## Muon HLT Efficiency Measurement with the "Reference-Trigger" Method

### Goal

Very useful to measure the efficiency of complex trigger paths, such as double-muon and dimuon triggers, combination of triggers, cross-triggers, etc. The description and examples below refer to the specific cases of (combinations of) double-muon trigger paths HLT_Mu17_Mu8, HLT_Mu17_TkMu8, HLT_Mu22_TkMu8.

### Ingredients

#### Choice of the binning

The measurement can be done in one single bin on in several bin depending of the use. The bins are just to be large enough to not lack of statistic for the efficiency computation.

#### Choice of the reference trigger

The measurement is based on the computation of the complex trigger paths efficiency for event passing a reference trigger. The reference trigger should be chosen to have a high efficiency on the events passing the complex trigger to not bias the result. For example, HLT_Mu17 is the best candidate for double muon triggers like HLT_Mu17_Mu8, HLT_Mu17_TkMu8, HLT_Mu22_TkMu8.

#### Reference trigger efficiency

The reference trigger efficiency on muon can be measured using the standard Tag and Probe method as documented here. The reference trigger efficiency on di-muon event can be computed from the reference trigger efficiency on muon as follow : if is the efficiency of the ref trigger on muon, then the efficiency on di-muon, , will be computed using this formula :

#### Complex efficiency after reference trigger

This is the efficiency of the complex trigger path on di-muon events triggered by the ref trigger. This efficiency can be obtained by fitting the Z peak as in usual Tag and Probe. It the reference trigger is pre-scaled, it is need to re-weight each event to compensate the pre-scale.

#### Complex trigger efficiency

The final number is obtained bin by bin by multiplying the efficiency of the reference trigger by the efficiency of the complex trigger after reference trigger.

### Recipe

#### Recipe for HLT_Mu17_Mu8 and HLT_Mu17_TkMu8 paths

This recipe is for use the TnP trees centrally produced, which are documentedhere?
##### Reference trigger efficiency
• The chosen reference trigger chosen is HLT_Mu17. As it is prescaled, it is needed to request the tag muon to be match with HLT_Mu17 in order measure an efficiency without prescale.
• The TnP trees are considering the case where more than one probe can be use for a tag : then the recommended action is to use the probe that will make the pair mass closest from the Z mass, it consist to ask in the TnP tree BestZ=1

The efficiencies of the reference trigger for event passing the POG loose working point (Global OR Tracker muon AND PF muon) are here :

plots here

##### Complex trigger efficiency after reference trigger
• The complex trigger efficiency (or soup of trigger) after reference trigger (EffSoup|ref) is computed also using TnP for di-muon event triggered by HLT_Mu17 (corresponding to the cut (tag_Mu17==1 || Mu17==1).
• Two specific features are to be taken in account :
• There is an additional cut on the tag : PFchargedIsolation/pt < 0.2 + matched with a L3 object ( pfIsolationR04().sumChargedHadronPt/pt < 0.2 && triggerObjectMatchesByCollection('hltL3MuonCandidates')). This introduce a overestimation of about 0.3% to be taken in account in the systematic error.
• As the TnP are designed to compute efficiency by muon, a TnP pair can be double counted with the Tag and the Probe role inverted. As the soup efficiency is computed by pair, it is needed to remove double counting by choosing randomly one of the 2 pairs. An example macro for this can be found here.

##### Final trigger efficiency
• This step only consist in multiplying the ref trigger efficiency time the efficiency after reference trigger.

##### Bias from the method :
• Bias from the additional cut on the tag :
• Bias from the difference of L1 seed :
• Bias from the choice of the binning :

This topic: Main > TWikiUsers > HuguesBrun > TestForTrigger
Topic revision: r6 - 2013-09-02 - HuguesBrun

Copyright &© 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback