New multilepton searches HLT2 lines
These are new lines for multi lepton searches. It mimics the dielectron and dimuon dark photon lines, but persists also information on other leptons in the events (muons and electrons). The goal is to search for dark photons with a lepton companion. Sorry if this come very late, I prepared it after recent discussions with theoreticians. It will require some tuning in the future, so this is just a first attempt. If there is no time for the review before the deadline do not worry, we can do it at the next iteration.
BW estimation
Just turbo lines are added, with leptons and muons tables as extra_output. Using https://gitlab.cern.ch/-/snippets/3438 for test.
Without changes Run over 10000 min bias events Turbo file-size = 1.91 MB Turbo events fired = 332 Turbo rate = 1290 kHz * 332/10000 = 42.8 kHz Turbo bandwidth = 1290 kHz * 1.91 MB/10000 = 246 MB/s
With changes Run over 10000 min bias events Turbo file-size = 1.92 MB Turbo events fired = 333 Turbo rate = 1290 kHz * 333/10000 = 42.8 kHz Turbo bandwidth = 1290 kHz * 1.92 MB/10000 = 248 MB/s
Conclusions
The change in the turbo rate is negligible, while the turbo BW increases of 2 MB/s
Merge request reports
Activity
assigned to @lsestini
requested review from @ahabdelm
requested review from @dzuliani
requested review from @rjhunter
mentioned in merge request !4367
119 PIDe_min=0.0, 120 pt_e=e_minpt, 121 maxipchi2=2.0, 122 trghostprob=0.25, 123 dielectron_ID="J/psi(1S)", 124 pt_diE=0 * MeV, 125 m_diE_min=5 * MeV, 126 m_diE_max=300 * MeV, 127 adocachi2cut=30, 128 bpvvdchi2=None, 129 vfaspfchi2ndof=10, 130 ) 131 132 133 @configurable 134 def dielectron_maker_displaced(e_minpt, isOS): -
I think this has been already defined here, so I guess it's no need to redefine it here (and this happens for several other functions)
100 F.CHI2DOF < vchi2pdof_max, F.OWNPVFDCHI2 > ownpvvdchi2_min 101 ) 102 if isOS: 103 DecayDescriptor = "J/psi(1S) -> mu+ mu-" 104 else: 105 DecayDescriptor = "[J/psi(1S) -> mu+ mu+]cc" 106 return ParticleCombiner( 107 Inputs=[filtered_muons, filtered_muons], 108 DecayDescriptor=DecayDescriptor, 109 CombinationCut=combination_code, 110 CompositeCut=vertex_code, 111 ) 112 113 114 @configurable 115 def dielectron_maker_prompt(e_minpt, isOS): -
see comment below
24 from RecoConf.algorithms_thor import ParticleCombiner, ParticleFilter 25 from RecoConf.event_filters import require_pvs 26 from RecoConf.reconstruction_objects import make_pvs, upfront_reconstruction 27 from RecoConf.standard_particles import ( 28 make_detached_dielectron_with_brem, 29 make_ismuon_long_muon, 30 make_long_electrons_no_brem, 31 make_long_electrons_with_brem, 32 make_prompt_dielectron_with_brem, 33 ) 34 35 full_lines = {} 36 turbo_lines = {} 37 38 # Mass limits now accepts all possible dilepton+lepton combination. May be reduced in the future. 39 comb_mass_limits = (0.0 * GeV, 14000.0 * GeV) 
Hi @lsestini , this is very interesting! Never thought about this. Since it mimics the dark photon lines (I didn't check the cuts one by one), I am thinking if it would be practical to just persist leptons in the existing dark photon lines? @acasaisv @kaaricha @fevolle @johndan
Edited by Andrii UsachovAfter thinking about it: as first attempt it could be ok to add these muon/electron containers to the existing dark photons lines. At least for now I am not going to differentiate them, so if necessary it can be done in the future. @fevolle could you please do it before the deadline (if it does not require too much time)? Thanks!
Hi @lsestini , due to various commitments, I won't have the time to implement and test it before tomorrow at lunch time. But I can give you some hints of how to do it : you can just implement the electrons and photons as
extra_output, as you do it already in this merge request and fill it in !4563 (diffs) . These are the new HLT2 lines, which take @acasaisv's new HLT1 developments into account (Allen!1890 (comment 9243990)). Please be aware that the lepton ID NN is not applied on the additionally saved leptons (so you might want to be a bit tighter in the lepton ID requirements). Furthermore, I wanted to flag up that the DP->ee HLT1 implementation only collects e+e- in the mass window of [5, 300]MeV. If you need a larger window, one would need different HLT1 lines as input.@lsestini for the
DiMuonNoIPlines, one could add a container to those lines !4539 (diffs) . I don't know which mass ranges you are particularly interested in ?
Hi @ausachov. It can be done but I would prefer to keep them separated: in fact in the future there may be different requirements for selecting the dimuon/dielectron resonance. There is a balance between the requirements on the resonance and those on the companion lepton that should be studied. At least for now I have put a hard cut on the companion lepton pT>5 GeV, for limiting the BW. But if we discover that we can reduce it, we could also release some of the cuts on the resonance. Consider the lines as they are now an experiment for studying the background (and hopefully for some preliminary limit expectation).
Edited by Lorenzo Sestini
Hey @lsestini, I glanced quickly at the code - please consider defining the lines in loops to avoid code duplication. See
drellyan.pyand theleplepmajorana lines in qee for inspiration on how to do this.