Rewrite particle output to use deal.II functionality #1869
Conversation
There was a problem hiding this comment.
I like it a lot that this will make the particle output look a lot like the regular field output. Nice work!
What happens to the current code -- does it continue to work if we have an older deal.II? I didn't see that backward compatibility shim, but maybe I missed where you do that.
| * of dim (the dimension in which this zero-dimensional particle lives). | ||
| */ | ||
| template<int dim> | ||
| class ParticleOutput : public dealii::DataOutInterface<0,dim> |
There was a problem hiding this comment.
Don't you have to guard this class with #if DEAL_II_VERSION_GTE(9,0) as well?
source/postprocess/particles.cc
Outdated
| { | ||
| template<int dim> | ||
| ParticleOutput<dim>::~ParticleOutput() | ||
| {} |
There was a problem hiding this comment.
do you need this destructor?
source/postprocess/particles.cc
Outdated
| dataset_names.push_back(field_name); | ||
| else | ||
| for (unsigned int component_index=0; component_index<n_components; ++component_index) | ||
| dataset_names.push_back(field_name + "[" + Utilities::to_string(component_index) + "]"); |
There was a problem hiding this comment.
I don't think [ and ] are valid names for data set names. You may have to just append the index with an underscore.
source/postprocess/particles.cc
Outdated
| for (unsigned int property_index = 0; property_index < properties.size(); ++property_index) | ||
| patches[i].data(property_index+1,0) = properties[property_index]; | ||
|
|
||
| } |
There was a problem hiding this comment.
no need for the empty line
gassmoeller
force-pushed
the
refactor_particle_output
branch
from
7b6db7f
to
48b929c
Compare
gassmoeller
force-pushed
the
refactor_particle_output
branch
2 times, most recently
from
be910e6
to
2f7639e
Compare
|
Yes, I did not check the old functionality. This is fixed now (should work with 8.5.0 and 9.0.pre). I also removed the empty destructor and updated the test results. |
|
Woah, I have been waiting for this day; compressed particle visualization data, oh man pretty excited :). I don;t want to slow things down but I am still interested in supporting writing out multiple formats for particles. I will export the relevant code of PR #1049 to this code branch. This will take me a day or two. |
|
Sounds good. From a first glance I would think it should fit neatly into the new structure, if you make output_format a list/vector of unique formats, and then loop over the if/else if branches in particles.cc:357-492. I will check in the meantime what happens with the test failures. |
gassmoeller
force-pushed
the
refactor_particle_output
branch
from
2f7639e
to
aa2fc84
Compare
…rmats Add functionality to choose a list of particle output formats
|
This should be good to go. Supersedes #1049. |
cookbooks/shell_simple_2d.prm
Outdated
| @@ -70,7 +70,7 @@ end | |||
|
|
|||
|
|
|||
| subsection Postprocess | |||
| set List of postprocessors = visualization, velocity statistics, temperature statistics, heat flux statistics, depth average | |||
| set List of postprocessors = visualization, velocity statistics, temperature statistics, heat flux statistics, depth average, Stokes residual | |||
There was a problem hiding this comment.
was this change on purpose?
|
|
||
| private: | ||
| /** | ||
| * Implementation of the base class. See there for information. |
There was a problem hiding this comment.
here and below, I think the proper text ought to be "Implementation of the corresponding interface of the base class."
We do let doxygen import the deal.II documentation. I wonder whether, if you just omit the comment, you get to see the corresponding documentation of the function in the base class. But that's really also not very important to address.
source/postprocess/particles.cc
Outdated
| { | ||
| const unsigned n_components = property_information.get_components_by_field_index(field_index); | ||
| const std::string field_name = property_information.get_field_name_by_index(field_index); | ||
| // If it is a 1D element, or a vector, print just the name, otherwise use [] |
There was a problem hiding this comment.
otherwise use [] -> otherwise, append the index after an underscore
| patches[i].n_subdivisions = 1; | ||
| patches[i].data.reinit(property_information.n_components()+1,1); | ||
|
|
||
| patches[i].data(0,0) = particle->second.get_id(); |
There was a problem hiding this comment.
We must be building the worst-suited data structure to represent particle data here :-) But that's what it is, nothing that can be done about it ;-)
There was a problem hiding this comment.
I think it might be worthwhile specializing Patch<0,spacedim> to avoid some of the member variables. But that can wait.
gassmoeller
force-pushed
the
refactor_particle_output
branch
from
a16f921
to
4191fea
Compare
|
I addressed the comments. Should be ready to go. |
This PR uses the recently introduced funtionality of deal.II (dealii/dealii#308) to output zero-dimensional output features (like particles). It essentially offers the same functionality that
Visualizationalready implements, but for the particles. With this PR theparticle/output/plugins will no longer be necessary once we require deal.II 9.0. @hlokavarapu: This will conflict with #1049. Are you still interested in that functionality? Then we should think about integrating that behavior into this PR.Things still to do: