Validation Suite

[carlu]

I'm currently working on a sort code to process the simulation output and produce spectra which can be directly compared to experiment for the purposes of validating the simulation. The beginnings of the code are completed, it currently produces energy spectra for each crystal. Still to do:

• Include realistic energy resolution.
• Include realistic trigger threshold.
• Model TIGRESS sub-crystal segmentation.
• Produce code to normalise total counts and overlay result with experimental spectra.

[bkatiemills]

Some goals for the validation suite were hashed out by the collaboration today:

• Three validation scales would be useful:
  • fast validation, a series of physics tests that can run in < 1 hour to be run before every pull request, designed to catch as much as possible qucikly.
  • in-depth validation, a series of physics tests that can take up to 1 day to run, to be run on a weekly basis to look for regressions that snuck past the fast validation.
  • versioning validation, a check to see if two outputs are exactly identical, intended to check if anything relevant has changed between Geant4 patches.

Several efficiency tests were proposed:

• GRIFFIN dead layer efficiencies
  • check efficiency, photopeak and totals
  • refit deadlayer v. efficiency to make sure it hasn't changed
• per-segment spectra; chi2 tests to detect differences
• hit pattern comparisons
• coincidence efficiency test on 60Co