Addition of a filter system and `[Filters]` block

nuclearkevin commented 1 month ago

Reason

This PR implements a filter system, an associated block in Cardinal's input syntax ([Problem/Filters]), and four new filters which tallies added through the [Problem/Tallies] block can access. The added filter objects are:

EnergyFilter: a thin wrapping of an OpenMC EnergyFilter.
PolarAngleFilter: a thin wrapping of an OpenMC PolarFilter.
AzimuthalAngleFilter: a thin wrapping of an OpenMC AzimuthalFilter.
FromXMLFilter: makes an arbitrary OpenMC filter (added in the tallies.xml file) available to Cardinal tallies through the filter's id.

The input syntax is similar to the recently added tally system:

[Problem]
  type = OpenMCCellAverageProblem
  verbose = true
  power = 1e4
  temperature_blocks = '100'
  cell_level = 0
  initial_properties = xml

  source_rate_normalization = 'kappa_fission'

  [Tallies]
    [Heating]
      type = CellTally
      score = 'kappa_fission'
      blocks = '100 200'
    []
    [Flux]
      type = CellTally
      score = 'flux'
      blocks = '100 200'
      filters = 'Energy Polar'
    []
  []

  [Filters]
    [Energy]
      type = EnergyFilter
      # CASMO 2 group structure for testing. May result in some missed particles
      energy_boundaries = '0.0 6.25e-1 2.0e7'
    []
    [Polar]
      type = PolarAngleFilter
      num_equal_divisions = 2
    []
  []
[]

(tests/neutronics/filters/multi_filter.i)

Design

The new filters system has the following objects other than the previously mentioned filters:

FilterBase: A class which wraps a single OpenMC filter. FilterBase classes are constructed by an OpenMCCellAverageProblem through the addFilter(...) function, which mirrors the tally system and other MOOSE object construction patterns. Filters derived from FilterBase must override the binName(...) function, which returns a unique name for each bin in the filter to enable output formatting. Otherwise, classes which derive from FilterBase must ensure that _filter and _filter_index are populated by the derived filter's constructor.
AddFilterAction: An Action which is responsible for adding FilterBase objects to an OpenMCCellAverageProblem. This action calls addFilter(...) when the simulation reaches the new add_filters task during execution. Filters are added before tallies and after the initialization of the problem.

The filters added through the filters system must use discrete bins as the tallies are still normalized by a sum over all bins. This restriction could be lifted, however this would require a substantial change to tally normalization to allow bins that contain coefficients to be skipped when computing the sum over the entire tally.

In addition to the above, this PR modifies the tally system to reduce the amount of duplicate code and fixes a bug in both MeshTally and CellTally which caused a memory leak when using a MOABSkinner. These changes can be removed from this PR and submitted in their own PR to expedite merging, if that's preferred.

Impact

Enables multi-physics coupling to other codes which require fast fluxes / fluences (such as Bison).
Allows for greater flexibility when post-processing tallies.
Opens the door to the generation of surface / volume sources for deterministic radiation transport (https://github.com/openmc-dev/openmc/issues/3031).
Allows for the comparison of multi-group reaction rates / fluxes between deterministic radiation transport codes (such as Griffin) and Cardinal.

Closes #602.

moosebuild commented 1 month ago

Job Documentation on 51cfa7f wanted to post the following:

View the site here

This comment will be updated on new commits.

moosebuild commented 1 month ago

Job Coverage on 51cfa7f wanted to post the following:

Coverage

	d46e57	#951 51cfa7
	Total	Total	+/-	New
Rate	93.40%	93.19%	-0.21%	86.55%
Hits	7472	7620	+148	206
Misses	528	557	+29	32

Diff coverage report

Full coverage report

Warnings

New new line coverage rate 86.55% is less than the suggested 90.0%

This comment will be updated on new commits.

tjlaboss commented 1 month ago

👀

moosebuild commented 1 week ago

Job Precheck on 544eedb wanted to post the following:

Your code requires style changes.

A patch was auto generated and copied here
You can directly apply the patch by running, in the top level of your repository:

curl -s https://mooseframework.inl.gov/cardinal/docs/PRs/951/clang_format/style.patch | git apply -v

Alternatively, with your repository up to date and in the top level of your repository:

git clang-format d46e57f4d038ff25544f4ab0fcd2f1bf66045212

aprilnovak commented 1 week ago

just needs a rebase and then I am good to merge

moosebuild commented 1 week ago

Job Documentation, step Sync to remote on e76ca06 wanted to post the following:

View the site here

This comment will be updated on new commits.

moosebuild commented 1 week ago

Job Coverage, step Generate coverage on e76ca06 wanted to post the following:

Coverage

	a458e5	#951 e76ca0
	Total	Total	+/-	New
Rate	93.42%	93.20%	-0.21%	86.55%
Hits	7477	7625	+148	206
Misses	527	556	+29	32

Diff coverage report

Full coverage report

Warnings

New new line coverage rate 86.55% is less than the suggested 90.0%

This comment will be updated on new commits.

nuclearkevin commented 1 week ago

@aprilnovak I've finished rebasing this, it should be good to go now!

neams-th-coe / cardinal