Re: [Libmesh-devel] NodeElems with ExodusII_IO::write_discontinuous_exodusII

[David K. <dav...@ak...>]

On Fri, Oct 21, 2016 at 5:22 PM, John Peterson <jwp...@gm...> wrote:

>
>
> On Fri, Oct 21, 2016 at 2:57 PM, David Knezevic <
> dav...@ak...> wrote:
>
>> It seems that ExodusII_IO::write_discontinuous_exodusII doesn't handle
>> NodeElems correctly. NodeElems get plotted as ExodusII SPHERE elements, and
>> this works fine with continuous plotting, but seems to be wrong in the
>> discontinuous plotting case.
>>
>> I've attached a modified version of systems_of_equations_ex6 that
>> illustrates this (change write_discontinuous_exodusII to
>> write_equation_systems in the example to see the correct locations).
>> Paraview plots the SPHERE elements as dots, so that is useful for checking
>> the result.
>>
>> John, I thought you may have an idea of where to look to fix this? My
>> first guess is that it might be related to the "FIXME" comment
>> in ExodusII_IO_Helper::write_elements?
>>
>
> You're talking about the FIXME on line 1449?  That should only be a
> problem if the mapping between your Mesh's node numbering and Exodus's node
> numbering, as stored in the libmesh_node_num_to_exodus, is something
> other than the trivial off-by-one map.
>
> I don't know if the presence of NodeElems, which are treated as Elems (?),
> would have an effect on this, but I suppose it's possible.  What is the
> error message you're getting?
>


I think I found the issue, see the PR I just made (#1128). Looks like the
code previously assumed num_nodes_per_elem was the same for the entire
mesh, so that didn't work properly for meshes (like my test case) with
different element types.

David