Tests, tests, tests

[johannbrehmer]

MadMiner should have as many (unit) tests as possible. So far, in tests/ we have two simple tests that check that all imports work, and that some of the ML functionality works.

Here are some ideas for different tests:

☐ the basic functionality of the MadMiner class: setting up parameter spaces and benchmarks, loading and saving MadMiner files
☐ the morphing functionality
☐ running MadGraph and Pythia from a MadMiner instance

☐ reading in LHE files with LHEProcessor
☐ running Delphes from DelphesProceessor
☐ different ways to define and evaluate observables, both in LHEProceessor and DelphesProcessor
☐ cuts in LHEProcessor and DelphesProcessor

☐ creating test samples with SampleAugmenter
☐ creating SALLY / SALLINO training samples with SampleAugmenter
☐ creating RASCAL / ALICES / ... training samples with SampleAugmenter
☐ creating SCANDAL / MAF training samples with SampleAugmenter

☐ training and evaluating a SALLY estimator
☑ training and evaluating a RASCAL / ALICES / ... estimator
☐ training and evaluating a SCANDAL / MAF estimator