[visit-users] Visit Input File
miller86 at llnl.gov
Mon Mar 15 12:10:56 EDT 2010
On Mon, 2010-03-15 at 07:18, Allen Barnett wrote:
> Hi: We are working to add a Visit input file option to the output
> choices from our analysis codes. Silo looks like the right choice but we
> are having difficulty getting the results we would like to see in Visit
> 1.12.2. So, I guess my question is: What's the right way to create a
> Visit file for our data? Our mesh is an unstructured tet mesh. The cells
> of the mesh belong to different material regions (i.e., a particular
> cell belongs to one and only one region). Various values are computed in
> each cell by the analysis codes; it appears sufficient to include the
> cell-average values.
> We would like to use the subsetting capability in Visit to select
> various material regions to display. We tried the MRG approach described
> in the Silo manual, but Visit will not show us the subsets we (think we)
> described in the Silo file. Basically, we supplied a Groupel list which
> divided the mesh cells up into segements and then assigned each segment
> to a Mrgtree node corresponding to a material reigon. It's quite likely
> we didn't get this right, but we're stumped now.
At present, you are also required to following the same naming scheme
for a few of the members of the MRGtree. If you run with '-debug 3' on
the command line, you should get engine debug logs that might suggest
any problems there. For example, the top node in mrg tree for amr
decomposition needs to be named 'amr_decomp', there needs to be a
'levels' child of it and a 'patches' child of it. The mrg variables
holding refinement ratios and logical extents need to be named with
'ratio' and 'ijk' occuring somewhere in them, repsectively.
If this is all true and the file is small enough, email it to me and I
can take a look.
> Also, some of our analysis tools are distributed-memory-parallel and use
> a spatial domain decomposition. We would like to see the mesh exploded
> based on the spatial decomposition (along with the material regions).
> The MRG seems ideal for supplying more than one grouping of the cells,
> but again, we can't quite make Visit show us the different mappings.
> Any help would be greatly appreciated.
Mark C. Miller, Lawrence Livermore National Laboratory
================!!LLNL BUSINESS ONLY!!================
miller86 at llnl.gov urgent: miller86 at pager.llnl.gov
T:8-6 (925)-423-5901 M/W/Th:7-12,2-7 (530)-753-851
More information about the visit-users