[MOAB-dev] Fwd: About reading vtk in parallel mode.

[Iulian Grindeanu]

Hi Ilya, 

----- Original Message -----

| Hi Iulian!
| Thank you for your answer.
| I solved my problem with reading h5m meshes, converted from vtk
| files. Utility addfield realy helped me. Thank you again!
| And about problem that the meshes don't cover the same domain. Yes,
| this is intentional. In my research I'm going to use the similar
| covering of meshes.
| Could you please explain or give advice if it is possible to use
| coupling solution only in the region, that is common for both
| meshes.
At the core of the method is a locate function, which tries to locate points of interest (from the target mesh) in the source mesh. You will have to call that yourself, and then handle the error. 

| In {$MOAB_dir}/tools/mbcoupler/mbcoupler_test.cpp on line 639 I found
| subtract function. Maybe I can use it in my situation or something
| analogous...
that subtract is a "moab::Range" subtract, not a geometry subtract. Range is a std-like container for EntityHandle. 

You can (probably) use "locate" to intersect the target and the source meshes. 

Anyway, this is the "geometric" part of the coupling. After you locate the target points in the source mesh, you have to decide how to transfer the field of interest from source to target. 

| P.S. Unfortunatly I can't check par_coupler_test because of problems
| with compilation on my cluster, but I check mbcoupler_test running
| with 2 processors, this the same, if I'm not wrong.

yes, it should be the same thing; 
What compilation problems? If tools/mbcoupler is compiled fine, test/parallel should compile fine too. Maybe there are issues with "make check", as it will try to run the tests on the cluster. But compilation should be fine. 

| Thank you. _______________________________
| Best regards, Ilya

Thanks, 
Iulian 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/moab-dev/attachments/20130909/098390e5/attachment.html>