Add data-driven SPE distributions for each channel (fax) (#595)

* Add spe_distribution.csv which contains pdf for sample channel. Modify simulation.py to pull from this distribution for the gains rather than truncated gaussian

* Small fixes

* Add spe_distribution.csv to Simulation.py and update the way the file is loaded using the utils.py intstead of hardcoding

* make it work (almost)

* Modify pdf of spe distribution so that the mean is equal to one.

* Small changes, think it works now?

* Add a space to comment in simulation.py

* Add SPE distributions for each channel

* Add SPE distributions for each channel

* Small text changes in comments

* Explicitly set SPE distribution to 0 for off PMTs

* Change ch->channel in simulation.get_gains()

* fix style issue

* further style issue

pax/config/Simulation.ini

@@ -17,4 +17,5 @@ decoder_plugin = None
 pre_dsp = 'ZLE.SoftwareZLE'
 
 [WaveformSimulator]
-output_PI_truth_info = False
+output_PI_truth_info = False
+

pax/config/XENON1T.ini

@@ -219,11 +219,12 @@ anode_voltage =                       4.0 * kV           # xenon:xenon1t:aalbers
 pmt_rise_time             =           3.0 * ns           # Daniel Mayani with scope
 pmt_fall_time             =           10.0 * ns          # Daniel Mayani with scope
 
-# Data-extracted Single-pe area Distribution PDF (not single-photon area, double-pe emission is a separate parameter)
+# Data-extracted Single-pe area Distribution PDFs (not single-photon area, double-pe emission is a separate parameter)
 # This distribution should have only positive values, and have a mean value of 1.
+##  TODO currently these distributions can have negative numbers (e.g. underamplified SPE on downward noise fluctuation)
+## Will this break things?? -Evan
 # The gains drawn from this will be multiplied by the per-channel gains.
-# In the future we could try to implement an SPE distribution for each individual channel
-photon_area_distribution = 'spe_distribution.csv'
+photon_area_distribution = 'spe_distributions.csv'
 
 # Custom PMT model extracted from data. See [note to be written]
 # interp1d(pe_pulse_ts, pue_pulse_ys) must give a function mapping time since max in ns -> amplitude