- UQ-TMC: Handle the NJOY dependency
- UQ-TMC: separate the preparation of run from the execution of the tmc loops (xml files outside the loop, tmc-manager like this: openmc-dev/openmc#3508)
- UQ-TMC: implement list of models in tmc engine
- UQ-TMC: the first realization should be the original xs "mean" (baseline)
- UQ-TMC: allow more options to user (e.g. library to perturb)
- UQ-TMC: allow a list of nuclides to perturb
- UQ-TMC: allow list of nuclides and perturbations from specifications file
- Add capability of merging multiple surface results for oktavian (should happen only with surface tallies)
- Add expected tally results shape in
schemaandspecificationsto deal with the point here above - Try to manage results as numpy arrays instead of pandas dataframes in
OpenmcBenchmark.postprocess() - Should we use
h5netcdfengine for saving/opening datasets inh5files or stick with the defaultnetcdf4? - Do we wanna use
makefundecorator for wrapped methods (e.g.OpenmcBenchmark.run())?
- Add tests to everything
- Use
specificationsvalidation againstschemafunction and script - Open and inspect a
specificationsfile (justmetadata) - Instantiate an
OpenmcBenchmarkobject, build model, run simulation - Postprocessing and visualization
- introduction and motivation
- V&V explanation
- Definition of benchmark
specifications - Benchmarks
- Oktavian description
- Oktavian results