-
Notifications
You must be signed in to change notification settings - Fork 11
Snowstorm Files & Details
This page contains details on the Snowstorm samples and production. The Snowstorm method is described in this paper from IceCube and represents a more computationally efficient method of simulating detector uncertainties.
This is the first Snowstorm sample, based on the software used for MiniRun6 (as much as possible). A total of 100 files were produced, giving 1E18 POT for the initial run. More files may be generated as needed and added to the same file locations.
Files are available at NERSC and FNAL. In each directory below there is a sub-directory for each stage of the 2x2 simulation (if present on the cluster). For example, CAFs can be found in MiniRun6_Snowstorm_1E19_RHC.caf.
/global/cfs/projectdirs/dune/www/data/2x2/simulation/productions/Snowstorm_RHC/MiniRun6_Snowstorm_1E19
/pnfs/dune/persistent/users/cuddandr/2x2_production/Snowstorm_RHC/MiniRun6_Snowstorm_1E19
A brief set of points about the simulation:
- Based on MiniRun6 tags/branches for 2x2 Simulation (uses the GENIE and edep-sim files from MR5)
- Charge-only simulation -- no light information
- Variations for this production were applied spill-by-spill (using the spill ID as the RNG seed)
The following parameters were varied in larnd-sim for this production (nominal values reproduced here):
+ consts/detector.py
- E_FIELD = 0.50 # kV/cm
- V_DRIFT = 0.1648 # cm / us
- ELECTRON_LIFETIME = 2.2e3 # us
- LONG_DIFF = 4.0e-6 # cm * cm / us
- TRAN_DIFF = 8.8e-6 # cm * cm / us
+ consts/physics.py
- BIRKS_Ab = 0.800
- BIRKS_kb = 0.0486 # g/cm2/MeV
- W_ION = 23.6e-6
Parameters were varied according to a Gaussian distribution using the spill ID as the RNG seed. Uncertainties are the following (mostly chosen semi-arbitrarily, with the Birks and W_ion parameters informed by some papers):
+ E_FIELD: +/- 2%
+ V_DRIFT: +/- 2% (this is not varied directly, but calculated using the varied e-field)
+ E_LIFET: +/- 5%
+ LN_DIFF: +/- 10%
+ TR_DIFF: +/- 10%
+ BIRKS_A: +/- 2.5%
+ BIRKS_K: +/- 17%
+ W_ION : +/- 2%
The varied parameter values for each spill are saved in the edep-sim HDF5, larnd-sim, ndlar-flow, and (structured) CAFs. For the HDF5 files, there is a new snowstorm dataset with the parameter values and the spill ID. For the CAFs, there is a new TTree called snowstorm with branches for the parameters and spill ID. In all cases there is one entry per spill.
The spill ID for a given interaction can be calculated using the following:
spill_id = i + 1000 * (vertex_id / 10000000) % 100000000
where i is the current entry/spill when iterating over the cafTree in the CAFs and the vertex_id can be accessed from mc.nu in the StandardRecord. The event_id in the MC truth datasets in the HDF5 files is the spill ID.