darrylmasson
commented
7 years ago A couple of things:
- After #603 was merged, I ran another half million waveforms through fax and saw how things differ. The kink seen in Fig 5b in my note that starts at -8cm is gone so the overall behavior is now much more similar to Fig 5a. However, the median lines still don't match. Fax still tends to have a more extreme trend (lower area_fraction_top at the cathode than data, and higher at the gate), and this hurts the acceptance.
- The map used in the S1AreaFractionTopProbability plugin shouldn't need to be changed - it's data-driven so shouldn't be subject to these issues in the simulation.
- The notch at 7 PE is from events with area_fraction_top = 0 in regions where that doesn't happen much in data. These are mainly events in the center of the population in Fig 5b that are left of the red line - these events have a systematically lower area_fraction_top than as seen in data. I'm not exactly sure on the very root cause, but after #603 the depth of the notch has decreased slightly (1% improvement or so).
- I am not confident in the performance of this cut at high energies. Once saturation becomes significant, if we can't desaturate to better than 1% error, it just won't work well. At high energies the band in z-aft space is very narrow, and deviations from the trend are heavily disfavored. When I get around to it I intend to relist the cut in lax as a "low-energy" cut, or at least put a caveat in the docstring that the acceptance does decrease significantly above 1000 PE.
Overall #603 doesn't change much in the FV. I have a couple of ideas on how to use data to check the acceptance out to high energies (rather than fax), but haven't had the chance to work on them.
pdeperio
@darrylmasson Does @l-althueser's S1 pattern map update (https://github.com/XENON1T/pax/pull/603, note) answer one of the issues in your note (Figure 5b)? If so, do you need to update the S1 AFT map again? Then can you please update your note with this and the following: