ComsicAnalysis

ROOT-based analysis to study cosmic muons.

MC

A through-going cosmic muon sample is stored at /eos/experiment/wcte/MC_Production/v1.4.1/cosmics/. The production and analysis environment can be accessed by

/eos/experiment/wcte/MC_Production/ContainerImage/run_container.sh /eos/experiment/wcte/MC_Production/ContainerImage/softwarecontainer_v1.4.2.sif

mPMT efficiency difference

To mimic the effect of reduced PMT efficiency observed in through going beam muons, MDT is used to reprocess the MC data. An ad-hoc detection efficiency factor (relative to WCSim) of 45% is applied on WUT ex-situ mPMTs, and 90% otherwise. Waveform simulation and hit finding are also applied based on self-trigger conditions.

The processed files are saved in /eos/experiment/wcte/MC_Production/v1.4.1/cosmics/mdt_reduced_eff/.

CreateBadChannelList.c

To create bad channel list per run, the good_wcte_pmts list in the metadata and the actual run data are used.

root -l -b -q CreateBadChannelList.c\(1766\)

This creates the bad channel list for run 1766.

fiTQun reconstruction is run with the bad channel being masked, and the output is stored in /eos/experiment/wcte/MC_Production/v1.4.1/cosmics/mdt/fiTQun/.

AnalyzeCosmics.c

To select muons that enter through the top cap and exit through the bottom cap, a set of plots is created in fig/ to study cut values on number of PMT hits and charge ratios in different sections (top cap, barrel, bottom cap).

root -l -b -q AnalyzeCosmics.c\(false,false\)
root -l -b -q AnalyzeCosmics.c\(false,true\)
root -l -b -q AnalyzeCosmics.c\(true,false\)

Then we derive a set of selection cuts:

• number of PMT hits > 700
• Top/Total charge < 0.07
• 0.45 < Barrel/Total charge < 0.7
• 0.25 < Bottom/Total charge < 0.5

To further improve selection purity, the fiTQun reconstructed vertex and direction are extrapolated to find the muon entrance and exit positions at the tank surface. Then we only keep muons with:

• reconstructed entrance point on the top cap
• reconstructed exit point on the top cap

After the selection, 25% of all the muons remain. The selection efficiency and purity of top-down events are 68% and 79% respectively. Meanwhile over 95% of the selected events are with truth cos(zenith_angle)>0.7.

The selected muons are stored in /eos/experiment/wcte/MC_Production/v1.4.1/cosmics/mdt_reduced_eff/muon_MC_selected.root.

convert_to_h5.py

Convert wcsim and fiTQun root files into a single h5 for ML studies. Note that the units of length, time and momentum are mm, ns and MeV/c respectively. And there is a coordinate transformation of $(x,y,z)\rightarrow (x,-z,y)$ applied.

The h5 file contains the following arrays:

• vertex: truth vertex (nevents,3)=shape, outside the tank
• direction: truth direction (nevents,3)
• entrance_pos: truth entrance point (nevents,3)
• exit_pos: truth exit point (nevents,3)
• is_top_down: truth top-down flag (nevents)
• momentum: truth momentum at vertex (nevents)
• pmtQ: total charge per PMT (nevents,npmts)
• pmtT: first hit time per PMT (nevents,npmts)
• fq_pos: fiTQun reconstructed vertex and time (nevents,4)
• fq_entrance_pos: fiTQun reconstructed entrance point (nevents,3)
• fq_exit_pos: fiTQun reconstructed exit point (nevents,3)
• fq_direction: fiTQun reconstructed direction (nevents,3)
• fq_momentum: fiTQun reconstructed momentum (nevents)
• pass_selection: pass selection flag (nevents)

To load the arrays,

import h5py
import torch

with h5py.File('wcte_cosmics_mc.h5','r') as f:
    entrance_pos = torch.as_tensor(f['entrance_pos'][:]) # shape=[nevents,3]
    exit_pos = torch.as_tensor(f['exit_pos'][:]) # [nevents,3]
    is_top_down = torch.as_tensor(f['is_top_down'][:]) # [nevents]
    pmtQ = torch.as_tensor(f['pmtQ'][:]) # [nevents,npmts]
    # etc.

Data

SelectCosmicCandidate.c

To search for cosmic muons in data, a 50 ns moving time window is used with a number of hit threshold (N50) equal to 700. Then the same charge ratio cuts (top/barrel/bottom over total charge) are applied. After that, the candidate event PMT hit information is written in a WCSim-like format for fiTQun processing.

In run 1766, which is a background run in pure water phase, there are 442k events with each event spanning 500 $\mu s$, giving a total live time of 211s. 80358 muon candidates are found.

The selected muon candidates and fiTQun outputs are stored in /eos/experiment/wcte/MC_Production/v1.4.1/cosmics/run1766/out_1766.root and fq_00*.root.

AnalyzeCosmics.c

After the fiTQun reconstructed entrance/exit point cuts are applied, 51093 muon candidates remain.

root -l -b -q AnalyzeCosmics.c\(true,false,false,\"/eos/experiment/wcte/MC_Production/v1.4.1/cosmics/run1766/out_1766.root\",\"/eos/experiment/wcte/MC_Production/v1.4.1/cosmics/run1766/fq_00*.root\",\"_1766\"\)

From MC studies, the fiTQun resolution on entrance and exit points are better than 20 cm. This is also cross-checked with beam crossing muon sample, which shows reasonable agreement between MC and data.

The selected muons are stored in /eos/experiment/wcte/MC_Production/v1.4.1/cosmics/run1766/muon_1766_selected.root.

FitAngularResponse.c

root -l -b -q FitAngularResponse.c 

For each selected muon, PMTs within the 42 deg Cherenkov cone from the entrance point are selected, and the time residual of each hit is calculated. To veto badly reconstructed muon tracks, events are rejected if more than 0.8% of charges are deposited with time residual < -2 ns. Then, only PMT hits with time residuals between -2 and +2 ns, and expected photon-PMT distance larger than 100 cm are selected. 2D histograms of charge*distance vs. costh are used to extract the angular response per mPMT (type). The extracted responses are compared with the ones from nominal MC to derive the efficiency correction curves per mPMT (type).