This code is part of the Liquid Argon Software (LArSoft) project. It contains simulation and reconstruction algorithms for LAr TPC detectors. If you have a problem, please log a redmine issue: https://cdcvs.fnal.gov/redmine/projects/larsoft/issues/new
0
stars
18
forks
source link
Bug in DepoFluxWriter when "incorrecting" the timing? #39
Unfortunately, the setup of ICARUS (and all other operating LArTPCs I am aware of) require "incorrecting" the time offsets to get SimChannel agreement with the output of the detector simulation and deconvolution. Following the directions in the nice documentation located here, resulted in nonsensical tdc times which don't appear to either be on the waveforms or anywhere close to pulses on those waveforms. In looking at the code I am fairly certain that the issue is at line 274 (here) in DepoFluxWriter.
If I modify this line to:
const unsigned int tdc = m_tbins.center(tbin)/m_tbins.binsize() + toffset_bin + m_tick_offsets[iplane];
Then I am able to get tdc values that are "in range" and agree with pulses seen on waveforms (with a caveat about differences in drift velocities leading to slight offsets between RawDigit peaks and SimChannel deposits).
How can we address this relatively quickly? The ICARUS ML effort really needs to be able to confirm that the output will allow them to continue with training of various networks in their code.
Unfortunately, the setup of ICARUS (and all other operating LArTPCs I am aware of) require "incorrecting" the time offsets to get SimChannel agreement with the output of the detector simulation and deconvolution. Following the directions in the nice documentation located here, resulted in nonsensical tdc times which don't appear to either be on the waveforms or anywhere close to pulses on those waveforms. In looking at the code I am fairly certain that the issue is at line 274 (here) in DepoFluxWriter.
If I modify this line to:
Then I am able to get tdc values that are "in range" and agree with pulses seen on waveforms (with a caveat about differences in drift velocities leading to slight offsets between RawDigit peaks and SimChannel deposits).
How can we address this relatively quickly? The ICARUS ML effort really needs to be able to confirm that the output will allow them to continue with training of various networks in their code.
Thanks!