WWZRun3 < Main

Main Web>TWikiUsers>KeeganDownham>WWZRun3 (18 Jan 2024, KeeganDownham) (raw view)Raw EditWysiwyg editAttachPrint versionPDF
WWZ Run 3 Analysis

%TOC%

---++ Meetings (weekly)

*Thursday, 7:00 am (CA) / 10:00 am (FL) / 4:00 pm (CERN) %BR%

Meetings are held in [[https://ufl.zoom.us/my/p.chang][Philip's Zoom room]]

Indico [[https://indico.cern.ch/event/1312765/][agenda]]

---

---++ Presentations

<br />%EDITTABLE{}%
| *Date* | *Keyword* | *Meeting* | *Who* | *Title* | *Summary* | *Link* | *Comments* |
| 5/22/23 | Tau ID vs Jets (follow up) | Weekly Meeting | Keegan | Tau ID vs jets (follow up) | Gen-matching study to estimate fake contribution in tau SRs | [[https://docs.google.com/presentation/d/1S6nvAvmY6Ukhoz2TH_DQtxA5X8JlD32yiAjN7PrEIGg/edit?usp=sharing][slides]] | ... |
| 5/16/23 | Tau ID vs Jets | None | Keegan | Tau ID vs jets | Yields comparison for DeepTau ID WPs vs jets | [[https://docs.google.com/presentation/d/1S6nvAvmY6Ukhoz2TH_DQtxA5X8JlD32yiAjN7PrEIGg/edit#slide=id.p][slides]] | ... |
| 5/15/23 | Tau ID vs Leps | Weekly Meeting | Keegan | Tau ID vs e/mu | Yields comparison for different DeepTau IDs vs light leptons | [[https://docs.google.com/presentation/d/10guWhn1UlVlRsqS-sMKv7w6r4G5ynJzNE0LDOznt5VM/edit#slide=id.p][slides]] | ... |
| 5/15/23 | Run3 DiMuon Analysis | Weekly Meeting | Matthew | Run3 DiMuon | MC and Data Comparison for DiMuon Triggered Events in Run3 era G | [[https://drive.google.com/file/d/16zt3SpKR7MtwEZ_qVGfqnsXSzoah_s4q/view?usp=share_link][slides]] | ... |
| 5/09/23 | Keegan's ATC Talk | None | Keegan | WWZTo4L Advancement Talk | Summary of the status of the WWZ-&gt;4L analysis | [[https://docs.google.com/presentation/d/17WQvBXTdEfVgWyTSTV9wn91aeym-aTsIKCl4CTiBia4/edit#slide=id.p][slides]] | ... |
| 5/3/23 | Tau Channels | None | Keegan | Tau Final States | Preliminary yields for tau final states and comparison with light lepton final states | [[https://docs.google.com/presentation/d/1Pe500kIsp57KEBHoQ7L9N0GsDxfGvGucD61XISJ3U2c/edit#slide=id.p][slides]] | ... |
| 3/13/23 | Lepton pT | Weekly Meeting | Keegan | Lepton pT optimization | Comparison of yields in SR for different lepton pT thresholds | [[http://cern.ch/go/6CZp][slides]] | ... |
| 3/6/23 | Signal xsec | Weekly Meeting | Keegan | Signal xsec | Comparison of cross sections for inclusive vs 4l sample | [[http://cern.ch/go/rcH6][slides]] | ... |
| 2/13/22 | Lepton ID Study | Weekly Meeting | Keegan | Lepton ID Study | Comparison of yields for new vs old lepton IDs | [[http://cern.ch/go/8QmM][slides]] | ... |
| 12/15/22 | pfMET vs PuppiMET | None | Keegan | Signal Efficiency for pfMET and PuppiMET | Determined cut on PuppiMET for optimal signal efficiency | [[http://cern.ch/go/9pLN][slides]] | ... |
| 12/08/22 | pfMET vs PuppiMET (cont.) | Weekly Meeting | Keegan | Background MET distribution | Compared data/background yields for pfMET vs PuppiMET | [[http://cern.ch/go/f66z][slides]] | ... |
| 12/1/2022 | Lepton ID | Weekly Meeting | Matthew | Lepton ID Study | made Loose/Z/W Candidate Efficiency plots for e/µ in pt v. eta | [[http://cern.ch/go/b6Rq][slides]] | ... |
| 12/1/2022 | pfMET vs PuppiMET | Weekly Meeting | Keegan | MET in ZZ CR | Compared data/ZZ background for PuppiMET vs pfMET in ZZ control region | [[http://cern.ch/go/FCX6][slides]] | ... |
| 11/17/2022 | ZZ Background | Weekly Meeting | Keegan | ZZ Background Yields | Compared ZZ background yields in ZZ CR between NanoAOD and MiniAOD | [[http://cern.ch/go/Lhf8][slides]] | ... |
| 11/10/22 | Signal Yields | Weekly Meeting | Keegan | Signal Yields | Comparison of signal yield between NanoAOD and MiniAOD | [[http://cern.ch/go/PB6F][slides]] | ... |

---++ People

Claudio Campagnari (UCSB)%BR% Philip Chang (UF)%BR% Keegan Downham (UCSB)%BR% Matthew Dittrich (UF)%BR%
---

---++ Data and MC Samples

A google spreadsheet containing the list of short names for all samples used in the analysis can be found [[https://docs.google.com/spreadsheets/d/1fTy4CwRkBf9XlJiudEtcUwJPKTbZuhKpFLmQjbb7TnE/edit#gid=0][here]]

The full list of samples (will full sample names) can be found [[https://github.com/cmstas/VVVNanoLooper/blob/698c634d97ae7aaf4febb6ce37291b2a74e2230d/analysis/run_analyzer.py][here]].
---

---++ Skimming

There will be 2 sets of skims for 2016-2018 data/MC: an inclusive 4-lepton skim and a 3-lepton skim. For a first skim, the selected leptons will satisfy the following criteria
   * *Muons* 
      * looseId = 1
      * |&eta;| &lt; 2.4
      * p<sub>T</sub> &gt; 10 GeV
      * pfIsoId &gt;= 1
   * *Electrons* 
      * mvaFall17V2noIso_WPL = 1
      * |&eta;| &lt; 2.5
      * p<sub>T</sub> &gt; 10 GeV
      * pfRelIso03_all &lt; 0.4

Skims for signal and background MC can be found in Keegan's ceph area on UAF: */ceph/cms/store/user/kdownham/skimOutput/WWZ_4L/*

A new set of skims, made for the purposes of testing the new lepton ID (see pre-selection for details), has been made. This is an inclusive 4-lepton selection that requires
   * *Muons* 
      * miniPFRelIso_all &lt; 0.4
      * mediumId = true
      * |&eta;| &lt; 2.4
      * p<sub>T</sub> &gt; 10 GeV
   * *Electrons* 
      * miniPFRelIso_all &lt; 0.4
      * |&eta;| &lt; 2.5
      * p<sub>T</sub> &gt; 10 GeV

Skims for testing the new lepton ID can be found in Keegan's ceph area on UAF: */ceph/cms/store/user/kdownham/skimOutput/WWZ_newLepID/*

---++++ Most recent version

The most up-to-date version of the skims is an inclusive 3-lepton skim that is used for both the 3-lepton+tau and 4-lepton analyses. It can be found in Keegan's storage area on ceph in the following directory:

*/ceph/cms/store/user/kdownham/skimOutput/3LepTau_4Lep/*

This set of skims is compatible with the new lepton ID (TOP-UL MVA ID) and utilizes the following selection criteria for leptons:

   * 3 or more leptons (electrons or muons) satisfying
---

---++ Pre-Selection (Old Lepton ID)

This analysis targets the 4-lepton final state. The pre-selection criteria is as follows:
   * *Preselection*: exactly 4 leptons satisfying the Common veto ID (see section "Skimming" for the definition)
   * *Lepton IDs*: Leptons are categorized as a Z-candidate or a W-candidate. The criteria for electron and muon IDs for W and Z candidates are 
      * *Electrons* 
         * *Z candidate (ZID)* 
            * ID: MVA POG Loose NoIso
            * |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; 4
            * I<sub>rel,R=0.3,EA,Lep</sub> &lt; 0.2 (Not used)
         * *W candidate (WID)* 
            * ID: MVA 90% Iso
            * |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; 4
            * I<sub>rel,R=0.3,EA,Lep</sub> &lt; 0.2 (Not used)
      * *Muons* 
         * *Z candidate (ZID)* 
            * ID: Medium
            * |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; 4
            * I<sub>rel,R=0.4,&Delta;&beta;</sub> &lt; 0.25
         * *W candidate (WID)* 
            * ID: Medium
            * |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; 4
            * I<sub>rel,R=0.4,&Delta;&beta;</sub> &lt; 0.15
   * *Z Candidate Selection*: The pair of leptons passing the ZID requirement with same flavor, opposite charge, that have an invariant mass closest to the Z boson mass. Both leptons are also subject to the following p<sub>T</sub> and m<sub>ll</sub> requirements: 
      * leading Z candidate lepton: p<sub>T</sub> &gt; 25 GeV
      * subleading Z candidate lepton: p<sub>T</sub> &gt; 15 GeV
      * |m<sub>ll</sub> - m<sub>Z</sub>| &lt; 10 GeV
   * *W Candidate Selection*: The remaining two leptons (the non-Z candidates) passing the WID requirements are then chosen as the W lepton candidates. We require the W lepton candidates to have opposite charge. Both leptons are subject to the following p<sub>T</sub> requirements: 
      * leading W candidate lepton: p<sub>T</sub> &gt; 25 GeV
      * subleading W candidate lepton: p<sub>T</sub> &gt; 15 GeV
   * *QCD low mass resonance veto*: Any opposite charge pair of leptons must have an invariant mass &gt; 12 GeV
   * *b-tagged jet veto*: selected events must have no b-tagged jets
---
---++ Pre-Selection (New Lepton ID)

This analysis targets the 4-lepton final state. The selection for leptons is as follows:
   * *Preselection*: exactly 4 leptons satisfying the following criteria 
      * *Muons*: 
         * p<sub>T </sub> &gt; 10 GeV
         * |&eta;| &lt; 2.4
         * |d<sub>xy</sub>| &lt; 0.05 cm
         * |d<sub>z</sub>| &lt; 0.1 cm
         * |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; = 8
         * miniPFRelIso_all &lt; 0.4
         * mediumId = true
         * MVA score &gt; 0.64
      * *Electrons*: 
         * p<sub>T</sub> &gt; 10 GeV
         * |&eta;| &lt; 2.5
         * |d<sub>xy</sub>| &lt; 0.05 cm
         * |d<sub>z</sub>| &lt; 0.1 cm
         * |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; = 8
         * miniPFRelIso_all &lt; 0.4
         * lostHits &lt; = 1
         * convVeto = true
         * tightCharge = 2
         * MVA score &gt; 0.81
   * <strong>Lepton p<sub>T </sub>: </strong>We require the leading (subleading) lepton in the event to have p<sub>T</sub> &gt; 25 (15) GeV. The third and fourth leptons in the event are both required to have p<sub>T</sub> &gt; 10 GeV.<sub style="font-weight: bold;"> </sub>
   * *Z Candidate Selection*: The pair of leptons passing the above Preselection requirement with same flavor, opposite charge, that have an invariant mass closest to the Z boson mass. Both leptons are also subject to the following requirements: 
      * |m<sub>ll</sub> - m<sub>Z</sub>| &lt; 10 GeV
      * both leptons: |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; 4
      * for electrons: relIso03_all &lt; 0.2
   * *W Candidate Selection*: The remaining two leptons (the non-Z candidates) passing the above Preselection requirements are then chosen as the W lepton candidates. We require the W lepton candidates to have opposite charge. Both leptons are subject to the following additional requirements: 
      * both leptons: |IP<sub>3D</sub>/&sigma;<sub>IP<sub>3D</sub></sub>| &lt; 4
      * for electrons: relIso03_all &lt; 0.2
   * *QCD low mass resonance veto*: Any opposite charge pair of leptons must have an invariant mass &gt; 12 GeV
   * *b-tagged jet veto*: selected events must have no b-tagged jets 
      * b-tagged jets: jets passing loose WP of btagDeepB algorithm
---++ Event Categorization

Events passing the pre-selection are categorized according to the flavors of the W candidate leptons. Two categories are defined: the opposite flavor category (e&mu;) and the same flavor category (ee/&mu;&mu;). Below is a brief description of the dominant backgrounds in each category and the selections made to address said backgrounds:
   * *e&mu;:* The main backgrounds in this category are ZZ--&gt; &tau;&tau; --&gt; e&mu; (for m<sub>ll,Wcands</sub> &lt; 100 GeV) and ttZ (for m<sub>ll,Wcands</sub> &gt; 100 GeV). The ZZ background is reduced heavily by cutting on the m<sub>T2</sub> variable, while the ttZ background is largely unaffected by cutting on m<sub>T2</sub> due to the similar kinematics as the signal. Because of this, events with m<sub>ll,Wcands</sub> &lt; 100 GeV are required to have m<sub>T2</sub> &gt; 25 GeV, while events with m<sub>ll,Wcands</sub> &gt; 100 GeV are not subject to any cut on m<sub>T2</sub>.
   * *ee/&mu;&mu;:* The main background in this category is ZZ--&gt; 4l. No additional selection is applied to this category prior to binning (see sec. "Binning").
---

---++ Signal Regions

Events in the e&mu; (ee/&mu;&mu;) category are further separated into four (three) signal regions (SRs) based on m<sub>ll,Wcands</sub> (p<sub>T,miss</sub>, p<sub>T,4l</sub>). The signal regions for both categories are given below:
   * *ee/&mu;&mu;:* 
      * Bin A: p<sub>T,miss</sub> &gt; 120 GeV
      * Bin B: 70 GeV &lt; p<sub>T,miss</sub> &lt; 120 GeV, 40 GeV &lt; p<sub>T,4l</sub> &lt; 70 GeV
      * Bin C: 70 GeV &lt; p<sub>T,miss</sub> &lt; 120 GeV, 70 GeV &lt; p<sub>T,4l</sub>
   * *e&mu;:* 
      * Bin 1: 0 GeV &lt; m<sub>ll,Wcands</sub> &lt; 40 GeV
      * Bin 2: 40 GeV &lt; m<sub>ll,Wcands</sub> &lt; 60 GeV
      * Bin 3: 60 GeV &lt; m<sub>ll,Wcands</sub> &lt; 100 GeV
      * Bin 4: 100 GeV &lt; m<sub>ll,Wcands</sub>
---

---++ Preliminary (Baseline) Results
---++++ Cut Based Analysis (4-lepton channel)

All plots for the baseline selection are stored [[http://uaf-10.t2.ucsd.edu/~kdownham/WWZ/070623_masterSync/Run2/][here]] on Keegan's public UAF area

The vvvtrees used to get these numbers are in the following area of Keegan's UAF directory:

*/ceph/cms/store/user/kdownham/VVVAnalysis/070623_masterSync/*

The baseline yields in the 4-lepton signal regions are as follows:
| SR Bins | NonResonant WWZ (signal) | ZH (signal) | &sum;bkgds | ZZ | ttZ | Higgs | WZ | Other |
| 1 | 0.694 ± 0.005 | 3.234 ± 0.012 | 1.528 ± 0.188 | 0.894 ± 0.023 | 0.339 ± 0.031 | 0.181 ± 0.168 | 0.107 ± 0.076 | 0.007 ± 0.007 |
| 2 | 0.796 ± 0.005 | 1.437 ± 0.008 | 1.787 ± 0.198 | 1.083 ± 0.026 | 0.442 ± 0.034 | 0.237 ± 0.193 | 0 ± 0 | 0.024 ± 0.009 |
| 3 | 1.574 ± 0.007 | 0.379 ± 0.004 | 1.753 ± 0.224 | 0.668 ± 0.020 | 0.864 ± 0.051 | -0.059 ± 0.164 | 0.176 ± 0.133 | 0.103 ± 0.052 |
| 4 | 5.310 ± 0.013 | 0.159 ± 0.004 | 3.949 ± 0.261 | 0.550 ± 0.018 | 2.586 ± 0.090 | 0.475 ± 0.167 | 0.240 ± 0.170 | 0.100 ± 0.052 |
| 5 | 2.513 ± 0.009 | 1.078 ± 0.009 | 3.833 ± 0.244 | 1.595 ± 0.031 | 1.486 ± 0.066 | 0.701 ± 0.230 | 0 ± 0 | 0.052 ± 0.038 |
| 6 | 1.836 ± 0.008 | 1.417 ± 0.010 | 6.386 ± 0.277 | 5.050 ± 0.054 | 1.032 ± 0.057 | 0.294 ± 0.265 | 0 ± 0 | 0.011 ± 0.009 |
| 7 | 0.560 ± 0.004 | 0.690 ± 0.005 | 3.992 ± 0.150 | 3.716 ± 0.046 | 0.242 ± 0.029 | 0.024 ± 0.139 | 0 ± 0 | 0.009 ± 0.007 |

<strong>Combined Significance: </strong>using signal+background MC in the 4-lepton SRs, we obtain a combined significance of <strong>4.46&sigma; </strong>(via Higgs Combine Tool)

---

---++ BDT Analysis
---+++ BDT Training Information:

To provide better rejection against the major backgrounds (ZZ and ttZ), a total of 4 different BDTs are trained (2 per signal process, and 2 per channel). Enumerating this, we have:

   * WWZ vs Backgrounds (ZZ,ttZ) in the opposite-flavor channel
   * ZH vs Backgrounds (ZZ,ttZ) in the opposite-flavor channel
   * WWZ vs Backgrounds (ZZ,ttZ) in the same-flavor channel
   * ZH vs Backgrounds (ZZ,ttZ) in the same-flavor channel

For the first version of the BDT (v1), the important information is listed below:

   * List of input variables that were trained on is [[http://uaf-10.t2.ucsd.edu/~kdownham/WWZ/BDT_Information/v1/inputs/Input_Vars_List_v1.pdf][here]]
   * xml files containing the weights for the [[http://uaf-10.t2.ucsd.edu/~kdownham/WWZ/BDT_Information/v1/weights/Opposite_Flavor/][opposite flavor]] and [[http://uaf-10.t2.ucsd.edu/~kdownham/WWZ/BDT_Information/v1/weights/Same_Flavor/][same flavor]] BDTs
   * Plots will be shown [[http://uaf-10.t2.ucsd.edu/~kdownham/WWZ/BDT_Information/v1/plots/][here]] (WIP)

---+++ Training Region (TR) Selection

Events passed to the BDT for training and testing are subject to the so-called "training region" selection, which is a subset of the full cut-based selection. A slide showing the training region selection can be found below:

   * Training region selection (v1): [[http://uaf-10.t2.ucsd.edu/~kdownham/WWZ/BDT_Information/v1/presentations/WWZ_BDT_Summary_092623.pdf][here]] (slide 3)
---
Topic revision: r47 - 18 Jan 2024 - 03:44:28 - KeeganDownham
Main
Webs
Main
Sandbox
Smurf
TWiki
Trash
TWiki Tip of the Day
Disabling individual WikiWords
Prevent a WikiWord from being linked by prefixing it with an exclamation point. Example: To escape SunOs ... Read on
Copyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback