XuyangGaoSandbox < Sandbox

-- Main.XuyangGao - 2019-11-20

---+ Electron short exercise

We would introduce the measurements on the efficiency of trigger, HEEP ID with tag & probe method.
   * Use tag & probe method to measure trigger efficiency for HEEP electron.
   * Use tag & probe method to measure HEEP ID efficiency for GSF electron.
   * Try to optimize each detailed cut of HEEP ID criteria according to N-1 efficiency.

---+++ Get object information from CMS dataset using IIHENtuple

You can skip this step next step if you have your own code to get object information from CMS dataset.

<br />%TWISTY{ mode="div"  <br />showlink="Show the recipe"<br />hidelink="Hide the recipe"  <br />showimgleft="%ICONURLPATH{toggleopen-small}%"  <br />hideimgleft="%ICONURLPATH{toggleclose-small}%"}%

The IIHENtuple code is a powerful framework developped by IIHE institute

Set up CMSSW area
<blockquote>

cmsrel CMSSW_10_2_13<br />cd CMSSW_10_2_13/src<br />cmsenv<br />git cms-init<br />git cms-merge-topic cms-egamma:EgammaPostRecoTools<br />git cms-addpkg EgammaAnalysis/ElectronTools<br />rm EgammaAnalysis/ElectronTools/data -rf<br />git clone git@github.com:cms-data/egammaanalysis-electrontools.git EgammaAnalysis/ElectronTools/data<br />git cms-addpkg RecoMET/METFilters<br />git cms-merge-topic cms-met:METFixEE2017_949_v2_backport_to_102X

scram b -j 16
</blockquote>

Clone IIHENtuple code
<blockquote>

git clone git@github.com:rgoldouz/miniaod_iihe_2018.git -b Legacy_CMSSW_10_2_13 UserCode/IIHETree<br />scram b -j 16
</blockquote>

Submit to crab3

<br />%ENDTWISTY%

Since there are more than 16K files in the whole 2018 datasets (run ABCD), It will take a very long time to run on all of them.

We thus have provided the output root files in:
<blockquote>

link : zzzzzzzzzzzzzzzzz
</blockquote>

You can use them for next step directly.
---+++ Tag & probe method

We would give a short introduction about the Tag and Probe method, we will use it in the following measurements on trigger efficiency and ID efficiency.

---+++ Produce preliminary tag & probe pair

First of all, we will produce a preliminary tag & probe pair, the "preliminary" means that the cut on tag and probe are not fully applied, because we would like to use these pairs as base of more than one measurements, so some other cuts will be added in related steps.

Here we loop each event and ask the following requirements for tag:
<blockquote>

ET &gt; 35 GeV

In barrel region

Pass HEEP ID
</blockquote>

If there are more than one tags in the event, we will store all possible tag & probe pair.

Here we have provided a python script as an example.
<blockquote>

python xxx.py
</blockquote>

The output root file is also provided since this step will take a lot of time without queue:
<blockquote>

path : xxxxxxxxxxxxxxxxxxxxxxxxx
</blockquote>
---+++ Measure trigger efficiency:

We will measure single electron trigger efficiency for HEEP electron in this step.

First we will add some cuts based on preliminary T&P pairs:
<blockquote>

tag must be matched with Ele35_WPTight trigger

T&P must be in [60, 120] GeV
</blockquote>

These cuts can make sure that we are making unbias measurements on real electron.
<blockquote>

Trigger efficiency = number of probe match with trigger final filter and pass HEEP ID / number of probe pass HEEP ID
</blockquote>

We loop each pair of T&P, get the number of denominator and numerator, we can of course split the events into different bins to get trigger efficiency curve as functions of different values, such as eta, ET, phi, number of vertex et al.

Here is an example python script to draw the plot:
<blockquote>

python xxx.py
</blockquote>

For double electron trigger, such as DoubleEle25 _MW, it is a bit more complicated, there are two legs in this trigger, one is L1 seeded leg, the other is unseeded leg. So if we want to measure efficiency, we need to add cuts:
<blockquote>

tag must be matched with Ele35_WPTight trigger

tag must be matched with L1 filter of DoubleEle25

T&P must be in [60, 120] GeV
</blockquote>

And as we know, each trigger has many filters, for example, L1, ET, ISO, ID, they are in linear, it means the object must pass the filters one by one. If we want to measure the efficiency of one of that filter, we can just change the cut on probe:

For example:
<blockquote>

Trigger DoubleEle25 ID efficiency = #probe pass HEEP, match with unseeded ID filter / #probe pass HEEP, match with unseeded ET filter
</blockquote>
---+++ Measure HEEP ID efficiency:

We will measure the HEEP ID efficiency for GSF electron in this step.

First we will add some cuts based on preliminary T&P pairs:
<blockquote>

tag must be matched with Ele35_WPTight trigger

T&P must be in [70, 110] GeV
</blockquote>

These cuts can make sure that we are making unbias measurements on real electron.
<blockquote>

HEEP ID efficiency = #probe pass HEEP ID / #probe
</blockquote>

Sometimes, we would like to do some more detailed study for each cut, then we can measure N-1 efficiency (N means all cut):
<blockquote>

HEEP ID N-1 efficiency = #probe pass HEEP ID / #probe pass HEEP ID except for one cut
</blockquote>
Topic revision: r2 - 2019-11-20 - XuyangGao
Account
- Log In
- Cern Search
- TWiki Search
- Google Search
Sandbox All webs
Edit
Attach
Copyright &© 2008-2023 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback