level3-filter-muon¶
Introduction¶
The muon L3 scripts contained in this project are aimed at developing a through-going muon sample for use at energies of a few hundred or higher. The scripts take events passing either the muon filter or the EHE filter, applies some general cuts for the northern and southern skies, and updates reconstructions for later event selection stages. Checks for icetop hits are included, but not cut on during this stage of the processing. Documentation of an early version of the IC86 muon L3 is available at Wiki: IC86_Muon_Level3.
The muon L3 processing is used by several high energy track selections in the collaboration. Most notably, it is the starting point for both the diffuse (northern tracks) and point source sample selections.
Usage¶
The primary script for users of the muon L3 is python/level3_Master.py. This scrip takes a number of arguments, most of which will not typically need to be changed by users. The options most likely to be modified by users are as follows:
Parser Option |
Description |
Default |
Good Default? |
–input, -i |
I3 file to process. Include multiple files by passing a comma-separated list |
None |
No |
–gcd, -g |
GCD file to use for processing. |
None |
No |
–output, -o |
Output file to write events to. |
None |
No |
–num, -n |
Number of events to process. Set to -1 to process all events. |
-1 (all events) |
Yes |
–hd5output |
HDF5 file to write event info. |
None |
Yes |
–rootoutput |
ROOT file to write event info. |
None |
Yes |
–gsiftp |
URL to use for FTP transfers. If given, this will be prepended to the gcdfile, infile, outfile, hd5output, and rootoutput paths. |
None |
Yes, if processing
files locally. If
using the grid,
set this to
|
There are also a few options used to custom photonics tables and splines used in reconstructions. These are generally set to the correct options and don’92t need to be modified unless the user does not have access to IceCube’92s cvmfs repository.
Parser Option |
Default |
–photonicsdir |
/cvmfs/icecube.opensciencegrid.org/data/photon-tables/SPICEMie/ |
–photonicsdriverdir |
/cvmfs/icecube.opensciencegrid.org/data/photon-tables/SPICEMie/driverfiles |
–photonicsdriverfile |
mu_photorec.list |
–infmuonampsplinepath |
/cvmfs/icecube.opensciencegrid.org/data/photon-tables/splines/InfBareMu_mie_abs_z20a10_V2.fits |
–infmuonprobsplinepath |
/cvmfs/icecube.opensciencegrid.org/data/photon-tables/splines/InfBareMu_mie_prob_z20a10_V2.fits |
–cascadeampsplinepath |
/cvmfs/icecube.opensciencegrid.org/data/photon-tables/splines/ems_mie_z20_a10.abs.fits |
–cascadeprobsplinepath |
/cvmfs/icecube.opensciencegrid.org/data/photon-tables/splines/ems_mie_z20_a10.prob.fits |
There is also one legacy option (-r or –restoretwformc), which was added to handle a bug present in some simulation sets where the SplitInIcePulsesTimeRange object was not written in the L2 processing. The bug is fixed in recent simulation sets, but is necessary for 2013+ simulation created before 2018. Enabling this option will add a TimeRange object into the frame if it is missing. Because the object is only added when missing, enabling this should not harm good events.
Cuts¶
The muon level3 cuts are primarily based around identifying good tracks using the HiveSplitter/IceHive suite. The cuts are as follows:
More detailed information about the processing and the individual scripts is given below.
Processing Stages¶
The primary entry script for the Muon L3 is python/level3_Master.py. This script passes all configuration options to MuonL3TraySegment.MuonL3, which runs the stages below.
level3_Functions.CleanInputStreams¶
https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_Functions.py#L78 - Reject events which do not pass either the MuonFilter or the EHEFilter L1 processing. Accept only the InIceSplit event stream.
MuonL3TraySegment.RestoreTimeWindow¶
https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/MuonL3TraySegment.py#L16
Re-runs trigger-splitter in order to recreate the SplitInIcePulsestimeRange object if missing. If the key is already present in the frame, do nothing.
level3_SplitHiveSplitter.SplitAndRecoHiveSplitter¶
Run the HiveSplitter module (Note: should be updated to use IceHive), which uses the geometry of the detector to efficiently split coincident events.
Remove afterpulses (which may have been split by HiveSplitter into a separate frame)
Attempt to identify and merge erroneously split frames (ie, frame where multiple sub-events were identified, but where only one true particle exists).
Remove any split frames produced solely from SLC hits.
Produce a set of time windows for the newly-split frames
Run basic L2 track reconstructions (2 LineFits, a single-iteration SPE, a 2-iteration SPE, and an MPE track fit) on the newly-split frames
level3_CalculateCutValues.CalculateCutValues¶
Calculate values using the CommonVariables project to get the number of total hits, direct hits, and some basic information about the best fit track https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_Functions.py#L7.
level3_Cuts.DoPrecuts¶
https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_Cuts.py#L88
Calculate the average charge-weighted distance to the track https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_Functions.py#L30 with and without DeepCore
Events with a total charge below 100 PE (ie, low energy events) or with an average charge-weighted distance above 90 PE*m (implying a poor fit) fail the precut
Events with a zenith angle less than 60 degrees pass if log10(qtot with DeepCore) >= 0.6(cosZenith-0.5)+2.6
Events with a zenith angle between [60, 78.5] degrees pass if log10(qtot with DeepCore) >= 3.9(cosZenith-0.5)+2.6
Events with a zenith angle larger than 78.5 degrees (ie, from the Northern hemisphere or horizon) pass the cut
level3_Cuts.DoLevel3Cuts¶
https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_Cuts.py#L43
Pass the event if at least 10,000 PE were observed, including hits from DeepCore.
Find the RlogL (reduced log-likelihood, logL/ndof) from the best fit. If the best-fit track is LineFit, set the RlogL to 1000. If RlogL < 9, pass the event.
Calculate a PlogL = RlogL*(number of hit DOMs-5)/(number of hit DOMs-3) for events with more than 5 hit DOMs. Set PlogL to 1000 for events with fewer than 5 hit DOMs. If PlogL < 7.5, pass the event.
Define a “direct hit ellipse” using the track length calculated from direct hits (LDir) and the number of direct DOMs (NDir), ellipse=(LDir/180)**2 + (NDir//10)**2. If the value is greater than 2 and at least 6 direct hits are observed, pass the event.
If an event passes at least one of the previous three conditions, and has a reconstructed zenith of at least 85 degrees (ie, Northern hemisphere), the event satisfies the L3 processing.
If an event passes at least one of the previous three conditions, has a reconstructed zenith of less than 85 degrees (Southern hemisphere + horizon), and the event satisfies a 2d cut in zenith and log10(total charge), the event satisfies the L3 processing.
level3_Reconstruct.DoReconstructions¶
https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_Reconstruct.py#L268
Rerun the L2 reconstructions using the new event splits, including MuEx, Cramer-Rao, and Paraboloid
Run Cramer-Rao and Paraboloid as angular error estimators using the the first good fit from [“MPEFit”,”SPEFit2”,”SPEFitSingle”,”LineFit”].
Run SplineMPE over the events
Run MuEX, TruncatedEnergy and Millipede over the events to obtain energy estimates
Test for IceTop hits using the shield module using several separate estimated in-ice reconstructed directions. Note that we are not cutting on these variables at this point!
level3_WriteOutput.WriteOutput¶
https://github.com/icecube/icetray/blob/main/level3-filter-muon/python/level3_WriteOutput.py#L15
Write basic MC information (like information about the primary neutrino and muon) to the frame, if available
Also run the final best-tracks through CommonVariables for use in later event selections
Write out information to i3, hdf5, and/or root files
To Do:¶
Add some figures to explain this more visually. Bullet points are really kind of ugly and the detail here doesn’t convey the steps well for new users at a glance.
Show a usage example (ie, what does a typical user pass to the command line?)
Add some python unit tests where reasonable. This’ll probably just be running and testing a few values using the full L3, since this is a (nearly) pure python script module
Add a verification file to I3_TESTDATA
Check whether the C++ modules are actually used anywhere. If they are not, then consider deleting them.
