Third analysis todo list¶
This is supposed to be a list of all the things that we ought to have done for 2nd ana but deferred. I'm sure this list will continue to grow. Please help to keep it up to date and categorize it appropriately. Eventually prioritize it.
- Fix mis-mapped FD channels (3x) and ND channels (2x) in DaqChannelMap rather than just masking them out
- DQ should add horn-current-study runs into list of bad runs removed manually (instead of removing all runs that week). Then the period/epoch page should be updated to have 2c go through ND run 10992. Production definitions like base_nd_period2 will also need to change to actually get these files back into the analysis pile.
- Develop data-based threshold corrections in calibrations
- Drift calibrations
- Improve x-view muon catcher attenuation fitting
- Try attenuation + rolloff + LOWESS fit for better attenuation fitting at the cell's rolloffs. not just attenuation + LOWESS fit.
- Detector alignment
- Remove all traces of PECorr outside the calibration itself
- Get all plots into automated validation system
- Tests for new calibcsvs releases/calibrator code
Read calibration csvs with custom code to eliminate need for chunking and improve FD job startup times
- Model beam timing structure and slipstacking
- Match beam conditions to data
- Fix cleanup of flux files at job completion; it's not clearing out /var/tmp when complete
- Move to G4numi flux
Detector and GENIE Simulations¶
- Update genie to pick up latest MEC models
- Update geant to pick up new neutron simulation
- Extend the noise model to sample the full ADC range
- Update to new noise model/Create new noise model
- Scheme to generate ND genierw faster, e.g. only on slices that pass analysis selections
- Update timing fit to new pulse shape parametrization, optimize parameters
- Update ReadoutSim parameters based on best info from data parameter tuning
- Simulate ND rock events more efficiently
- Cherenkov light
- Difference between X & Y view
- New geometry files
- Investigate size improvements from CellHit pointing to RawDigit
- Switch to HoughVertex
- Better CAF flag for "truly in detector". Also flag for in M/C.
- Move over environment variables to fcl parameters where possible, eg pid library version, calibration version
- New CAFs
- Move modules over to Table, replacing ParameterSet for better fcl error detection
- Reduce memory footprint of art jobs - Making good progress (E. Niner)
- Make it easier to upgrade nutools versions (novadaq and novaddt dependencies)
- Reduce (or make configurable) log file verbosity.
- Teach IFDBSpillInfo to correctly handle missing variables
Update defman for new naming schemes/metadata, teach it about decafs- Replaced with defgen.sh Only declare files after confirming successful copy-back. Add job-specific metadata (Cluster, Job, fcl, NGU version, project name, hostname) Fix metadata defaulting in metadata module- Not totally reworked, but operational issues resolved Better handling of default "none" for skim, special, systematic, etc.- Not totally reworked, but operational issues resolved Reduce number of decafs Merge nus caf/decaf making with standard production
- Produce attenprofs etc as part of std keepup
- Production metadata cleanups (especially auditing duplicate fields)
- remove "nova.label: beta" constraint from official SAM definitions
- Testing suite for NGU
- Put sandboxes together in a ProdSandPaths.fcl to avoid repeating everywhere, so nothing gets missed.
- Generate new LEM library with GeVs
- Separate versioning for LEM
- Optimize division of labor between LEMClient and LEMWorker
- Base MRBrem on MRCC tools, make it actually fast.
- Numu hadronic energy for FD has different fits for different run periods. Currently, the ideal conditions MC run numbers are also hardcoded in. In the future, this would probably be more robust if it instead checked for things like gain setting and coarse/fine timing.
- Disentangle numu containment from cosmic rejection, making them separate modules.
- Retrain remid on modern MC, and switch to a more memory-efficient MVA method at the same time