Adjustments to missing pt
This MR implements several changes to missing pT:
- EE noise mitigation in 2017 is enabled (#48 (closed)). Implemented as discussed in #49 (closed) and this message.
- Propagation of JER smearing into missing pT can now be switched off in the master configuration. It's left on by default.
- In the standard implementation of the type 1 correction, jets with a high electromagnetic energy fraction are skipped and constituent muons are removed from jets (see #49 (closed)). Allow approximating this behaviour with a flag in the master configuration. Switched off by default.
- Consider in the type 1 correction also jets that fail jet ID, which agrees with the standard implementation.
I checked missing pT computed with the following configuration
ptmiss:
jer: false
pog_jets: true
in a few events in 2016. For this check, I also disabled the propagation of the electron and muon scale corrections into missing pT. Ideally, the result should be identical to the precomputed missing pT stored in branches MET_*
of NanoAOD. In reality their magnitudes typically agree within 0.1 GeV, although there are O(1) GeV differences in some events. There are several reasons that could explain these differences:
- Reduced numeric precision in NanoAOD.
- EM energy fractions are not stored in NanoAOD for soft jets, so the corresponding filtering cannot be applied to them.
- The muon subtraction is implemented with rescaling of the momentum of the jet, i.e. it doesn't change the jet's direction.
- The geometrical jet-lepton cleaning is not exactly the same. Overall, the observed differences are small enough to validate the type 1 correction.