Closed fjnicolas closed 3 months ago
As for validation, the current PDS MC paper draft reflects the current performance up to Reco1 using the efficiencies introduced by this PR. As for Reco2, since the CI will not really pick any possible drop in performaces due to this change, I made some validation plots for the relevant higher level variables that rely on PDS information (OpT0Finder, SimpleFlashMatch and CRUMBS). The plots compare current (black histograms) with the updated version including the mesh attenuation (red crosses)
CRUMBS score (neutrinos left, cosmics right):
OpT0Finder chi2 (neutrinos left, cosmics right):
OpT0Finder PEFractional (neutrinos left, cosmics right):
OpT0Finder PEMeasured (neutrinos left, cosmics right):
This one is obviously the one that is affected the most, since we are indeed reducing the amount of deteceted of photons by an overall scaling factor of 0.8. Since the plot above may be a bit hard to read, some numbers:
SimpleFlashMatch score (neutrinos left, cosmics right):
The ratio plots look sensible to me, so we are probably good to go without having to re-train the input weights for the tools. I'm asking the main developers of these tools to review this! @henrylay97 @lynnt20 @Robert-Darby
CI build and sequential tests can be found in this link. Expected changes in the product sizes for all the light related data products (generally a drop given the lower ScintPreScale and detection efficiency). The lower efficiency can be also seen in the OpT0 PE related variables:
473: rec.slc[0].opt0.hypoPE differs: 109979.6328125 vs 87983.703125
474: rec.slc[0].opt0.measPE differs: 123453.671875 vs 99095.1484375
Updating the Photon Detection Efficiencies after the new measurements for the X-ARAPUCAs (#412) we realized that the light attenuation due to the APA grounding mesh was not accounted for the PMTs. This component is not present in the GDML file for the SBND geometry (i.e. not accounted by the semi-analytic model corrections), so it needs to be included in the overal detection efficiency. This PR is to include its effect in the simulation of the PMT signals.
The mesh transparency is 85% [ref], but the light attenuation depends on the photons incident angle, as can be seen in the figure below
The distribution of incident angles for the two components VUV and visible are shown below
We see that the maximum of the distribution is at 42º for the VUV and at 37º for the visible light. A zoom of the attenuation factor around those values can be seen below
According to these results we will use in our simulations an attenuation factor of 0.8