From 99c8f22ca1ec2816ac1f37bb4449735f83f7c8e4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Franck=20P=C3=A9rignon?= <franck.perignon@imag.fr>
Date: Wed, 13 Jul 2011 11:32:15 +0000
Subject: [PATCH] Parmes
---
HySoP/INSTALL | 12 +++
HySoP/src/main/Domain.f90 | 4 +-
HySoP/src/main/FieldsComputation.f90 | 18 +++-
HySoP/src/main/NavierStokes3D.f90 | 87 ++++++++-------
HySoP/src/main/Particles.f90 | 151 +++++++++++++++++++--------
HySoP/src/main/Penalize.f90 | 8 +-
HySoP/src/main/Topology.f90 | 6 +-
HySoP/src/main/io_vtk.f90 | 8 +-
8 files changed, 200 insertions(+), 94 deletions(-)
diff --git a/HySoP/INSTALL b/HySoP/INSTALL
index e69de29bb..ede41e12c 100644
--- a/HySoP/INSTALL
+++ b/HySoP/INSTALL
@@ -0,0 +1,12 @@
+
+1- create a "build" directory anywhere
+
+
+2- cmake path-to-sources-of-Parmes -DPPMCore_DIR=/home/perignon/install/ppmcore/share/CMake/ -DPPMNumerics_DIR=/home/perignon/install/ppmnumerics/share/CMake/
+
+3- make -j 8
+
+4- mpirun -np 8 ./parmesRun
+
+
+
diff --git a/HySoP/src/main/Domain.f90 b/HySoP/src/main/Domain.f90
index 7f6d864c3..2487885d4 100755
--- a/HySoP/src/main/Domain.f90
+++ b/HySoP/src/main/Domain.f90
@@ -40,7 +40,7 @@ contains
domain_maxCoords(2) = zFolke
! 0.3: boundary layer
domain_minCoords(3) = -zFolke - 0.3
- domain_maxCoords(3) = zFolke + 0.3
+ domain_maxCoords(3) = zFolke + 0.3
! Boundary conditions and ghosts
allocate(domain_bc(2*dime), domain_ghostsize(dime), stat = istat)
@@ -56,7 +56,7 @@ contains
allocate(grid_resolution(dime),grid_step(dime),stat=istat)
if(istat.ne.0) stop 'grid_resolution, allocation failed'
- grid_resolution = 65
+ grid_resolution = 129
grid_step = (domain_maxCoords - domain_minCoords)/(real(grid_resolution,mk)-1.)
boundary_layer_min = 0
boundary_layer_max = 0
diff --git a/HySoP/src/main/FieldsComputation.f90 b/HySoP/src/main/FieldsComputation.f90
index 70ce7f45c..a4234bb7d 100755
--- a/HySoP/src/main/FieldsComputation.f90
+++ b/HySoP/src/main/FieldsComputation.f90
@@ -9,7 +9,7 @@ module FieldsComputation
private
- public :: fit_velocity, computeFlowRate
+ public :: fit_velocity, perturb
contains
@@ -55,8 +55,22 @@ contains
print *, computed_flowrate, '//',surf
vel(1,:,:,:,1) = vel(1,:,:,:,1) + constant
-
+
end subroutine fit_velocity
+
+ subroutine perturb(vel, time)
+ ! Velocity, intent(inout)
+ real(mk), dimension(:,:,:,:,:), pointer :: vel
+ ! current time
+ real(mk) :: time
+
+ real(mk) :: pi
+
+ pi = 4.0*atan(1.0_mk)
+
+ vel(2,:,:,:,1) = sin(pi*(time-3.0))
+
+ end subroutine perturb
subroutine computeFlowRate(vel, stepSize, topo, flowRate)
diff --git a/HySoP/src/main/NavierStokes3D.f90 b/HySoP/src/main/NavierStokes3D.f90
index dc9f723e7..2c872d60f 100755
--- a/HySoP/src/main/NavierStokes3D.f90
+++ b/HySoP/src/main/NavierStokes3D.f90
@@ -20,7 +20,7 @@ module ppmNavierStokes3D
use penalisation, only : penalisation_init, penalise_velocity, get_drag
- use FieldsComputation, only : fit_velocity,computeFlowRate
+ use FieldsComputation, only : fit_velocity,perturb
use vectorcalculus, only: curldf4, computeRHS, norm1, norm2, normInf
@@ -55,7 +55,7 @@ contains
! debug mode
integer :: debug
! error status
- integer :: info
+ integer :: info,rank,sizeZ
!======================
! Init ppm
@@ -82,6 +82,7 @@ contains
! Fields allocation
!======================
call init_fields(domain_ghostsize, topo)
+
!======================
! Init Physics
!======================
@@ -94,12 +95,30 @@ contains
velocity = 0.0_mk
call init_vorticity(vorticity,topo,grid_step,boundary_layer_min,boundary_layer_max)
- call penalisation_init(topo,grid_step,boundary_layer_min,boundary_layer_max)
+ call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
+!!$ vorticity = 0.0_mk
+!!$ if(rank==0) then
+!!$ sizeZ = size(vorticity,4)
+!!$ print *, 'shape : ', shape(vorticity), sizeZ
+!!$ vorticity(2,4,5,5,:) = 12.0
+!!$ else
+!!$ vorticity(2,4,5,1,:) = 12.0
+!!$ end if
+!!$ call MPI_BARRIER(MPI_COMM_WORLD,info)
- !======================
- ! Init Particles
- !======================
+ call ppm_map_field_ghost_get(topo%ID,topo%mesh(1)%ID,domain_ghostsize,info)
+ call ppm_map_field_push(topo%ID,topo%mesh(1)%ID,vorticity, 3, info)
+ call ppm_map_field_send(info)
+ call ppm_map_field_pop(topo%ID,topo%mesh(1)%ID,vorticity,3,domain_ghostsize,info)
+ !============================================
+ ! Init penalization : set chi functions
+ !============================================
+ call penalisation_init(topo,grid_step,boundary_layer_min,boundary_layer_max)
+
+ !============================================
+ ! Build a first set of particles
+ !============================================
call init_particles(vorticity,topo%ID,topo%mesh(1)%ID,velocity)
print *, "end of parmes:ppm:init_client"
@@ -120,12 +139,14 @@ contains
Re = 133.4
nu = 1./Re
initial_time = 0.0
- final_time = 10.0
- time_step = 0.05
+ final_time = 60.0
+ time_step = 0.04
current_time = initial_time
iter = 1
- max_iter= 10
-! call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
+ max_iter= 10000
+
+
+ call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
! call MPI_COMM_SIZE(MPI_COMM_WORLD,nbProcs,info)
! Initialisation stuff
call init_client()
@@ -136,17 +157,9 @@ contains
! Init multigrid works in ppm (or seems to ...) but solve fails.
!call init_multigrid(topo%ID, topo%mesh(1)%ID, domain_ghostsize, domain_bc)
- call MPI_BARRIER(MPI_COMM_WORLD,info)
-
! Analytical solutions
!call rhs_analytic(topo, grid_step, stream_function,vel_ex,vorticity,nu)
- ! Init ghosts ...
- call ppm_map_field_ghost_get(topo%ID,topo%mesh(1)%ID,domain_ghostsize,info)
- call ppm_map_field_push(topo%ID,topo%mesh(1)%ID,vorticity, 3, info)
- call ppm_map_field_send(info)
- call ppm_map_field_pop(topo%ID,topo%mesh(1)%ID,vorticity,3,domain_ghostsize,info)
-
derive = ppm_poisson_drv_curl_fd4
call init_poisson_solver(vorticity,velocity,topo%ID, topo%mesh(1)%ID,derive)
@@ -155,13 +168,15 @@ contains
print *, '[',rank,'] pas espace/temps ', grid_step, time_step
print *, '[',rank,'] mesh ', shape(topo%mesh(1)%nnodes), topo%mesh(1)%nnodes
- allocate(drag(max_iter))
-
- open(10,file='drag')
-
+ !if(rank==0) then
+ allocate(drag(max_iter))
+ open(10,file='drag')
+ !end if
+ call MPI_BARRIER(MPI_COMM_WORLD,info)
+ print *, 'start iterations ...'
! Time loop
do while(current_time < final_time .and. iter <= max_iter)
- print *, '================ Start iter ',iter,' ================'
+! print *, '================ Start iter ',iter,' ================'
! Compute velocity from vorticity
! Two steps:
@@ -169,22 +184,24 @@ contains
! to be done in ppm routine; indeed we do not have to deal with stream_function.
! - update velocity to fit with required boundary values
call solve_poisson(vorticity, velocity, topo%ID, topo%mesh(1)%ID,domain_ghostsize)
- call MPI_BARRIER(MPI_COMM_WORLD,info)
- !call fit_velocity(velocity, grid_step,lengths,topo,boundary_layer_min,boundary_layer_max)
+ call fit_velocity(velocity, grid_step,lengths,topo,boundary_layer_min,boundary_layer_max)
+ if( current_time>3.0 .and. current_time<4.0) then
+ call perturb(velocity, current_time)
+ end if
! Penalize velocity
call penalise_velocity(velocity,topo,time_step)
call ppm_map_field_ghost_get(topo%ID,topo%mesh(1)%ID,domain_ghostsize,info)
call ppm_map_field_push(topo%ID,topo%mesh(1)%ID,velocity, 3, info)
call ppm_map_field_send(info)
call ppm_map_field_pop(topo%ID,topo%mesh(1)%ID,velocity,3,domain_ghostsize,info)
-
- call get_drag(drag(iter))
- !print *, 'drag...', drag(iter,rank)
- write(10,'(e14.5,e14.5)') current_time, drag(iter)
-!!$ ! Compute the new "penalized" vorticity
+ call get_drag(drag(iter))
+ if(rank == 0) then
+ write(10,'(e14.5,e14.5)') current_time, drag(iter)
+ end if
+ ! Compute the new "penalized" vorticity
call curlDF4(velocity, vorticity,topo, grid_step)
-
+
! Computation of rhs of vorticity equation (on the grid) ...
call computeRHS(velocity, vorticity, rhs, topo, grid_step, nu)
call MPI_BARRIER(MPI_COMM_WORLD,info)
@@ -206,12 +223,13 @@ contains
! Remesh
call remesh(topo%ID, topo%mesh(1)%ID,domain_ghostsize,vorticity)
-! call printToVTK("post",iter, current_time,velocity,vorticity,topo,grid_step)
-
+ if(mod(iter, 10).eq.0) then
+ call printToVTK("post",iter, current_time,velocity,vorticity,topo,grid_step)
+ end if
current_time = current_time + time_step
!current_time = final_time
iter = iter+1
- print *, 'end of iteration ', iter-1
+ !print *, 'end of iteration ', iter-1
call MPI_BARRIER(MPI_COMM_WORLD,info)
enddo
@@ -219,7 +237,6 @@ contains
close(10)
call ppm_finalize(info)
print *, 'end ppm simulation'
-
deallocate(drag)
end subroutine main_client
diff --git a/HySoP/src/main/Particles.f90 b/HySoP/src/main/Particles.f90
index 4e3199e75..b9f9377f5 100755
--- a/HySoP/src/main/Particles.f90
+++ b/HySoP/src/main/Particles.f90
@@ -2,9 +2,9 @@
module Particles
use ppm_module_rmsh, only : ppm_rmsh_create_part,ppm_interp_m2p, ppm_interp_p2m !, ppm_rmsh_remesh
-
+ use ppm_module_impose_part_bc
use client_data, only: mk, dime
-! use mpi
+ use mpi
use ppm_module_data, only : ppm_param_rmsh_kernel_mp4
use ppm_module_map_part
use ppm_module_util_dbg
@@ -47,7 +47,7 @@ contains
! topo and mesh ids
integer, intent(in) :: topoid
integer, intent(in) :: meshid
- integer :: info, newNpart
+ integer :: info, newNpart,rank,i
! velocity on grid
real(mk), dimension(:,:,:,:,:), pointer :: vel
@@ -62,31 +62,66 @@ contains
npart = 0
last_npart = npart
first_call = .TRUE.
-
+ call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
+
! --------> The following call will allocate memory for xp, omp and velo_p, right?
! --------> And deallocation will be handled by ppm also?
call ppm_rmsh_create_part(topoid,meshid,xp,npart,omp,dime,field_on_grid,cutoff,info,.TRUE.,&
field_wp=vel,wp=velo_p,lda2=dime)
+ print *, '[',rank,'] initialisation with ', npart,' particles'
!> --------> Other variables carried by particles but not allocated during ppm_rmsh_create call
! --------> is there a better way to do this, with map_push or other ppm routines?
allocate(rhsp(dime,npart),xp0(dime,npart))
-
- ! --------> Is the following (map_part_global) required? I would say yes but I'm not sure ...
- newNpart = 0
- call ppm_map_part_global(topoid,xp,npart,info) ! positions
- call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
- call ppm_map_part_push(rhsp,dime,npart,info) ! rhs
- call ppm_map_part_push(omp,dime,npart,info) ! vorticity
- call ppm_map_part_push(xp0,dime,npart,info)
- call ppm_map_part_send(npart,newNpart,info) ! send
- call ppm_map_part_pop(xp0,dime,npart,newNpart,info)
- call ppm_map_part_pop(omp,dime,npart,newNpart,info)
- call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
- call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
- call ppm_map_part_pop(xp,dime,npart,newNpart,info)
- npart = newNpart
+ !call ppm_dbg_print_d(topoid,0.0_mk,0,3,info,xp,npart)
+
+! print *, '[',rank,'] shapes ...', shape(xp),'//', shape(omp), '//', shape(velo_p),'//',shape(rhsp),'//', shape(xp0)
+!!$ ! --------> Is the following (map_part_global) required? I would say no but I'm not sure ...
+!!$ newNpart = 0
+!!$ call ppm_impose_part_bc(topoid,xp,npart,info)
+!!$ call ppm_map_part_global(topoid,xp,npart,info) ! positions
+!!$ call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
+!!$ call ppm_map_part_push(rhsp,dime,npart,info) ! rhs
+!!$ call ppm_map_part_push(omp,dime,npart,info) ! vorticity
+!!$ call ppm_map_part_push(xp0,dime,npart,info)
+!!$ call ppm_map_part_send(npart,newNpart,info) ! send
+!!$ call ppm_map_part_pop(xp0,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(omp,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(xp,dime,npart,newNpart,info)
+!!$
+!!$ npart = newNpart
+
+ ! print *, '[',rank,'] post init with ', npart,' particles'
+ ! print *, '[',rank,'] post init xp/omp ', xp(3,:),velo_p(:,:),omp
+!!$ if(rank==1) then
+!!$ do i =1,size(xp,2)
+!!$ xp(3,i) = xp(3,i) - 1.0_mk
+!!$ end do
+!!$ else
+!!$ xp(3,:) = xp(3,:) +1.5
+!!$ end if
+!!$ !call ppm_impose_part_bc(topoid,xp,npart,info)
+!!$ call ppm_map_part_partial(topoid,xp,npart,info) ! positions
+!!$ call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
+!!$ !call ppm_map_part_push(rhsp,dime,npart,info) ! rhs
+!!$ !call ppm_map_part_push(omp,dime,npart,info) ! vorticity
+!!$ !call ppm_map_part_push(xp0,dime,npart,info)
+!!$ call ppm_map_part_send(npart,newNpart,info) ! send
+!!$ !call ppm_map_part_pop(xp0,dime,npart,newNpart,info)
+!!$ !call ppm_map_part_pop(omp,dime,npart,newNpart,info)
+!!$ !call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(xp,dime,npart,newNpart,info)
+!!$ !call ppm_impose_part_bc(topoid,xp,npart,info)
+!!$ npart = newNpart
+!!$ print *, '[',rank,'] post 2 init with ', npart,' particles'
+!!$ print *, '[',rank,'] post 2 init xp/omp ', xp(3,:)!,velo_p(:,:)
+!!$
+!!$
+!!$
end subroutine init_particles
!> Create particles distribution
@@ -98,7 +133,7 @@ contains
! topo and mesh ids
integer, intent(in) :: topoid
integer, intent(in) :: meshid
- integer :: info, newNpart
+ integer :: info, newNpart,rank
integer, dimension(:), pointer :: ghostsize
! velocity on grid
real(mk), dimension(:,:,:,:,:), pointer :: vel
@@ -107,27 +142,33 @@ contains
info = 0
- !call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
+ call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
! Call ppm func to create the particles distribution
last_npart = npart
call ppm_rmsh_create_part(topoid,meshid,xp,npart,omp,dime,field_on_grid,cutoff,info,resetpos,&
field_wp=vel,wp=velo_p,lda2=dime)
!if(associated(xp)) print *, "xp shape:", rank, " ", shape(xp), npart
-
- newNpart = 0
- call ppm_map_part_partial(topoid,xp,npart,info) ! positions
- call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
- call ppm_map_part_send(npart,newNpart,info) ! send
- call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
- call ppm_map_part_pop(xp,dime,npart,newNpart,info)
- npart = newNpart
-
- ! --------> rhsp and xp0 size was computed before rmsh_create and part_pop => must be updated
- ! --------> deallocate/allocate may be done by some ppm routine?
- if (last_npart.lt.npart) then
- deallocate(rhsp,xp0) ! Todo : use ppm_alloc stuff
+ if(last_npart.lt.npart) then
+ deallocate(rhsp,xp0)
allocate(rhsp(dime,npart),xp0(dime,npart))
end if
+! newNpart = 0
+!!$ call ppm_impose_part_bc(topoid,xp,npart,info)
+!!$ call ppm_map_part_global(topoid,xp,npart,info) ! positions
+!!$ call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
+!!$ call ppm_map_part_push(rhsp,dime,npart,info) ! rhs
+!!$ call ppm_map_part_push(omp,dime,npart,info) ! vorticity
+!!$ call ppm_map_part_push(xp0,dime,npart,info)
+!!$ call ppm_map_part_send(npart,newNpart,info) ! send
+!!$ call ppm_map_part_pop(xp0,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(omp,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
+!!$ call ppm_map_part_pop(xp,dime,npart,newNpart,info)
+
+ !print *, 'UPDATE ', npart, newNpart, shape(omp) , '\\', shape(velo_p)
+
+ ! npart = newNpart
! --------> Ok, from here all vars carried by particles must have the following shape : dime,npart
@@ -136,7 +177,7 @@ contains
!> --------> mesh to particles for rhs of vorticity eq. This routine does not resize rhsp, it must be of the right size before the call?
call ppm_interp_m2p(topoid, meshid, xp, npart, rhsp, dime, kernel, ghostsize, rhs,info)
- print *, 'End of update particles'
+! print *, 'End of update particles'
end subroutine update_particles
@@ -156,8 +197,8 @@ contains
real(mk), dimension(:,:,:,:,:), pointer :: rhs
!> local loop indices
- integer :: i, newNpart
- ! call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
+ integer :: i, newNpart,rank
+ call MPI_COMM_RANK(MPI_COMM_WORLD, rank, info)
do i=1,npart
xp0(1,i) = xp(1,i)
@@ -170,7 +211,9 @@ contains
end do
newNpart = 0
-
+
+ ! Particles positions have changed ... we must map between domains
+ ! call ppm_impose_part_bc(topoid,xp,npart,info) ! Useless according to doc, required according to Omar ...
call ppm_map_part_partial(topoid,xp,npart,info) ! positions
call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
call ppm_map_part_push(rhsp,dime,npart,info) ! rhs
@@ -179,10 +222,17 @@ contains
call ppm_map_part_send(npart,newNpart,info) ! send
call ppm_map_part_pop(xp0,dime,npart,newNpart,info)
call ppm_map_part_pop(omp,dime,npart,newNpart,info)
- call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
- call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
+ call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
+ call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
call ppm_map_part_pop(xp,dime,npart,newNpart,info)
-
+ ! print *, 'PUSH 0 ', npart, newNpart, shape(omp) , '\\', shape(velo_p)
+
+ ! Update sizes ...
+! if (newNpart.lt.npart) then
+ ! deallocate(rhsp,xp0) ! Todo : use ppm_alloc stuff
+ ! allocate(rhsp(dime,newNpart),xp0(dime,newNpart))
+ ! end if
+
npart = newNpart
! --------> No need to call any realloc or resize process for xp0, rhsp ... that was done during map_part_pop, right?
@@ -203,11 +253,24 @@ contains
end do
newNpart = 0
+ !call ppm_impose_part_bc(topoid,xp,npart,info)
call ppm_map_part_partial(topoid,xp,npart,info) ! positions
- call ppm_map_part_push(omp,dime,npart,info) ! strengths
+ call ppm_map_part_push(velo_p,dime,npart,info) ! velocity
+ call ppm_map_part_push(rhsp,dime,npart,info) ! rhs
+ call ppm_map_part_push(omp,dime,npart,info) ! vorticity
+ call ppm_map_part_push(xp0,dime,npart,info)
call ppm_map_part_send(npart,newNpart,info) ! send
- call ppm_map_part_pop(omp,dime,npart,newNpart,info) ! strengths
- call ppm_map_part_pop(xp,dime,npart,newNpart,info) ! positions
+ call ppm_map_part_pop(xp0,dime,npart,newNpart,info)
+ call ppm_map_part_pop(omp,dime,npart,newNpart,info)
+ call ppm_map_part_pop(rhsp,dime,npart,newNpart,info)
+ call ppm_map_part_pop(velo_p,dime,npart,newNpart,info)
+ call ppm_map_part_pop(xp,dime,npart,newNpart,info)
+! print *, 'PUSH 1 ', npart, newNpart, shape(omp) , '\\', shape(velo_p)
+! if (newNpart.lt.npart) then
+! deallocate(rhsp,xp0) ! Todo : use ppm_alloc stuff
+! allocate(rhsp(dime,newNpart),xp0(dime,newNpart))
+! end if
+
npart = newNpart
end subroutine push_particles
diff --git a/HySoP/src/main/Penalize.f90 b/HySoP/src/main/Penalize.f90
index 839516870..7543abd80 100755
--- a/HySoP/src/main/Penalize.f90
+++ b/HySoP/src/main/Penalize.f90
@@ -94,13 +94,14 @@ contains
chi_sphere(i,j,k,isub) = lambda
end if
if( (z> boundary_layer_max(3)).or.(z< boundary_layer_min(3))) then
- chi(i,j,k,isub) = lambda
+ chi(i,j,k,isub) = lambda
end if
+ chi(i,j,k,isub) = chi(i,j,k,isub) + chi_sphere(i,j,k,isub)
end do
end do
end do
end do
- chi = chi + chi_sphere
+
nullify(isublist, ndata)
local_drag = 0.0
@@ -149,11 +150,12 @@ subroutine get_drag(drag)
real(mk), intent(out) :: drag
integer :: info
real(mk) :: coef,uinf
+! call MPI_BARRIER(MPI_COMM_WORLD,info)
uinf = 1.0
coef = 2./(uinf**2*2.*sphere_radius)
local_drag = coef*local_drag
- call MPI_AllReduce(local_drag, drag, 1, MPI_DOUBLE_PRECISION, MPI_SUM, MPI_COMM_WORLD,info)
+ call MPI_AllReduce(local_drag, drag, 1, MPI_DOUBLE_PRECISION, MPI_SUM,MPI_COMM_WORLD,info)
end subroutine get_drag
diff --git a/HySoP/src/main/Topology.f90 b/HySoP/src/main/Topology.f90
index c4eaf7272..c9026f6a0 100755
--- a/HySoP/src/main/Topology.f90
+++ b/HySoP/src/main/Topology.f90
@@ -4,7 +4,7 @@ module client_topology
use ppm_module_typedef, only : ppm_t_topo
use ppm_module_mktopo
use ppm_module_data, only : ppm_param_assign_internal, ppm_param_decomp_cartesian, ppm_topo
-! use mpi
+ use mpi
use client_data, only: mk,dime
@@ -48,7 +48,7 @@ contains
istart, ndata, resolution, info)
if(info.ne.0) stop 'Topology:init initialisation failed'
-! call MPI_COMM_RANK(MPI_COMM_WORLD,rank,info)
+ call MPI_COMM_RANK(MPI_COMM_WORLD,rank,info)
! print *, '[',rank,'], topo :', shape(ndata), '//', ndata
! Get the created topology ...
topo=>ppm_topo(topoid)%t
@@ -61,7 +61,7 @@ contains
! sub2proc[i] = global number of the processus on which subdomain i is attached
!
!!$
- print *, '[', rank,'], subs : ', topo%nsublist, '//', shape(topo%isublist),'//', topo%isublist
+ print *, '[', rank,'], subs : ', topo%nsublist, '//', shape(topo%isublist),'//', topo%isublist(1)
!!$ print *, '======================= [', rank,'] ', shape(topo%sub2proc)
!!$ print *, '======================= [', rank,'] ', topo%sub2proc
!!$
diff --git a/HySoP/src/main/io_vtk.f90 b/HySoP/src/main/io_vtk.f90
index 926943d8b..78435c511 100755
--- a/HySoP/src/main/io_vtk.f90
+++ b/HySoP/src/main/io_vtk.f90
@@ -3,11 +3,9 @@ module io_vtk
use client_data, only:mk
use client_topology, only: ppm_t_topo
- !use mpi
+ use mpi
implicit none
-
-
contains
!> print velocity and vorticity to vtk file
@@ -30,7 +28,7 @@ contains
integer :: isub,isubl, rank
! local resolution
integer :: nx,ny,nz
-! integer :: info
+ integer :: info
!local filename (depends on subproc number)
character(len=30) :: localname,buffer1, buffer2,localname2
@@ -45,7 +43,7 @@ contains
!call res_vtk_vit(trim(name_vit))
!call res_vtk_omega(trim(name_omg))
- !call MPI_COMM_RANK(MPI_COMM_WORLD,rank,info)
+ call MPI_COMM_RANK(MPI_COMM_WORLD,rank,info)
do isub=1,topo%nsublist
isubl=topo%isublist(isub)
--
GitLab