Hi Ali,
first of all, you should not rely on comparing indices from the vtk
output with indices that you generate via the dofMapper. In sequential
settings, they might coincide (but there is no requirement for this),
while, as you observe, in parallel settings they might be completely
different.
Please have a look at the DUNE data decomposition model for parallel
grids, see for example Section 8.1 in
http://www.dune-project.org/doc/grid-howto/grid-howto.pdf
In parallel, there are vertices (edges, facets or even elements) that
have the same geometry but are stored on more than one process. In
general, they will have different indices. You could try to work with ID
sets instead:
http://www.dune-project.org/doc/doxygen/dune-grid-html/class_dune_1_1_id_set.html
But maybe that's too much effort. You could also not only print the
index, but in addition also the process rank
this->gridView().comm().rank()
In a crude approach, you simply run your chemistry stuff multiple times
on the vertices shared by more than one process. You will do more work
than you have to, but at least things should stay simple.
Kind regards
Bernd
On 07/14/2014 06:19 AM, Ali Nowamooz wrote:
Dear DuMuX developer,
I try to write gas-phase saturations of all nodes (in 2p2c model) in a
*.txt file for each time integration, and use them next in another
code to conduct some geochemical simulations.
I followed the way that results are written in vtk files and prepared
void postTimeStep as shown below (I didn't show it here, but
(*Saturation_...) are all written into a text file).
It works perfectly when I run my simulations on one (1) processor.
However, on several processors, I obtain a larger global Idx than
those I find in vtk files for each rank. For example, in rank 0,
maximum global Idx is 7340 in *.txt file while in vtk file is only
5251. I tried to figure out what is the cause of this difference, and
I mentioned that in *.txt files for all nodes located on the interface
of two ranks (processors), I have two different global Idxs, for
example (I give here the value of saturation on a vertex:
globalIdx, x(m), y(m), z(m), and gas-phase saturation(-):
445, -10.53, 24, 4.8, 0.55
and
5346, -10.53, 24, 4.8, 0
As can be seen, for the same global position I have two different
Idxs. How can it be possible? How can we have two different indxs for
the same global position? The gas-phase saturation of globalIdx 445
corresponds to the value stored in the vtk file for the same position
(-10.53, 24, 4.8). What does the second index represent? Is there
any problem in the code that I used for writing the results?
Hope that I was clear in my explanations
Please let me know if you need more details
Thanks in advance for your help
Best regard
Ali NOWAMOOZ
void postTimeStep() {
FVElementGeometry fvGeometry;
ElementVolumeVariables elemVolVars;
// get the number of degrees of freedom
unsigned numDofs = this->model().numDofs();
typedef Dune::BlockVector<Dune::FieldVector<double, 1> > ScalarField;
ScalarField *Saturation_nw =
this->resultWriter().allocateManagedBuffer(numDofs);
ScalarField *Saturation_x =
this->resultWriter().allocateManagedBuffer(numDofs);
ScalarField *Saturation_y =
this->resultWriter().allocateManagedBuffer(numDofs);
ScalarField *Saturation_z =
this->resultWriter().allocateManagedBuffer(numDofs);
ScalarField *Saturation_scv =
this->resultWriter().allocateManagedBuffer(numDofs);
(*Saturation_nw) = 0;
(*Saturation_x) = 0;
(*Saturation_y) = 0;
(*Saturation_z) = 0;
(*Saturation_scv) = 0;
ElementIterator eIt = this->gridView().template begin<0>();
ElementIterator eEndIt = this->gridView().template end<0>();
for (; eIt != eEndIt; ++eIt){
elemVolVars.update(*this,*eIt,fvGeometry,false);
fvGeometry.update(this->gridView(), *eIt);
for (int scvIdx = 0; scvIdx < fvGeometry.numScv; ++scvIdx){
int gIobaldx = this->model().dofMapper().map(*eIt, scvIdx, dofCodim);
const GlobalPosition gpos = eIt->geometry().corner(scvIdx);
(*Saturation_nw)[ gIobaldx] = elemVolVars[scvIdx].saturation(gPhaseIdx);
(*Saturation_x)[ gIobaldx] = gpos[0];
(*Saturation_y)[ gIobaldx] = gpos[1];
(*Saturation_z)[ gIobaldx] = gpos[2];
(*Saturation_scv)[ gIobaldx] += 1;
}
}
}
_______________________________________________
Dumux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
--
_______________________________________________________________
Bernd Flemisch phone: +49 711 685 69162
IWS, Universität Stuttgart fax: +49 711 685 60430
Pfaffenwaldring 61 email: [email protected]
D-70569 Stuttgart url: www.hydrosys.uni-stuttgart.de
_______________________________________________________________
_______________________________________________
Dumux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
