109
[4444]	110	#if defined( __parallel )
	111	USE indices, &
	112	ONLY: mg_loc_ind, nbgp, nx_a, nx_o, ny_a, ny_o
	113	#endif
	114
[1320]	115	USE kinds
	116
	117	USE pegrid
[3999]	118
[4444]	119	#if defined( __parallel )
[3999]	120	USE pmc_interface, &
	121	ONLY: nested_run
[4444]	122
[1833]	123	USE spectra_mod, &
[4444]	124	ONLY: calculate_spectra
[1833]	125
[2259]	126	USE synthetic_turbulence_generator_mod, &
[2938]	127	ONLY: id_stg_left, id_stg_north, id_stg_right, id_stg_south, &
	128	use_syn_turb_gen
[4444]	129	#endif
[2259]	130
[1320]	131	USE transpose_indices, &
[4444]	132	ONLY: nxl_y, nxl_z, nxr_y, nxr_z, nyn_x, nyn_z, nys_x,&
	133	nys_z, nzb_x, nzb_y, nzt_x, nzt_y
[667]	134
[4444]	135	#if defined( __parallel )
	136	USE transpose_indices, &
	137	ONLY: nxl_yd, nxr_yd, nzb_yd, nzt_yd
	138
[2365]	139	USE vertical_nesting_mod, &
	140	ONLY: vnested, vnest_init_pegrid_domain, vnest_init_pegrid_rank
[4444]	141	#endif
[2365]	142
[1]	143	IMPLICIT NONE
	144
[3552]	145	INTEGER(iwp) :: i !< running index over number of processors or number of multigrid level
[4444]	146	#if defined( __parallel )
[3552]	147	INTEGER(iwp) :: id_inflow_l !< ID indicating processors located at the left inflow boundary
[2050]	148	INTEGER(iwp) :: id_outflow_l !< local value of id_outflow
	149	INTEGER(iwp) :: id_outflow_source_l !< local value of id_outflow_source
[3552]	150	INTEGER(iwp) :: id_recycling_l !< ID indicating processors located at the recycling plane
[2938]	151	INTEGER(iwp) :: id_stg_left_l !< left lateral boundary local core id in case of turbulence generator
	152	INTEGER(iwp) :: id_stg_north_l !< north lateral boundary local core id in case of turbulence generator
	153	INTEGER(iwp) :: id_stg_right_l !< right lateral boundary local core id in case of turbulence generator
	154	INTEGER(iwp) :: id_stg_south_l !< south lateral boundary local core id in case of turbulence generator
[3552]	155	INTEGER(iwp) :: ind(5) !< array containing the subdomain bounds
[4444]	156	#endif
[3552]	157	INTEGER(iwp) :: j !< running index, used for various loops
	158	INTEGER(iwp) :: k !< number of vertical grid points in different multigrid level
	159	INTEGER(iwp) :: maximum_grid_level_l !< maximum number of grid level without switching to PE 0
	160	INTEGER(iwp) :: mg_levels_x !< maximum number of grid level allowed along x-direction
	161	INTEGER(iwp) :: mg_levels_y !< maximum number of grid level allowed along y-direction
	162	INTEGER(iwp) :: mg_levels_z !< maximum number of grid level allowed along z-direction
	163	INTEGER(iwp) :: mg_switch_to_pe0_level_l !< maximum number of grid level with switching to PE 0
[4444]	164	#if defined( __parallel )
[3552]	165	INTEGER(iwp) :: nnx_y !< quotient of number of grid points along x-direction and number of PEs used along y-direction
	166	INTEGER(iwp) :: nny_x !< quotient of number of grid points along y-direction and number of PEs used along x-direction
	167	INTEGER(iwp) :: nny_z !< quotient of number of grid points along y-direction and number of PEs used along x-direction
	168	INTEGER(iwp) :: nnz_x !< quotient of number of grid points along z-direction and number of PEs used along x-direction
	169	INTEGER(iwp) :: nnz_y !< quotient of number of grid points along z-direction and number of PEs used along x-direction
	170	INTEGER(iwp) :: numproc_sqr !< square root of the number of processors
[4444]	171	#endif
[3552]	172	INTEGER(iwp) :: nxl_l !< lower index bound along x-direction on subdomain and different multigrid level
	173	INTEGER(iwp) :: nxr_l !< upper index bound along x-direction on subdomain and different multigrid level
	174	INTEGER(iwp) :: nyn_l !< lower index bound along y-direction on subdomain and different multigrid level
	175	INTEGER(iwp) :: nys_l !< upper index bound along y-direction on subdomain and different multigrid level
[4444]	176	#if defined( __parallel )
[3552]	177	INTEGER(iwp) :: nzb_l !< lower index bound along z-direction on subdomain and different multigrid level
[4444]	178	#endif
[3552]	179	INTEGER(iwp) :: nzt_l !< upper index bound along z-direction on subdomain and different multigrid level
	180	!$ INTEGER(iwp) :: omp_get_num_threads !< number of OpenMP threads
[1]	181
[4444]	182	#if defined( __parallel )
[3552]	183	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: ind_all !< dummy array containing index bounds on subdomain, used for gathering
	184	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nxlf !< lower index bound allong x-direction for every PE
	185	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nxrf !< upper index bound allong x-direction for every PE
	186	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nynf !< lower index bound allong y-direction for every PE
	187	INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nysf !< lower index bound allong y-direction for every PE
[1]	188
[3552]	189	INTEGER(iwp), DIMENSION(2) :: pdims_remote !< number of PEs used for coupled model (only in atmospher-ocean coupling)
[2372]	190	INTEGER(iwp) :: lcoord(2) !< PE coordinates of left neighbor along x and y
	191	INTEGER(iwp) :: rcoord(2) !< PE coordinates of right neighbor along x and y
[4444]	192	#endif
[667]	193
[1]	194	!
	195	!-- Get the number of OpenMP threads
	196	!$OMP PARALLEL
	197	!$ threads_per_task = omp_get_num_threads()
	198	!$OMP END PARALLEL
	199
	200
	201	#if defined( __parallel )
[667]	202
[3885]	203	CALL location_message( 'creating virtual PE grids + MPI derived data types', 'start' )
[1764]	204
[1]	205	!
[2177]	206	!-- Determine the processor topology or check it, if prescribed by the user
[1779]	207	IF ( npex == -1 .AND. npey == -1 ) THEN
[1]	208
	209	!
[2177]	210	!-- Automatic determination of the topology
[1779]	211	numproc_sqr = SQRT( REAL( numprocs, KIND=wp ) )
	212	pdims(1) = MAX( numproc_sqr , 1 )
[2180]	213	DO WHILE ( MOD( numprocs , pdims(1) ) /= 0 )
[1779]	214	pdims(1) = pdims(1) - 1
	215	ENDDO
[2180]	216	pdims(2) = numprocs / pdims(1)
[1]	217
[1779]	218	ELSEIF ( npex /= -1 .AND. npey /= -1 ) THEN
[1]	219
	220	!
[1779]	221	!-- Prescribed by user. Number of processors on the prescribed topology
	222	!-- must be equal to the number of PEs available to the job
	223	IF ( ( npex * npey ) /= numprocs ) THEN
[3045]	224	WRITE( message_string, * ) 'number of PEs of the prescribed ', &
[3046]	225	'topology (', npex*npey,') does not match & the number of ', &
[1779]	226	'PEs available to the job (', numprocs, ')'
	227	CALL message( 'init_pegrid', 'PA0221', 1, 2, 0, 6, 0 )
	228	ENDIF
	229	pdims(1) = npex
	230	pdims(2) = npey
[1]	231
[1779]	232	ELSE
[1]	233	!
[1779]	234	!-- If the processor topology is prescribed by the user, the number of
	235	!-- PEs must be given in both directions
[3045]	236	message_string = 'if the processor topology is prescribed by th' // &
[3046]	237	'e user & both values of "npex" and "npey" must be given' // &
[1779]	238	' in the &NAMELIST-parameter file'
	239	CALL message( 'init_pegrid', 'PA0222', 1, 2, 0, 6, 0 )
[1]	240
	241	ENDIF
	242
	243	!
	244	!-- If necessary, set horizontal boundary conditions to non-cyclic
[722]	245	IF ( bc_lr /= 'cyclic' ) cyclic(1) = .FALSE.
	246	IF ( bc_ns /= 'cyclic' ) cyclic(2) = .FALSE.
[1]	247
[807]	248
[1]	249	!
	250	!-- Create the virtual processor grid
	251	CALL MPI_CART_CREATE( comm_palm, ndim, pdims, cyclic, reorder, &
	252	comm2d, ierr )
	253	CALL MPI_COMM_RANK( comm2d, myid, ierr )
[1468]	254	WRITE (myid_char,'(''_'',I6.6)') myid
[1]	255
	256	CALL MPI_CART_COORDS( comm2d, myid, ndim, pcoord, ierr )
	257	CALL MPI_CART_SHIFT( comm2d, 0, 1, pleft, pright, ierr )
	258	CALL MPI_CART_SHIFT( comm2d, 1, 1, psouth, pnorth, ierr )
	259	!
[2372]	260	!-- In case of cyclic boundary conditions, a y-shift at the boundaries in
	261	!-- x-direction can be introduced via parameter y_shift. The shift is done
	262	!-- by modifying the processor grid in such a way that processors located
	263	!-- at the x-boundary communicate across it to processors with y-coordinate
	264	!-- shifted by y_shift relative to their own. This feature can not be used
	265	!-- in combination with an fft pressure solver. It has been implemented to
	266	!-- counter the effect of streak structures in case of cyclic boundary
	267	!-- conditions. For a description of these see Munters
	268	!-- (2016; dx.doi.org/10.1063/1.4941912)
	269	!--
	270	!-- Get coordinates of left and right neighbor on PE grid
	271	IF ( y_shift /= 0 ) THEN
[4301]	272	IF ( bc_lr == 'cyclic' ) THEN
	273	IF ( TRIM( psolver ) /= 'multigrid' .AND. &
	274	TRIM( psolver ) /= 'multigrid_noopt') &
	275	THEN
	276	message_string = 'y_shift /= 0 requires a multigrid pressure solver '
	277	CALL message( 'check_parameters', 'PA0468', 1, 2, 0, 6, 0 )
	278	ENDIF
[2372]	279
[4301]	280	CALL MPI_CART_COORDS( comm2d, pright, ndim, rcoord, ierr )
	281	CALL MPI_CART_COORDS( comm2d, pleft, ndim, lcoord, ierr )
[2372]	282
[4301]	283	!
	284	!-- If the x(y)-coordinate of the right (left) neighbor is smaller (greater)
	285	!-- than that of the calling process, then the calling process is located on
	286	!-- the right (left) boundary of the processor grid. In that case,
	287	!-- the y-coordinate of that neighbor is increased (decreased) by y_shift.
	288	!-- The rank of the process with that coordinate is then inquired and the
	289	!-- neighbor rank for MPI_SENDRECV, pright (pleft) is set to it.
	290	!-- In this way, the calling process receives a new right (left) neighbor
	291	!-- for all future MPI_SENDRECV calls. That neighbor has a y-coordinate
	292	!-- of y+(-)y_shift, where y is the original right (left) neighbor's
	293	!-- y-coordinate. The modulo-operation ensures that if the neighbor's
	294	!-- y-coordinate exceeds the grid-boundary, it will be relocated to
	295	!-- the opposite part of the grid cyclicly.
	296	IF ( rcoord(1) < pcoord(1) ) THEN
	297	rcoord(2) = MODULO( rcoord(2) + y_shift, pdims(2) )
	298	CALL MPI_CART_RANK( comm2d, rcoord, pright, ierr )
	299	ENDIF
[2372]	300
[4301]	301	IF ( lcoord(1) > pcoord(1) ) THEN
	302	lcoord(2) = MODULO( lcoord(2) - y_shift, pdims(2) )
	303	CALL MPI_CART_RANK( comm2d, lcoord, pleft, ierr )
	304	ENDIF
	305
	306	ELSE
[2372]	307	!
[4301]	308	!-- y-shift for non-cyclic boundary conditions is only implemented
	309	!-- for the turbulence recycling method in inflow_turbulence.f90
	310	IF ( .NOT. turbulent_inflow ) THEN
	311	message_string = 'y_shift /= 0 is only allowed for cyclic ' // &
	312	'boundary conditions in both directions ' // &
	313	'or with turbulent_inflow == .TRUE.'
	314	CALL message( 'check_parameters', 'PA0467', 1, 2, 0, 6, 0 )
	315	ENDIF
[2372]	316	ENDIF
	317	ENDIF
	318	!
[2365]	319	!-- Vertical nesting: store four lists that identify partner ranks to exchange
	320	!-- data
	321	IF ( vnested ) CALL vnest_init_pegrid_rank
	322
	323	!
[1]	324	!-- Determine sub-topologies for transpositions
	325	!-- Transposition from z to x:
	326	remain_dims(1) = .TRUE.
	327	remain_dims(2) = .FALSE.
	328	CALL MPI_CART_SUB( comm2d, remain_dims, comm1dx, ierr )
	329	CALL MPI_COMM_RANK( comm1dx, myidx, ierr )
	330	!
	331	!-- Transposition from x to y
	332	remain_dims(1) = .FALSE.
	333	remain_dims(2) = .TRUE.
	334	CALL MPI_CART_SUB( comm2d, remain_dims, comm1dy, ierr )
	335	CALL MPI_COMM_RANK( comm1dy, myidy, ierr )
	336
	337
	338	!
[1003]	339	!-- Calculate array bounds along x-direction for every PE.
[3045]	340	ALLOCATE( nxlf(0:pdims(1)-1), nxrf(0:pdims(1)-1), nynf(0:pdims(2)-1), &
[1003]	341	nysf(0:pdims(2)-1) )
[1]	342
[1003]	343	IF ( MOD( nx+1 , pdims(1) ) /= 0 ) THEN
[3045]	344	WRITE( message_string, * ) 'x-direction: gridpoint number (',nx+1,') ', &
[4264]	345	'is not an& integral multiple of the number',&
	346	' of processors (',pdims(1),')'
[254]	347	CALL message( 'init_pegrid', 'PA0225', 1, 2, 0, 6, 0 )
[1]	348	ELSE
[1003]	349	nnx = ( nx + 1 ) / pdims(1)
[4264]	350	ENDIF
[1]	351
	352	!
	353	!-- Left and right array bounds, number of gridpoints
	354	DO i = 0, pdims(1)-1
	355	nxlf(i) = i * nnx
	356	nxrf(i) = ( i + 1 ) * nnx - 1
	357	ENDDO
	358
	359	!
	360	!-- Calculate array bounds in y-direction for every PE.
[1003]	361	IF ( MOD( ny+1 , pdims(2) ) /= 0 ) THEN
[274]	362	WRITE( message_string, * ) 'y-direction: gridpoint number (',ny+1,') ', &
[4264]	363	'is not an& integral multiple of the number',&
	364	' of processors (',pdims(2),')'
[254]	365	CALL message( 'init_pegrid', 'PA0227', 1, 2, 0, 6, 0 )
[1]	366	ELSE
[1003]	367	nny = ( ny + 1 ) / pdims(2)
[4264]	368	ENDIF
[1]	369
	370	!
	371	!-- South and north array bounds
	372	DO j = 0, pdims(2)-1
	373	nysf(j) = j * nny
	374	nynf(j) = ( j + 1 ) * nny - 1
	375	ENDDO
	376
	377	!
	378	!-- Local array bounds of the respective PEs
[1003]	379	nxl = nxlf(pcoord(1))
	380	nxr = nxrf(pcoord(1))
	381	nys = nysf(pcoord(2))
	382	nyn = nynf(pcoord(2))
	383	nzb = 0
	384	nzt = nz
	385	nnz = nz
[1]	386
	387	!
[707]	388	!-- Set switches to define if the PE is situated at the border of the virtual
	389	!-- processor grid
	390	IF ( nxl == 0 ) left_border_pe = .TRUE.
	391	IF ( nxr == nx ) right_border_pe = .TRUE.
	392	IF ( nys == 0 ) south_border_pe = .TRUE.
	393	IF ( nyn == ny ) north_border_pe = .TRUE.
	394
	395	!
[1]	396	!-- Calculate array bounds and gridpoint numbers for the transposed arrays
	397	!-- (needed in the pressure solver)
	398	!-- For the transposed arrays, cyclic boundaries as well as top and bottom
	399	!-- boundaries are omitted, because they are obstructive to the transposition
	400
	401	!
	402	!-- 1. transposition z --> x
[1001]	403	!-- This transposition is not neccessary in case of a 1d-decomposition along x
[2938]	404	IF ( psolver == 'poisfft' .OR. calculate_spectra ) THEN
[1304]	405
[1922]	406	IF ( pdims(2) /= 1 ) THEN
	407	IF ( MOD( nz , pdims(1) ) /= 0 ) THEN
[4264]	408	WRITE( message_string, * ) 'transposition z --> x:& ', &
	409	'nz=',nz,' is not an integral multiple ',&
	410	'of pdims(1)=',pdims(1)
[1922]	411	CALL message( 'init_pegrid', 'PA0230', 1, 2, 0, 6, 0 )
	412	ENDIF
[1]	413	ENDIF
[1922]	414
	415	nys_x = nys
	416	nyn_x = nyn
	417	nny_x = nny
	418	nnz_x = nz / pdims(1)
	419	nzb_x = 1 + myidx * nnz_x
	420	nzt_x = ( myidx + 1 ) * nnz_x
	421	sendrecvcount_zx = nnx * nny * nnz_x
	422
[1]	423	ENDIF
	424
[1922]	425
[4264]	426	IF ( psolver == 'poisfft' ) THEN
[1]	427	!
[1922]	428	!-- 2. transposition x --> y
	429	IF ( MOD( nx+1 , pdims(2) ) /= 0 ) THEN
[4264]	430	WRITE( message_string, * ) 'transposition x --> y:& ', &
	431	'nx+1=',nx+1,' is not an integral ', &
	432	'multiple of pdims(2)=',pdims(2)
[1922]	433	CALL message( 'init_pegrid', 'PA0231', 1, 2, 0, 6, 0 )
	434	ENDIF
[1]	435
[1922]	436	nnz_y = nnz_x
	437	nzb_y = nzb_x
	438	nzt_y = nzt_x
	439	nnx_y = (nx+1) / pdims(2)
	440	nxl_y = myidy * nnx_y
	441	nxr_y = ( myidy + 1 ) * nnx_y - 1
	442	sendrecvcount_xy = nnx_y * nny_x * nnz_y
[1]	443	!
[4264]	444	!-- 3. transposition y --> z
[1922]	445	!-- (ELSE: x --> y in case of 1D-decomposition along x)
	446	nxl_z = nxl_y
	447	nxr_z = nxr_y
	448	nny_z = (ny+1) / pdims(1)
	449	nys_z = myidx * nny_z
	450	nyn_z = ( myidx + 1 ) * nny_z - 1
	451	sendrecvcount_yz = nnx_y * nny_z * nnz_y
[1304]	452
[1922]	453	IF ( pdims(2) /= 1 ) THEN
[1]	454	!
[1922]	455	!-- y --> z
	456	!-- This transposition is not neccessary in case of a 1d-decomposition
	457	!-- along x, except that the uptream-spline method is switched on
	458	IF ( MOD( ny+1 , pdims(1) ) /= 0 ) THEN
[4264]	459	WRITE( message_string, * ) 'transposition y --> z:& ', &
	460	'ny+1=',ny+1,' is not an integral ', &
	461	'multiple of pdims(1)=',pdims(1)
[1922]	462	CALL message( 'init_pegrid', 'PA0232', 1, 2, 0, 6, 0 )
	463	ENDIF
[1]	464
[1922]	465	ELSE
[1]	466	!
[1922]	467	!-- x --> y
	468	!-- This condition must be fulfilled for a 1D-decomposition along x
	469	IF ( MOD( ny+1 , pdims(1) ) /= 0 ) THEN
[4264]	470	WRITE( message_string, * ) 'transposition x --> y:& ', &
	471	'ny+1=',ny+1,' is not an integral ', &
	472	'multiple of pdims(1)=',pdims(1)
[1922]	473	CALL message( 'init_pegrid', 'PA0233', 1, 2, 0, 6, 0 )
	474	ENDIF
	475
[1]	476	ENDIF
	477
	478	ENDIF
	479
	480	!
	481	!-- Indices for direct transpositions z --> y (used for calculating spectra)
[1922]	482	IF ( calculate_spectra ) THEN
[1003]	483	IF ( MOD( nz, pdims(2) ) /= 0 ) THEN
[1922]	484	WRITE( message_string, * ) 'direct transposition z --> y (needed ', &
[4264]	485	'for spectra):& nz=',nz,' is not an ', &
	486	'integral multiple of pdims(2)=',pdims(2)
[254]	487	CALL message( 'init_pegrid', 'PA0234', 1, 2, 0, 6, 0 )
[1]	488	ELSE
[1003]	489	nxl_yd = nxl
	490	nxr_yd = nxr
	491	nzb_yd = 1 + myidy * ( nz / pdims(2) )
	492	nzt_yd = ( myidy + 1 ) * ( nz / pdims(2) )
	493	sendrecvcount_zyd = nnx * nny * ( nz / pdims(2) )
[1]	494	ENDIF
	495	ENDIF
	496
[1922]	497	IF ( psolver == 'poisfft' .OR. calculate_spectra ) THEN
[1]	498	!
[1922]	499	!-- Indices for direct transpositions y --> x
	500	!-- (they are only possible in case of a 1d-decomposition along x)
	501	IF ( pdims(2) == 1 ) THEN
	502	nny_x = nny / pdims(1)
	503	nys_x = myid * nny_x
	504	nyn_x = ( myid + 1 ) * nny_x - 1
	505	nzb_x = 1
	506	nzt_x = nz
	507	sendrecvcount_xy = nnx * nny_x * nz
	508	ENDIF
	509
[1]	510	ENDIF
	511
[1922]	512	IF ( psolver == 'poisfft' ) THEN
[1]	513	!
[1922]	514	!-- Indices for direct transpositions x --> y
	515	!-- (they are only possible in case of a 1d-decomposition along y)
	516	IF ( pdims(1) == 1 ) THEN
	517	nnx_y = nnx / pdims(2)
	518	nxl_y = myid * nnx_y
	519	nxr_y = ( myid + 1 ) * nnx_y - 1
	520	nzb_y = 1
	521	nzt_y = nz
	522	sendrecvcount_xy = nnx_y * nny * nz
	523	ENDIF
	524
[1]	525	ENDIF
	526
	527	!
	528	!-- Arrays for storing the array bounds are needed any more
	529	DEALLOCATE( nxlf , nxrf , nynf , nysf )
	530
[807]	531
[145]	532	!
	533	!-- Collect index bounds from other PEs (to be written to restart file later)
	534	ALLOCATE( hor_index_bounds(4,0:numprocs-1) )
	535
	536	IF ( myid == 0 ) THEN
	537
	538	hor_index_bounds(1,0) = nxl
	539	hor_index_bounds(2,0) = nxr
	540	hor_index_bounds(3,0) = nys
	541	hor_index_bounds(4,0) = nyn
	542
	543	!
	544	!-- Receive data from all other PEs
	545	DO i = 1, numprocs-1
	546	CALL MPI_RECV( ibuf, 4, MPI_INTEGER, i, MPI_ANY_TAG, comm2d, status, &
	547	ierr )
	548	hor_index_bounds(:,i) = ibuf(1:4)
	549	ENDDO
	550
	551	ELSE
	552	!
	553	!-- Send index bounds to PE0
	554	ibuf(1) = nxl
	555	ibuf(2) = nxr
	556	ibuf(3) = nys
	557	ibuf(4) = nyn
	558	CALL MPI_SEND( ibuf, 4, MPI_INTEGER, 0, myid, comm2d, ierr )
	559
	560	ENDIF
	561
[807]	562
[1]	563	#if defined( __print )
	564	!
	565	!-- Control output
	566	IF ( myid == 0 ) THEN
	567	PRINT, ' processor topology *'
	568	PRINT*, ' '
	569	PRINT*, 'myid pcoord left right south north idx idy nxl: nxr',&
	570	&' nys: nyn'
	571	PRINT*, '------------------------------------------------------------',&
	572	&'-----------'
	573	WRITE (*,1000) 0, pcoord(1), pcoord(2), pleft, pright, psouth, pnorth, &
	574	myidx, myidy, nxl, nxr, nys, nyn
	575	1000 FORMAT (I4,2X,'(',I3,',',I3,')',3X,I4,2X,I4,3X,I4,2X,I4,2X,I3,1X,I3, &
	576	2(2X,I4,':',I4))
	577
	578	!
[108]	579	!-- Receive data from the other PEs
[1]	580	DO i = 1,numprocs-1
	581	CALL MPI_RECV( ibuf, 12, MPI_INTEGER, i, MPI_ANY_TAG, comm2d, status, &
	582	ierr )
	583	WRITE (*,1000) i, ( ibuf(j) , j = 1,12 )
	584	ENDDO
	585	ELSE
	586
	587	!
	588	!-- Send data to PE0
	589	ibuf(1) = pcoord(1); ibuf(2) = pcoord(2); ibuf(3) = pleft
	590	ibuf(4) = pright; ibuf(5) = psouth; ibuf(6) = pnorth; ibuf(7) = myidx
	591	ibuf(8) = myidy; ibuf(9) = nxl; ibuf(10) = nxr; ibuf(11) = nys
	592	ibuf(12) = nyn
	593	CALL MPI_SEND( ibuf, 12, MPI_INTEGER, 0, myid, comm2d, ierr )
	594	ENDIF
	595	#endif
	596
[667]	597	!
[709]	598	!-- Determine the number of ghost point layers
[3347]	599	IF ( scalar_advec == 'ws-scheme' .OR. &
[3999]	600	momentum_advec == 'ws-scheme' .OR. nested_run ) THEN
[667]	601	nbgp = 3
	602	ELSE
	603	nbgp = 1
[709]	604	ENDIF
[667]	605
[102]	606	!
[4241]	607	!-- Check that the number of computational grid points is not smaller than the number of
	608	!-- ghost points.
	609	IF ( nnx < nbgp ) THEN
	610	WRITE( message_string, * ) 'number of subdomain grid points along x (', nnx, ') is smaller',&
	611	'than the number of ghost points (', nbgp, ')'
	612	CALL message( 'init_pegrid', 'PA0682', 1, 2, 0, 6, 0 )
	613	ENDIF
	614	IF ( nny < nbgp ) THEN
	615	WRITE( message_string, * ) 'number of subdomain grid points along y (', nny, ') is smaller',&
	616	'than the number of ghost points (', nbgp, ')'
	617	CALL message( 'init_pegrid', 'PA0683', 1, 2, 0, 6, 0 )
	618	ENDIF
	619
	620	!
[709]	621	!-- Create a new MPI derived datatype for the exchange of surface (xy) data,
	622	!-- which is needed for coupled atmosphere-ocean runs.
	623	!-- First, calculate number of grid points of an xy-plane.
[667]	624	ngp_xy = ( nxr - nxl + 1 + 2 * nbgp ) * ( nyn - nys + 1 + 2 * nbgp )
[102]	625	CALL MPI_TYPE_VECTOR( ngp_xy, 1, nzt-nzb+2, MPI_REAL, type_xy, ierr )
	626	CALL MPI_TYPE_COMMIT( type_xy, ierr )
[667]	627
[2365]	628	IF ( TRIM( coupling_mode ) /= 'uncoupled' .AND. .NOT. vnested ) THEN
[667]	629
	630	!
	631	!-- Pass the number of grid points of the atmosphere model to
	632	!-- the ocean model and vice versa
	633	IF ( coupling_mode == 'atmosphere_to_ocean' ) THEN
	634
	635	nx_a = nx
	636	ny_a = ny
	637
[709]	638	IF ( myid == 0 ) THEN
	639
	640	CALL MPI_SEND( nx_a, 1, MPI_INTEGER, numprocs, 1, comm_inter, &
	641	ierr )
	642	CALL MPI_SEND( ny_a, 1, MPI_INTEGER, numprocs, 2, comm_inter, &
	643	ierr )
	644	CALL MPI_SEND( pdims, 2, MPI_INTEGER, numprocs, 3, comm_inter, &
	645	ierr )
	646	CALL MPI_RECV( nx_o, 1, MPI_INTEGER, numprocs, 4, comm_inter, &
	647	status, ierr )
	648	CALL MPI_RECV( ny_o, 1, MPI_INTEGER, numprocs, 5, comm_inter, &
	649	status, ierr )
	650	CALL MPI_RECV( pdims_remote, 2, MPI_INTEGER, numprocs, 6, &
[667]	651	comm_inter, status, ierr )
	652	ENDIF
	653
[709]	654	CALL MPI_BCAST( nx_o, 1, MPI_INTEGER, 0, comm2d, ierr )
	655	CALL MPI_BCAST( ny_o, 1, MPI_INTEGER, 0, comm2d, ierr )
	656	CALL MPI_BCAST( pdims_remote, 2, MPI_INTEGER, 0, comm2d, ierr )
[667]	657
	658	ELSEIF ( coupling_mode == 'ocean_to_atmosphere' ) THEN
	659
	660	nx_o = nx
	661	ny_o = ny
	662
	663	IF ( myid == 0 ) THEN
[709]	664
	665	CALL MPI_RECV( nx_a, 1, MPI_INTEGER, 0, 1, comm_inter, status, &
	666	ierr )
	667	CALL MPI_RECV( ny_a, 1, MPI_INTEGER, 0, 2, comm_inter, status, &
	668	ierr )
	669	CALL MPI_RECV( pdims_remote, 2, MPI_INTEGER, 0, 3, comm_inter, &
	670	status, ierr )
	671	CALL MPI_SEND( nx_o, 1, MPI_INTEGER, 0, 4, comm_inter, ierr )
	672	CALL MPI_SEND( ny_o, 1, MPI_INTEGER, 0, 5, comm_inter, ierr )
	673	CALL MPI_SEND( pdims, 2, MPI_INTEGER, 0, 6, comm_inter, ierr )
[667]	674	ENDIF
	675
	676	CALL MPI_BCAST( nx_a, 1, MPI_INTEGER, 0, comm2d, ierr)
	677	CALL MPI_BCAST( ny_a, 1, MPI_INTEGER, 0, comm2d, ierr)
	678	CALL MPI_BCAST( pdims_remote, 2, MPI_INTEGER, 0, comm2d, ierr)
	679
	680	ENDIF
	681
[709]	682	ngp_a = ( nx_a+1 + 2 * nbgp ) * ( ny_a+1 + 2 * nbgp )
	683	ngp_o = ( nx_o+1 + 2 * nbgp ) * ( ny_o+1 + 2 * nbgp )
[667]	684
	685	!
[709]	686	!-- Determine if the horizontal grid and the number of PEs in ocean and
	687	!-- atmosphere is same or not
	688	IF ( nx_o == nx_a .AND. ny_o == ny_a .AND. &
[667]	689	pdims(1) == pdims_remote(1) .AND. pdims(2) == pdims_remote(2) ) &
	690	THEN
	691	coupling_topology = 0
	692	ELSE
	693	coupling_topology = 1
	694	ENDIF
	695
	696	!
	697	!-- Determine the target PEs for the exchange between ocean and
	698	!-- atmosphere (comm2d)
[709]	699	IF ( coupling_topology == 0 ) THEN
	700	!
	701	!-- In case of identical topologies, every atmosphere PE has exactly one
	702	!-- ocean PE counterpart and vice versa
	703	IF ( TRIM( coupling_mode ) == 'atmosphere_to_ocean' ) THEN
[667]	704	target_id = myid + numprocs
	705	ELSE
	706	target_id = myid
	707	ENDIF
	708
	709	ELSE
	710	!
	711	!-- In case of nonequivalent topology in ocean and atmosphere only for
	712	!-- PE0 in ocean and PE0 in atmosphere a target_id is needed, since
[709]	713	!-- data echxchange between ocean and atmosphere will be done only
	714	!-- between these PEs.
	715	IF ( myid == 0 ) THEN
	716
	717	IF ( TRIM( coupling_mode ) == 'atmosphere_to_ocean' ) THEN
[667]	718	target_id = numprocs
	719	ELSE
	720	target_id = 0
	721	ENDIF
[709]	722
[667]	723	ENDIF
[709]	724
[667]	725	ENDIF
	726
	727	ENDIF
	728
[2365]	729	!
	730	!-- Store partner grid point co-ordinates as lists.
	731	!-- Create custom MPI vector datatypes for contiguous data transfer
	732	IF ( vnested ) CALL vnest_init_pegrid_domain
[667]	733
[1]	734	#else
	735
	736	!
	737	!-- Array bounds when running on a single PE (respectively a non-parallel
	738	!-- machine)
[1003]	739	nxl = 0
	740	nxr = nx
	741	nnx = nxr - nxl + 1
	742	nys = 0
	743	nyn = ny
	744	nny = nyn - nys + 1
	745	nzb = 0
	746	nzt = nz
	747	nnz = nz
[1]	748
[145]	749	ALLOCATE( hor_index_bounds(4,0:0) )
	750	hor_index_bounds(1,0) = nxl
	751	hor_index_bounds(2,0) = nxr
	752	hor_index_bounds(3,0) = nys
	753	hor_index_bounds(4,0) = nyn
	754
[1]	755	!
	756	!-- Array bounds for the pressure solver (in the parallel code, these bounds
	757	!-- are the ones for the transposed arrays)
[1003]	758	nys_x = nys
	759	nyn_x = nyn
	760	nzb_x = nzb + 1
	761	nzt_x = nzt
[1]	762
[1003]	763	nxl_y = nxl
	764	nxr_y = nxr
	765	nzb_y = nzb + 1
	766	nzt_y = nzt
[1]	767
[1003]	768	nxl_z = nxl
	769	nxr_z = nxr
	770	nys_z = nys
	771	nyn_z = nyn
[1]	772
	773	#endif
	774
	775	!
	776	!-- Calculate number of grid levels necessary for the multigrid poisson solver
	777	!-- as well as the gridpoint indices on each level
[1575]	778	IF ( psolver(1:9) == 'multigrid' ) THEN
[1]	779
	780	!
	781	!-- First calculate number of possible grid levels for the subdomains
	782	mg_levels_x = 1
	783	mg_levels_y = 1
	784	mg_levels_z = 1
	785
	786	i = nnx
	787	DO WHILE ( MOD( i, 2 ) == 0 .AND. i /= 2 )
	788	i = i / 2
	789	mg_levels_x = mg_levels_x + 1
	790	ENDDO
	791
	792	j = nny
	793	DO WHILE ( MOD( j, 2 ) == 0 .AND. j /= 2 )
	794	j = j / 2
	795	mg_levels_y = mg_levels_y + 1
	796	ENDDO
	797
[181]	798	k = nz ! do not use nnz because it might be > nz due to transposition
	799	! requirements
[1]	800	DO WHILE ( MOD( k, 2 ) == 0 .AND. k /= 2 )
	801	k = k / 2
	802	mg_levels_z = mg_levels_z + 1
	803	ENDDO
[2197]	804	!
	805	!-- The optimized MG-solver does not allow odd values for nz at the coarsest
	806	!-- grid level
	807	IF ( TRIM( psolver ) /= 'multigrid_noopt' ) THEN
	808	IF ( MOD( k, 2 ) /= 0 ) mg_levels_z = mg_levels_z - 1
[3057]	809	!
	810	!-- An odd value of nz does not work. The finest level must have an even
	811	!-- value.
	812	IF ( mg_levels_z == 0 ) THEN
	813	message_string = 'optimized multigrid method requires nz to be even'
[3058]	814	CALL message( 'init_pegrid', 'PA0495', 1, 2, 0, 6, 0 )
[3057]	815	ENDIF
[2197]	816	ENDIF
[1]	817
	818	maximum_grid_level = MIN( mg_levels_x, mg_levels_y, mg_levels_z )
[3890]	819	!
	820	!-- Check if subdomain sizes prevents any coarsening.
	821	!-- This case, the maximum number of grid levels is 1, i.e. effectively
	822	!-- a Gauss-Seidel scheme is applied rather than a multigrid approach.
[3897]	823	!-- Give a warning in this case.
	824	IF ( maximum_grid_level == 1 .AND. mg_switch_to_pe0_level == -1 ) THEN
	825	message_string = 'No grid coarsening possible, multigrid ' // &
	826	'approach effectively reduces to a Gauss-Seidel ' //&
	827	'scheme.'
	828
	829	CALL message( 'poismg', 'PA0648', 0, 1, 0, 6, 0 )
[3890]	830	ENDIF
[1]	831
	832	!
	833	!-- Find out, if the total domain allows more levels. These additional
[709]	834	!-- levels are identically processed on all PEs.
[197]	835	IF ( numprocs > 1 .AND. mg_switch_to_pe0_level /= -1 ) THEN
[709]	836
[1]	837	IF ( mg_levels_z > MIN( mg_levels_x, mg_levels_y ) ) THEN
[709]	838
[1]	839	mg_switch_to_pe0_level_l = maximum_grid_level
	840
	841	mg_levels_x = 1
	842	mg_levels_y = 1
	843
	844	i = nx+1
	845	DO WHILE ( MOD( i, 2 ) == 0 .AND. i /= 2 )
	846	i = i / 2
	847	mg_levels_x = mg_levels_x + 1
	848	ENDDO
	849
	850	j = ny+1
	851	DO WHILE ( MOD( j, 2 ) == 0 .AND. j /= 2 )
	852	j = j / 2
	853	mg_levels_y = mg_levels_y + 1
	854	ENDDO
	855
	856	maximum_grid_level_l = MIN( mg_levels_x, mg_levels_y, mg_levels_z )
	857
	858	IF ( maximum_grid_level_l > mg_switch_to_pe0_level_l ) THEN
	859	mg_switch_to_pe0_level_l = maximum_grid_level_l - &
	860	mg_switch_to_pe0_level_l + 1
	861	ELSE
	862	mg_switch_to_pe0_level_l = 0
	863	ENDIF
[709]	864
[1]	865	ELSE
[3057]	866
[1]	867	mg_switch_to_pe0_level_l = 0
	868	maximum_grid_level_l = maximum_grid_level
[709]	869
[1]	870	ENDIF
	871
	872	!
	873	!-- Use switch level calculated above only if it is not pre-defined
	874	!-- by user
	875	IF ( mg_switch_to_pe0_level == 0 ) THEN
	876	IF ( mg_switch_to_pe0_level_l /= 0 ) THEN
	877	mg_switch_to_pe0_level = mg_switch_to_pe0_level_l
	878	maximum_grid_level = maximum_grid_level_l
	879	ENDIF
	880
	881	ELSE
	882	!
	883	!-- Check pre-defined value and reset to default, if neccessary
[3045]	884	IF ( mg_switch_to_pe0_level < mg_switch_to_pe0_level_l .OR. &
[1]	885	mg_switch_to_pe0_level >= maximum_grid_level_l ) THEN
[3045]	886	message_string = 'mg_switch_to_pe0_level ' // &
[2271]	887	'out of range and reset to 0'
[254]	888	CALL message( 'init_pegrid', 'PA0235', 0, 1, 0, 6, 0 )
[1]	889	mg_switch_to_pe0_level = 0
	890	ELSE
	891	!
	892	!-- Use the largest number of possible levels anyway and recalculate
	893	!-- the switch level to this largest number of possible values
	894	maximum_grid_level = maximum_grid_level_l
	895
	896	ENDIF
[709]	897
[1]	898	ENDIF
	899
	900	ENDIF
	901
[1056]	902	ALLOCATE( grid_level_count(maximum_grid_level), &
	903	nxl_mg(0:maximum_grid_level), nxr_mg(0:maximum_grid_level), &
	904	nyn_mg(0:maximum_grid_level), nys_mg(0:maximum_grid_level), &
	905	nzt_mg(0:maximum_grid_level) )
[1]	906
	907	grid_level_count = 0
[1056]	908	!
	909	!-- Index zero required as dummy due to definition of arrays f2 and p2 in
	910	!-- recursive subroutine next_mg_level
	911	nxl_mg(0) = 0; nxr_mg(0) = 0; nyn_mg(0) = 0; nys_mg(0) = 0; nzt_mg(0) = 0
[778]	912
[1]	913	nxl_l = nxl; nxr_l = nxr; nys_l = nys; nyn_l = nyn; nzt_l = nzt
	914
	915	DO i = maximum_grid_level, 1 , -1
	916
	917	IF ( i == mg_switch_to_pe0_level ) THEN
[1804]	918	#if defined( __parallel )
[1]	919	!
	920	!-- Save the grid size of the subdomain at the switch level, because
	921	!-- it is needed in poismg.
	922	ind(1) = nxl_l; ind(2) = nxr_l
	923	ind(3) = nys_l; ind(4) = nyn_l
	924	ind(5) = nzt_l
	925	ALLOCATE( ind_all(5*numprocs), mg_loc_ind(5,0:numprocs-1) )
	926	CALL MPI_ALLGATHER( ind, 5, MPI_INTEGER, ind_all, 5, &
	927	MPI_INTEGER, comm2d, ierr )
	928	DO j = 0, numprocs-1
	929	DO k = 1, 5
	930	mg_loc_ind(k,j) = ind_all(k+j*5)
	931	ENDDO
	932	ENDDO
	933	DEALLOCATE( ind_all )
	934	!
[709]	935	!-- Calculate the grid size of the total domain
[1]	936	nxr_l = ( nxr_l-nxl_l+1 ) * pdims(1) - 1
	937	nxl_l = 0
	938	nyn_l = ( nyn_l-nys_l+1 ) * pdims(2) - 1
	939	nys_l = 0
	940	!
	941	!-- The size of this gathered array must not be larger than the
	942	!-- array tend, which is used in the multigrid scheme as a temporary
[778]	943	!-- array. Therefore the subdomain size of an PE is calculated and
	944	!-- the size of the gathered grid. These values are used in
	945	!-- routines pres and poismg
	946	subdomain_size = ( nxr - nxl + 2 * nbgp + 1 ) * &
	947	( nyn - nys + 2 * nbgp + 1 ) * ( nzt - nzb + 2 )
[3045]	948	gathered_size = ( nxr_l - nxl_l + 3 ) * ( nyn_l - nys_l + 3 ) * &
[1]	949	( nzt_l - nzb + 2 )
	950
[1804]	951	#else
[3045]	952	message_string = 'multigrid gather/scatter impossible ' // &
[1]	953	'in non parallel mode'
[254]	954	CALL message( 'init_pegrid', 'PA0237', 1, 2, 0, 6, 0 )
[1]	955	#endif
	956	ENDIF
	957
	958	nxl_mg(i) = nxl_l
	959	nxr_mg(i) = nxr_l
	960	nys_mg(i) = nys_l
	961	nyn_mg(i) = nyn_l
	962	nzt_mg(i) = nzt_l
	963
	964	nxl_l = nxl_l / 2
	965	nxr_l = nxr_l / 2
	966	nys_l = nys_l / 2
	967	nyn_l = nyn_l / 2
	968	nzt_l = nzt_l / 2
[3890]	969
[1]	970	ENDDO
	971
[780]	972	!
[3045]	973	!-- Temporary problem: Currently calculation of maxerror in routine poismg crashes
[780]	974	!-- if grid data are collected on PE0 already on the finest grid level.
	975	!-- To be solved later.
	976	IF ( maximum_grid_level == mg_switch_to_pe0_level ) THEN
	977	message_string = 'grid coarsening on subdomain level cannot be performed'
	978	CALL message( 'poismg', 'PA0236', 1, 2, 0, 6, 0 )
	979	ENDIF
	980
[1]	981	ELSE
	982
[667]	983	maximum_grid_level = 0
[1]	984
	985	ENDIF
	986
[722]	987	!
	988	!-- Default level 0 tells exchange_horiz that all ghost planes have to be
	989	!-- exchanged. grid_level is adjusted in poismg, where only one ghost plane
	990	!-- is required.
	991	grid_level = 0
[1]	992
[1804]	993	#if defined( __parallel )
[1]	994	!
	995	!-- Gridpoint number for the exchange of ghost points (y-line for 2D-arrays)
[667]	996	ngp_y = nyn - nys + 1 + 2 * nbgp
[1]	997
	998	!
[709]	999	!-- Define new MPI derived datatypes for the exchange of ghost points in
	1000	!-- x- and y-direction for 2D-arrays (line)
[1968]	1001	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2*nbgp, nbgp, ngp_y, MPI_REAL, type_x, &
[709]	1002	ierr )
[1]	1003	CALL MPI_TYPE_COMMIT( type_x, ierr )
	1004
[667]	1005	CALL MPI_TYPE_VECTOR( nbgp, ngp_y, ngp_y, MPI_REAL, type_y, ierr )
	1006	CALL MPI_TYPE_COMMIT( type_y, ierr )
[1968]	1007	!
	1008	!-- Define new MPI derived datatypes for the exchange of ghost points in
[3542]	1009	!-- x- and y-direction for 2D-INTEGER arrays (line) - on normal grid.
	1010	!-- Define types for 32-bit and 8-bit Integer. The 8-bit Integer are only
	1011	!-- required on normal grid, while 32-bit Integer may be also required on
	1012	!-- coarser grid level in case of multigrid solver.
	1013	!
	1014	!-- 8-bit Integer
	1015	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2*nbgp, nbgp, ngp_y, MPI_BYTE, &
	1016	type_x_byte, ierr )
	1017	CALL MPI_TYPE_COMMIT( type_x_byte, ierr )
	1018
	1019	CALL MPI_TYPE_VECTOR( nbgp, ngp_y, ngp_y, MPI_BYTE, &
	1020	type_y_byte, ierr )
	1021	CALL MPI_TYPE_COMMIT( type_y_byte, ierr )
	1022	!
	1023	!-- 32-bit Integer
[1968]	1024	ALLOCATE( type_x_int(0:maximum_grid_level), &
	1025	type_y_int(0:maximum_grid_level) )
[3542]	1026
[1968]	1027	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2*nbgp, nbgp, ngp_y, MPI_INTEGER, &
	1028	type_x_int(0), ierr )
	1029	CALL MPI_TYPE_COMMIT( type_x_int(0), ierr )
[667]	1030
[1968]	1031	CALL MPI_TYPE_VECTOR( nbgp, ngp_y, ngp_y, MPI_INTEGER, type_y_int(0), ierr )
	1032	CALL MPI_TYPE_COMMIT( type_y_int(0), ierr )
[1]	1033	!
	1034	!-- Calculate gridpoint numbers for the exchange of ghost points along x
	1035	!-- (yz-plane for 3D-arrays) and define MPI derived data type(s) for the
	1036	!-- exchange of ghost points in y-direction (xz-plane).
	1037	!-- Do these calculations for the model grid and (if necessary) also
	1038	!-- for the coarser grid levels used in the multigrid method
[2696]	1039	ALLOCATE ( ngp_xz(0:maximum_grid_level), &
	1040	ngp_xz_int(0:maximum_grid_level), &
	1041	ngp_yz(0:maximum_grid_level), &
	1042	ngp_yz_int(0:maximum_grid_level), &
	1043	type_xz(0:maximum_grid_level), &
	1044	type_xz_int(0:maximum_grid_level), &
	1045	type_yz(0:maximum_grid_level), &
	1046	type_yz_int(0:maximum_grid_level) )
[1]	1047
	1048	nxl_l = nxl; nxr_l = nxr; nys_l = nys; nyn_l = nyn; nzb_l = nzb; nzt_l = nzt
[709]	1049
[667]	1050	!
	1051	!-- Discern between the model grid, which needs nbgp ghost points and
	1052	!-- grid levels for the multigrid scheme. In the latter case only one
	1053	!-- ghost point is necessary.
[709]	1054	!-- First definition of MPI-datatypes for exchange of ghost layers on normal
[667]	1055	!-- grid. The following loop is needed for data exchange in poismg.f90.
	1056	!
	1057	!-- Determine number of grid points of yz-layer for exchange
	1058	ngp_yz(0) = (nzt - nzb + 2) * (nyn - nys + 1 + 2 * nbgp)
[709]	1059
[667]	1060	!
[709]	1061	!-- Define an MPI-datatype for the exchange of left/right boundaries.
	1062	!-- Although data are contiguous in physical memory (which does not
	1063	!-- necessarily require an MPI-derived datatype), the data exchange between
	1064	!-- left and right PE's using the MPI-derived type is 10% faster than without.
[667]	1065	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2nbgp, nbgp(nzt-nzb+2), ngp_yz(0), &
[709]	1066	MPI_REAL, type_xz(0), ierr )
[667]	1067	CALL MPI_TYPE_COMMIT( type_xz(0), ierr )
[1]	1068
[709]	1069	CALL MPI_TYPE_VECTOR( nbgp, ngp_yz(0), ngp_yz(0), MPI_REAL, type_yz(0), &
	1070	ierr )
[667]	1071	CALL MPI_TYPE_COMMIT( type_yz(0), ierr )
[709]	1072
[667]	1073	!
[2696]	1074	!-- Define data types for exchange of 3D Integer arrays.
	1075	ngp_yz_int(0) = (nzt - nzb + 2) * (nyn - nys + 1 + 2 * nbgp)
	1076
	1077	CALL MPI_TYPE_VECTOR( nxr-nxl+1+2nbgp, nbgp(nzt-nzb+2), ngp_yz_int(0), &
	1078	MPI_INTEGER, type_xz_int(0), ierr )
	1079	CALL MPI_TYPE_COMMIT( type_xz_int(0), ierr )
	1080
	1081	CALL MPI_TYPE_VECTOR( nbgp, ngp_yz_int(0), ngp_yz_int(0), MPI_INTEGER, &
	1082	type_yz_int(0), ierr )
	1083	CALL MPI_TYPE_COMMIT( type_yz_int(0), ierr )
	1084
	1085	!
[709]	1086	!-- Definition of MPI-datatypes for multigrid method (coarser level grids)
[1575]	1087	IF ( psolver(1:9) == 'multigrid' ) THEN
[667]	1088	!
[709]	1089	!-- Definition of MPI-datatyoe as above, but only 1 ghost level is used
	1090	DO i = maximum_grid_level, 1 , -1
[1968]	1091	!
[2696]	1092	!-- For 3D-exchange on different multigrid level, one ghost point for
	1093	!-- REAL arrays, two ghost points for INTEGER arrays
[1575]	1094	ngp_xz(i) = (nzt_l - nzb_l + 2) * (nxr_l - nxl_l + 3)
[667]	1095	ngp_yz(i) = (nzt_l - nzb_l + 2) * (nyn_l - nys_l + 3)
	1096
[2696]	1097	ngp_xz_int(i) = (nzt_l - nzb_l + 2) * (nxr_l - nxl_l + 3)
	1098	ngp_yz_int(i) = (nzt_l - nzb_l + 2) * (nyn_l - nys_l + 3)
	1099	!
	1100	!-- MPI data type for REAL arrays, for xz-layers
	1101	CALL MPI_TYPE_VECTOR( nxr_l-nxl_l+3, nzt_l-nzb_l+2, ngp_yz(i), &
[709]	1102	MPI_REAL, type_xz(i), ierr )
[667]	1103	CALL MPI_TYPE_COMMIT( type_xz(i), ierr )
[1]	1104
[2696]	1105	!
	1106	!-- MPI data type for INTEGER arrays, for xz-layers
	1107	CALL MPI_TYPE_VECTOR( nxr_l-nxl_l+3, nzt_l-nzb_l+2, ngp_yz_int(i), &
	1108	MPI_INTEGER, type_xz_int(i), ierr )
	1109	CALL MPI_TYPE_COMMIT( type_xz_int(i), ierr )
	1110
	1111	!
	1112	!-- MPI data type for REAL arrays, for yz-layers
[709]	1113	CALL MPI_TYPE_VECTOR( 1, ngp_yz(i), ngp_yz(i), MPI_REAL, type_yz(i), &
	1114	ierr )
[667]	1115	CALL MPI_TYPE_COMMIT( type_yz(i), ierr )
[2696]	1116	!
	1117	!-- MPI data type for INTEGER arrays, for yz-layers
	1118	CALL MPI_TYPE_VECTOR( 1, ngp_yz_int(i), ngp_yz_int(i), MPI_INTEGER, &
	1119	type_yz_int(i), ierr )
	1120	CALL MPI_TYPE_COMMIT( type_yz_int(i), ierr )
[667]	1121
[1968]	1122
	1123	!-- For 2D-exchange of INTEGER arrays on coarser grid level, where 2 ghost
[3542]	1124	!-- points need to be exchanged. Only required for 32-bit Integer arrays.
	1125	CALL MPI_TYPE_VECTOR( nxr_l-nxl_l+5, 2, nyn_l-nys_l+5, MPI_INTEGER, &
[1968]	1126	type_x_int(i), ierr )
	1127	CALL MPI_TYPE_COMMIT( type_x_int(i), ierr )
	1128
	1129
[3542]	1130	CALL MPI_TYPE_VECTOR( 2, nyn_l-nys_l+5, nyn_l-nys_l+5, MPI_INTEGER, &
[1968]	1131	type_y_int(i), ierr )
	1132	CALL MPI_TYPE_COMMIT( type_y_int(i), ierr )
	1133
[667]	1134	nxl_l = nxl_l / 2
	1135	nxr_l = nxr_l / 2
	1136	nys_l = nys_l / 2
	1137	nyn_l = nyn_l / 2
	1138	nzt_l = nzt_l / 2
[709]	1139
[667]	1140	ENDDO
[709]	1141
	1142	ENDIF
[1677]	1143
[1]	1144	#endif
	1145
[1804]	1146	#if defined( __parallel )
[1]	1147	!
[1933]	1148	!-- Setting of flags for inflow/outflow/nesting conditions.
[1964]	1149	IF ( pleft == MPI_PROC_NULL ) THEN
[3182]	1150	IF ( bc_lr == 'dirichlet/radiation' .OR. bc_lr == 'nested' .OR. &
	1151	bc_lr == 'nesting_offline' ) THEN
	1152	bc_dirichlet_l = .TRUE.
[1964]	1153	ELSEIF ( bc_lr == 'radiation/dirichlet' ) THEN
[3182]	1154	bc_radiation_l = .TRUE.
[1]	1155	ENDIF
	1156	ENDIF
[1933]	1157
[1964]	1158	IF ( pright == MPI_PROC_NULL ) THEN
	1159	IF ( bc_lr == 'dirichlet/radiation' ) THEN
[3182]	1160	bc_radiation_r = .TRUE.
	1161	ELSEIF ( bc_lr == 'radiation/dirichlet' .OR. bc_lr == 'nested' .OR. &
	1162	bc_lr == 'nesting_offline' ) THEN
	1163	bc_dirichlet_r = .TRUE.
[1]	1164	ENDIF
	1165	ENDIF
	1166
[1964]	1167	IF ( psouth == MPI_PROC_NULL ) THEN
	1168	IF ( bc_ns == 'dirichlet/radiation' ) THEN
[3182]	1169	bc_radiation_s = .TRUE.
	1170	ELSEIF ( bc_ns == 'radiation/dirichlet' .OR. bc_ns == 'nested' .OR. &
	1171	bc_ns == 'nesting_offline' ) THEN
	1172	bc_dirichlet_s = .TRUE.
[1]	1173	ENDIF
	1174	ENDIF
	1175
[1964]	1176	IF ( pnorth == MPI_PROC_NULL ) THEN
[3182]	1177	IF ( bc_ns == 'dirichlet/radiation' .OR. bc_ns == 'nested' .OR. &
	1178	bc_ns == 'nesting_offline' ) THEN
	1179	bc_dirichlet_n = .TRUE.
[1964]	1180	ELSEIF ( bc_ns == 'radiation/dirichlet' ) THEN
[3182]	1181	bc_radiation_n = .TRUE.
[1]	1182	ENDIF
	1183	ENDIF
[2938]	1184	!
	1185	!-- In case of synthetic turbulence geneartor determine ids.
	1186	!-- Please note, if no forcing or nesting is applied, the generator is applied
	1187	!-- only at the left lateral boundary.
	1188	IF ( use_syn_turb_gen ) THEN
[3182]	1189	IF ( bc_dirichlet_l ) THEN
[2938]	1190	id_stg_left_l = myidx
	1191	ELSE
	1192	id_stg_left_l = 0
	1193	ENDIF
[3182]	1194	IF ( bc_dirichlet_r ) THEN
[2938]	1195	id_stg_right_l = myidx
	1196	ELSE
	1197	id_stg_right_l = 0
	1198	ENDIF
[3182]	1199	IF ( bc_dirichlet_s ) THEN
[2938]	1200	id_stg_south_l = myidy
	1201	ELSE
	1202	id_stg_south_l = 0
	1203	ENDIF
[3182]	1204	IF ( bc_dirichlet_n ) THEN
[2938]	1205	id_stg_north_l = myidy
	1206	ELSE
	1207	id_stg_north_l = 0
	1208	ENDIF
[1968]	1209
[2938]	1210	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1211	CALL MPI_ALLREDUCE( id_stg_left_l, id_stg_left, 1, MPI_INTEGER, &
	1212	MPI_SUM, comm1dx, ierr )
	1213
	1214	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1215	CALL MPI_ALLREDUCE( id_stg_right_l, id_stg_right, 1, MPI_INTEGER, &
	1216	MPI_SUM, comm1dx, ierr )
	1217
	1218	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1219	CALL MPI_ALLREDUCE( id_stg_south_l, id_stg_south, 1, MPI_INTEGER, &
	1220	MPI_SUM, comm1dy, ierr )
	1221
	1222	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1223	CALL MPI_ALLREDUCE( id_stg_north_l, id_stg_north, 1, MPI_INTEGER, &
	1224	MPI_SUM, comm1dy, ierr )
	1225
	1226	ENDIF
	1227
[151]	1228	!
	1229	!-- Broadcast the id of the inflow PE
[3182]	1230	IF ( bc_dirichlet_l ) THEN
[163]	1231	id_inflow_l = myidx
[151]	1232	ELSE
	1233	id_inflow_l = 0
	1234	ENDIF
[622]	1235	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
[151]	1236	CALL MPI_ALLREDUCE( id_inflow_l, id_inflow, 1, MPI_INTEGER, MPI_SUM, &
	1237	comm1dx, ierr )
	1238
[163]	1239	!
	1240	!-- Broadcast the id of the recycling plane
	1241	!-- WARNING: needs to be adjusted in case of inflows other than from left side!
[3884]	1242	IF ( turbulent_inflow ) THEN
	1243
[4045]	1244	IF ( NINT( recycling_width / dx, KIND=idp ) >= nxl .AND. &
	1245	NINT( recycling_width / dx, KIND=idp ) <= nxr ) THEN
[3884]	1246	id_recycling_l = myidx
	1247	ELSE
	1248	id_recycling_l = 0
	1249	ENDIF
	1250	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1251	CALL MPI_ALLREDUCE( id_recycling_l, id_recycling, 1, MPI_INTEGER, MPI_SUM, &
	1252	comm1dx, ierr )
	1253
[163]	1254	ENDIF
	1255
[2050]	1256	!
	1257	!-- Broadcast the id of the outflow PE and outflow-source plane
	1258	IF ( turbulent_outflow ) THEN
	1259
[3182]	1260	IF ( bc_radiation_r ) THEN
[2050]	1261	id_outflow_l = myidx
	1262	ELSE
	1263	id_outflow_l = 0
	1264	ENDIF
	1265	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1266	CALL MPI_ALLREDUCE( id_outflow_l, id_outflow, 1, MPI_INTEGER, MPI_SUM, &
	1267	comm1dx, ierr )
	1268
	1269	IF ( NINT( outflow_source_plane / dx ) >= nxl .AND. &
	1270	NINT( outflow_source_plane / dx ) <= nxr ) THEN
	1271	id_outflow_source_l = myidx
	1272	ELSE
	1273	id_outflow_source_l = 0
	1274	ENDIF
	1275	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	1276	CALL MPI_ALLREDUCE( id_outflow_source_l, id_outflow_source, 1, &
	1277	MPI_INTEGER, MPI_SUM, comm1dx, ierr )
	1278
	1279	ENDIF
	1280
[3885]	1281	CALL location_message( 'creating virtual PE grids + MPI derived data types', 'finished' )
[1384]	1282
[1804]	1283	#else
[1159]	1284	IF ( bc_lr == 'dirichlet/radiation' ) THEN
[3182]	1285	bc_dirichlet_l = .TRUE.
	1286	bc_radiation_r = .TRUE.
[1159]	1287	ELSEIF ( bc_lr == 'radiation/dirichlet' ) THEN
[3182]	1288	bc_radiation_l = .TRUE.
	1289	bc_dirichlet_r = .TRUE.
[1]	1290	ENDIF
	1291
[1159]	1292	IF ( bc_ns == 'dirichlet/radiation' ) THEN
[3182]	1293	bc_dirichlet_n = .TRUE.
	1294	bc_radiation_s = .TRUE.
[1159]	1295	ELSEIF ( bc_ns == 'radiation/dirichlet' ) THEN
[3182]	1296	bc_radiation_n = .TRUE.
	1297	bc_dirichlet_s = .TRUE.
[1]	1298	ENDIF
	1299	#endif
[807]	1300
[106]	1301	!
[978]	1302	!-- At the inflow or outflow, u or v, respectively, have to be calculated for
	1303	!-- one more grid point.
[3182]	1304	IF ( bc_dirichlet_l .OR. bc_radiation_l ) THEN
[106]	1305	nxlu = nxl + 1
	1306	ELSE
	1307	nxlu = nxl
	1308	ENDIF
[3182]	1309	IF ( bc_dirichlet_s .OR. bc_radiation_s ) THEN
[106]	1310	nysv = nys + 1
	1311	ELSE
	1312	nysv = nys
	1313	ENDIF
[1]	1314
[114]	1315	!
	1316	!-- Allocate wall flag arrays used in the multigrid solver
[1575]	1317	IF ( psolver(1:9) == 'multigrid' ) THEN
[114]	1318
	1319	DO i = maximum_grid_level, 1, -1
	1320
	1321	SELECT CASE ( i )
	1322
	1323	CASE ( 1 )
	1324	ALLOCATE( wall_flags_1(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 ( 2 )
	1329	ALLOCATE( wall_flags_2(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 ( 3 )
	1334	ALLOCATE( wall_flags_3(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 ( 4 )
	1339	ALLOCATE( wall_flags_4(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 ( 5 )
	1344	ALLOCATE( wall_flags_5(nzb:nzt_mg(i)+1, &
	1345	nys_mg(i)-1:nyn_mg(i)+1, &
	1346	nxl_mg(i)-1:nxr_mg(i)+1) )
	1347
	1348	CASE ( 6 )
	1349	ALLOCATE( wall_flags_6(nzb:nzt_mg(i)+1, &
	1350	nys_mg(i)-1:nyn_mg(i)+1, &
	1351	nxl_mg(i)-1:nxr_mg(i)+1) )
	1352
	1353	CASE ( 7 )
	1354	ALLOCATE( wall_flags_7(nzb:nzt_mg(i)+1, &
	1355	nys_mg(i)-1:nyn_mg(i)+1, &
	1356	nxl_mg(i)-1:nxr_mg(i)+1) )
	1357
	1358	CASE ( 8 )
	1359	ALLOCATE( wall_flags_8(nzb:nzt_mg(i)+1, &
	1360	nys_mg(i)-1:nyn_mg(i)+1, &
	1361	nxl_mg(i)-1:nxr_mg(i)+1) )
	1362
	1363	CASE ( 9 )
	1364	ALLOCATE( wall_flags_9(nzb:nzt_mg(i)+1, &
	1365	nys_mg(i)-1:nyn_mg(i)+1, &
	1366	nxl_mg(i)-1:nxr_mg(i)+1) )
	1367
	1368	CASE ( 10 )
	1369	ALLOCATE( wall_flags_10(nzb:nzt_mg(i)+1, &
	1370	nys_mg(i)-1:nyn_mg(i)+1, &
	1371	nxl_mg(i)-1:nxr_mg(i)+1) )
	1372
	1373	CASE DEFAULT
[254]	1374	message_string = 'more than 10 multigrid levels'
	1375	CALL message( 'init_pegrid', 'PA0238', 1, 2, 0, 6, 0 )
[114]	1376
	1377	END SELECT
	1378
	1379	ENDDO
	1380
	1381	ENDIF
	1382
[1]	1383	END SUBROUTINE init_pegrid

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

Download in other formats:

| Impressum | ©Leibniz Universität Hannover |