This lesson is being piloted (Beta version)

Event Display and Particle Flow Exercise:

permalink: /aio/index.html —

Particle Flow event reconstruction in CMS

Overview

Teaching: 20 min
Exercises: 15 min
Questions
Objectives

Follow the slides!

Learn a little bit about how CMS converts detector readings into muons, electrons, photons, and hadrons by reading the slides from the exercise talk here!

Key Points

FireworksWeb Introduction

Overview

Teaching: 20 min
Exercises: 15 min
Questions

  • What is FireworksWeb?

  • How can FireworksWeb be used to analyze CMS event data?

  • What are the benefits of using FireworksWeb over traditional Fireworks?

Objectives

  • Understand the purpose and functionality of FireworksWeb.

  • Learn how to access and use FireworksWeb for CMS event data analysis.

The Fireworks team has created a web-based fireworks application. Fireworks is a longstanding CMS event display tool that interacts with MiniAOD files and allows the user to significantly customize the event display. The following 5 “explorations” can be done offline by anyone referencing this short exercise TWiki.

Introduction

Fireworks often requires (somewhat) cumbersome installation or heavy network usage over X11 forwarding, which creates some hurdle in quickly analyzing the events of interest. The FireworksWeb allows users to circumvent all of the installation and simply use web to directly open up a file such as /store/.../some/miniaod.root and directly start analyzing the event.

Please take a look at the CMS Offline and Computing week presentation from Alja for overview of the project.

Opening FireworksWeb

There are two instances of the website running currently.

The main features are the same. The first is the more general use case and accesses files stored on the CERN EOS space. The latter can access files from the CERN Open Data Portal.

Try Me!

Visit the https://fireworks-open.cern.ch/cms/ webpage, and click “View Fireworks” for one of the example samples. You will see some messages on the next screen as Fireworks loads the file, and then an event viewer will appear. Click and drag on the main image to manipulate it, and explore the various buttons and fields that you can see in the viewer.

Key Points

  • FireworksWeb is a web-based application for CMS event display.

  • It allows users to analyze CMS event data without cumbersome installations.

  • You can access files stored on CERN EOS or the CERN Open Data Portal depending on the FireworksWeb instance you use.

EW Physics

Overview

Teaching: 60 min
Exercises: 30 min
Questions
Objectives

Open http://fireworks.cern.ch and take a look at /store/group/upgrade/visualization/dy.root.

Once the path is provided click the Load File EOS button. Once it opens you should something like the following:

fireworksweb_5.png

To get a better sense of the event’s geometry, you can add the ECAL barrel to the event view in the following way:

fireworksweb_24.png

Before starting to play around, we can move to a more interesting event. You can scan through the events by clicking the event navigation buttons (right arrow), or enter the Run/Lumi/Event number directly. In this exercise, each file has only one Run, you just need to edit the lumi/event number to go to the new event.

Skip to the third event in the file, run/lumi/event is 1/416042/83208204.

fireworksweb_6.png

Try to click the interesting objects in the 3D view window and answer the below question:

Question 1

How many different kinds of objects are there? What are they?

Try to explore this event with different views with Fireworks. To swap the "main" panel with other view, click the button with the left-pointing arrow in the panel you wish to make your main one. You can look at other available views by clicking Views in the top-left and then the selecting the view you wish to add from the drop-down menu.

Question 2

Using the Table View on the bottom right, from the drop down menu below Choose Collection: choose Muons. How many additional Muons are not being displayed? (i.e. grayed out) Why are they hidden (Hint: press the "pencil-shaped" button in the left panel "Muons" section)?

Solution

In this configuration, the muon collection is filtered with expression “pt()>5 & isLooseMuon()”, meaning that each muon is only displayed if its pT is larger than 5 !GeV, and it passes the Loose ID requirements. fireworksweb_7

Question 3

Change the filter expression to display only muons with pT lower than 5 GeV (revert the change afterwards)

Solution

You should see this: fireworksweb_8

Question 4

Use the Table View in the bottom right corner to learn more details about the Muons collection. The table doesn't show dZ, to add it: in the table panel, go to the Muons collection, and click on the "Edit Collection Columns" button; then in the expression text box, write i.track()->dz(), give a title "dz", and define precision "3" and click "Add". Why does the 5.9 GeV muon stand out? Could you have seen this by zooming in on the Rho-Z View?

Solution

You should see this: fireworksweb_9

Question 5

Go to the 9th event in the file, run/lumi/event is 1/416042/83208226. What kind of event is it?

Solution

Z->ee event

Question 6

Uncheck the "Electrons" box to prevent them from being displayed: are there track pointing to both ECAL deposits?

Solution

Yes.

Question 7

Go to the next event, the 10th event in the file, run/lumi/event is 1/416042/83208227. What kind of event is it? What is the MET in this event? What is the transverse mass (mT) of the muon+MET+jet? What about for the muon+MET+LeadingTrack? This is MC, so you can peak at the GenParticles to check your hypothesis. You will have to add pdgId to the column to get a better understanding of the event. From the Table view once the PrunedGenParticles is selected, edit the table to add "i.pdgId()" expression with "pdgId" title with precision of 0.

NOTE: The legacy fireworks had kinematic variable calculator provided with the app. The feature is being worked on and will be added in the near future.

Show Answer

MET is et = 46.6 !GeV, phi = -0.414
Muon has pt = 11.0 !GeV, eta = 1.292, phi = -0.580
Jet has pt = 39.6 !GeV, eta = 1.453, phi = 2.865
Track has pt = 32.6 !GeV,eta = 1.456, phi = 2.869

mT(met, muon, jet) = 95.2 !GeV
mT(met, muon, track) = 86.4 !GeV

The event is likely a Z->tautau, with one tau decaying leptonically (muon + MET) and another in a one-prong hadronic tau decay. The one prong tau track is clustered in a jet with pT = 39.6 !GeV. When looking at the PrunedGenParticles list, one will find that there are pdgId 15 and -15 particle (grayed out) listed confirming that the event is indeed Z->tautau. The track pT is closest to the origina pion (pdgId=211) pT of 31.8 !GeV and also the eta. fireworksweb_10

Now let us use event filter by clicking the event filter button called FilterDialog. Search for events with large MET (i.e. >60 GeV). Select only the MET filter to be Active. Then click ApplyFilters button. Once the filter runs the FilterStatus: will report the number of events passing.

Question 8

How many events are selected? Can you guess what is the cause?

Show Answer

In total, six events are selected. All are Z-->TauTau events except events 83208351, 8365, 8506. Event 83208351 is a Z(ee)+jets event, the electrons are reconstructed properly (the invariant mass matches the Z). There is a below-threshold jet (pT=24.2 GeV) in the direction opposite to the MET, which could be mis-measured or coming from pile-up.

Event 83208365 is a Z(mm)+jets event with one muon out of acceptance (pt = 45 GeV, eta = 2.4). The lost muon explains most of the MET, while the rest can be ascribed to jet mis-measurement.

Event 83208506 is a Z(mm)+jets event with both muons out of acceptance and two very forward jets. Lost muons and mis-measured jets lead to the MET.

fireworksweb_14

Key Points

Top Physics

Overview

Teaching: 60 min
Exercises: 30 min
Questions
Objectives

The LHC has a very large ttbar production cross section, it is often called a top factory. Understanding ttbar events would be an important task for standard model measurement or searches for new physics. Now let us scan the ttbar file /store/group/upgrade/visualization/ttjets.root

Since events start to become complicated, a few hints before we start:

Hint 1: Top quarks decay into W+b, so you can look for b-tagged jets. To do this, create a table for the jet collection and add the b-tag discriminator method, i.bDiscriminator(\"pfCombinedInclusiveSecondaryVertexV2BJetTags\"). A "medium" b-tagging working point requires that the discriminator value is larger than 0.8. Alternatively, add a new Jet collection ("Add Collection") with a bDiscriminator filter applied.

NOTE: Please do not forget the escape character "\".

Show/Hide

fireworksweb_11

Hint 2 (MC only): use the PrunedGenParticles collection to make sense of the event: filter the collection with "!isHardProcess" to get the relevant particles (rather than the default filter).

First, go to the first event, 36/24518.

Question 9

What is this event? Can you find the combination of jets from Ws and from tops?

Show/Hide

This is a particularly difficult ttbar event, with one W decaying to leptons (muon+neutrino), and other to hadrons.

The transverse mass (mT) of the muon+MET is 98 GeV, very high for a W, presumably due to MET resolution. To make things worse, one of the quarks from the hadronic W is very low pT and does not form a jet. fireworksweb_12

Next, let's see event 2744/1876862

Question 10

What are the objects in the event? Are the muons close to jets?

Show Answer

This is a nice dilepton (e+mu) ttbar event. There are three muons, but two of them are close to jets. Looking at the truth record, we confirm that those two jets are coming from the b quarks. The truth record says that there are following two b-quarks:

pt = 87.4 !GeV, eta = -0.49, phi = 2.39,
pt = 52.6 !GeV, eta = 1.96, phi = 1.71

While the two jets are
pt = 66.5 !GeV, eta = -0.721, phi = 2.287
pt = 38.4 !GeV, eta = 2.012, phi = 1.622

The eta / phi directions show good agreement indicating that these indeed are the b-jets. fireworksweb_13

Key Points

Higgs Physics

Overview

Teaching: 60 min
Exercises: 30 min
Questions
Objectives

Exploration 3: Higgs Physics

Now let us scan the Higgs file /store/group/upgrade/visualization/ggh4l.root

First go to event 1278/254045.

Question 11

How many leptons in this event? What is this event? Can you calculate the invariant mass of the higgs? Are both Zs on-shell and if not which is the off-shell one?

Show/Hide

This event is H->ZZ->2e2mu. Invariant mass of the Higgs is 125.9 GeV, the off-shell Z is the one from the muon pair with mass 27.6 GeV (electrons make » an invariant mass of 92.4 GeV)

Go to event 1278/254054.

Use the muon table view to get the right charges. (i.e. add a column with i.track()->charge() or i.charge()) Let's add isolation column. The more proper PF-based isolation is currently not accessible. So we will use what is available. Add a column with (i.caloIso()+i.trackIso())/i.pt(). Once you do that you'll see something like this:

Result of adding charge and isolation fireworksweb_15

Also for isolation (i.userIsolation(\"PfChargedHadronIso\")+i.userIsolation(\"PfNeutralHadronIso\")+i.userIsolation(\"PfGammaIso\"))/i.pt() can be used as well.

Question 12

What is this event? Can you pair the muons and get the invariant mass of the two Zs? Are all 4 muons isolated (Isolation<0.2)?

Show/Hide

It is H->ZZ->4mu. Invariant mass of the Higgs is 126.9 GeV. From the table you can see the charge of the muons and try the combinations with opposite charges. Turns out that muons 0 and 2 are from an on-shell Z with mass 92.1 GeV, muons 1 and 4 from an off-shell Z with mass 27.0 GeV. All of the muons are isolated.

Go to event 1278/254110.

Question 14

What is this event? Are the electrons isolated? Can you pair electrons in the correct way? Why is the 4th electron not drawn?

Hint: Isolation can be accessed via (i.userIsolation(\"PfChargedHadronIso\")+i.userIsolation(\"PfNeutralHadronIso\")+i.userIsolation(\"PfGammaIso\"))/i.pt()

Show Answer

It is H->ZZ->4e. Invariant mass of the Higgs is 121 GeV. All electrons are fairly isolated, i.e. below 20%, except 1. From the electron table you can see the charge of the electrons and try the combinations with opposite charges. Based on the truth information the first and third electrons are the correct pair, and the 2nd and 4th, each with 23.5 !GeV and 61.9 !GeV respectively.

Key Points

13 TeV Collision Events

Overview

Teaching: 60 min
Exercises: 30 min
Questions
Objectives

In this section, we will look at collision events at 13TeV. These events are selected from SingleElectron and SingleMuon datasets. The events with two or more leptons are selected.

There are a few relevant files. Open the single electron data file, /eos/uscms/store/user/cmsdas/2020/short_exercises/Visualization/data_SingleElectron.root

Go to event 274200:48:72793234

Question 15

What kind of event is this? How can you justify your answer?

Show/Hide

This is a probably a Z->ee event, as can be understood from the invariant mass of the two electrons (90.647 GeV). fireworksweb_16

Go to event 274200:48:71915470

Question 16

What kind of event is this? How can you justify your answer?

Show Answer

The invariant mass of the two electrons (84.667 GeV) and that of the two jets (104.962 GeV). This could even be a ZZ->2e2q event, although this process is extremely rare with respect to the more frequent Z+2jets. fireworksweb_17

Go to event 274200:48:72636084

Question 17

What kind of event is this? How can you justify your answer?

Show Answer

This is probably a Z+jet event, as can be understood from the invariant mass of the two isolated electrons (90.312 GeV).

Hint: Isolation can be accessed via (i.userIsolation(\"PfChargedHadronIso\")+i.userIsolation(\"PfNeutralHadronIso\")+i.userIsolation(\"PfGammaIso\"))/i.pt() The 2nd electron with very high isolation value is likely part of a quark or gluon (i.e. ISR). fireworksweb_18

Often we find events that are not trivial to interpret, and usually there are at least a couple of processes which could have generated the event. If you got this far in the exercise you have enough experience to look at some more difficult events and try to guess what they come from. Typical questions to address are: how many AK4 jets are there in this event? Are they all isolated from leptons? Are they b-jets? Are there any electrons in the event with pt>30GeV? If so, check their isolation. Are there any muons in the event with pt>30GeV? If so, check their isolation. Is this a single lepton event? If so, what is the lepton? Can you identify any likely top candidates?

Question 18/19

Look at some interesting events in this dataset and try to interpret them: 1023/1523919869, 1023/1523757530.

You can also explore the single muon data file /eos/uscms/store/user/cmsdas/2020/short_exercises/Visualization/data_SingleMuon.root

You can go through the events and characterize each based on the objects you find.

If you have time at the end (after the next section), explore the events stored in

/eos/uscms/store/user/cmsdas/2020/short_exercises/Visualization/SinglePhoton*.root, and see if you can find anything interesting.

Key Points

Scanning Noisy Events

Overview

Teaching: min
Exercises: min
Questions
Objectives

In this section, we will have a look at a few events one does not want to end up with after having applied an analysis selection. These events can be of different natures:

Let's now have a look at some of those events:

Go to event 274316:385:698231955 of /store/group/upgrade/visualization/BadEvents_Run2016B1.root

Question 22

What is this event?

Hint

Take a look at the CSC-segments. Due to white background, the CSC view is a bit hard to catch. Change the color to see it a little better. If one hovers the mouse one can see click on them to highlight to make it more visible. Also, Ctrl- click will allow user to click multiple at a time.

Show/Hide

Beam halo – On the Rho-Z view, you observe a series of CSC hits forming a straight line. You can also notice that the HCAL deposit actually consists of several HCAL towers. If you now go to the 3D view, you can see that the HCAL towers lie on the same phi coordinate as the CSC hits. To get even more convinced of the phi correspondence of the calorimeter hits, you can also use the Lego view and zoom on to the area of interest. image19

Go to event 275074:259:417685155 of /store/group/upgrade/visualization/BadEvents_Run2016B2.root

Question 23

What is this event?

Show Answer

Beam halo – This is again a beam halo event. In this case, you actually have CSC hits on both sides of the detector (forward and backward). 20

Go to event 276244:428:658649950 of /store/group/upgrade/visualization/BadEvents_Run2016C1.root

Question 24

What is this event? Hint: The DT-Segments are not available in FireworksWeb for now. (Will be supported in the future.) So this one will be more easier with Legacy Fireworks. However, try adding collections (Click Add Collections -> Search "cosmic" -> Select "Muons", "muonsFromCosmics" -> AddCollection

Show/Hide

Cosmics – Looking at the Rho-Phi view, you see that some of the DT hits are forming a straight line. This can also be seen from the Rho-Z view. Going to the 3D view and playing a bit with the point of view, you can observe that the DT line coincides with the ECAL deposits. This is a muon hitting ECAL electronics to create a fake high energy ECAL deposit with no track: a photon (as can be seen from the photon collection).

!FireworksWeb with cosmic muon collection:

22 Legacy Fireworks screenshot: 21

Go to event 276437:1940:3426602195 of /store/group/upgrade/visualization/BadEvents_Run2016D1.root

Question 25

What is this event?

Show/Hide

Cosmics – The justification is the same as for the previous event. The straight DT line is quite obvious here. Notice that Fireworks tries to draw tracks coming from the center: it always assumes particles come from the center of the detector. 23

Go to event 276544:83:153985051 of /store/group/upgrade/visualization/BadEvents_Run2016D2.root

Question 26

What is this event?

Show/Hide

Beam halo – Another halo event.

Go to event 276585:73:117748568 of /store/group/upgrade/visualization/BadEvents_Run2016D3.root

Question 27

What is this event?

Show/Hide

Cosmics – Another cosmics event.

Go to event 276811:2460:3888980376 of /store/group/upgrade/visualization/BadEvents_Run2016D4.root

Question 28

What is this event?

Show/Hide

Cosmics – Another cosmics event.

Go to event 276582:638:1144780530 of /store/group/upgrade/visualization/BadEvents_Run2016D5_AOD.root

The fwc file defines all the parameter of the Fireworks display: the colors, the tables and views, the collections shown, etc. You can always change the parameters while looking at an event and then save your configuration in “File→Save Configuration”.

Question 29

Let's now have a look at this event. Can you spot any peculiar feature?
Hint : Look at the “Muons” table.

Show/Hide

What balances the large MET of 1.9 TeV is the Jets 0, which is made by a single charged hadron track. (i.e. single track line + single hcal/ecal tower).
On the other hand, you notice that a muon (Muon_0) passes right at the same place.
This muon has an insanely large track pT error (much larger than the track pT itself which is already at 2 TeV),
and is not identified by the particle flow algorithm as a muon.
Instead, its inner track is associated with the fake charged hadron we just described above.
This makes up a fake event where a charged hadron balances large MET.
Those are reconstruction issues that happen from time to time and that must be filtered out in order to keep the tail of the MET spectrum clean.

Key Points

iSpy Event Viewer

Overview

Teaching: min
Exercises: min
Questions
Objectives

The iSpy event display platform is used widely for CMS outreach and Open Data. It does not read MiniAOD (or any data tier) files directly, but converts them first into “.ig” files using an EDAnalyzer. This event display tool has stronger 3D capabilities (including 3D model exports) and more web-friendly aesthetics, but gives the user less control over the physics displayed than Fireworks.

Accessing iSpy

There are several instances of the iSpy viewer for different purposes. Some capabilities differ between the sites, but the main difference is which data files are pre-loaded for viewing.

Using the iSpy controls

The iSpy viewer has a panel of menu buttons across the top of the page, and a panel of Detector/Physics viewing options down the right-hand-side of the page. As you hover over each menu button a description will pop up.

In the Detector and Physics panel you can click on any element to open or close the drop-down menu. For each individual item you can control:

Calculating invariant mass

For educational purposes, mis a keyboard shortcut to display the invariant mass a selected object or pair of objects.

For example, in the Developer iSpy version, open the DoubleElectron_Run2012C_0.ig file and navigate to the 6th event (Events/Run_202016/Event_616963273). This is a nice Z boson candidate, decaying to 2 electrons.

To improve visibility of the electrons, open the controls for Tracks and click off the "Show" button. The two remaining tracks point directly to the ECAL hits and represent the electrons. Hover over one electron track and click when it turns grey. Hover over the other electron track and click when it turns grey. You should see both tracks turn grey. Strike the m key and the invariant mass will appear!

Producing an iSpy input file

Any CMS event can be processed to produce an iSpy input file.

  1. If you are working in NanoAOD, use the dasgoclient tools to find a RECO, AOD, or MiniAOD file the contains the event(s) you wish to study.
  2. Follow the instructions on the iSpy Analyzers Github to create a .ig file containing your events. *Read the entire README first* to understand how the EDAnalyzer functions and which cmsRun configuration file is best for your use case.
  3. Download the .ig file to your local computer and upload it into iSpy using the File menu.

3D and VR extras

3D file export and view

If you have a scene prepared in the display and want to export it to a 3D file press Shift+E. This will export the scene to a glTF (GL Transmission Format) binary file (with a .glb extension). You can view this file by dragging and dropping it to for example, this viewer https://cms3d.web.cern.ch/gltf-test.

VR test

This works best on a mobile device with a viewer but if you don't have a viewer you can get the idea by opening https://ispy-dev.web.cern.ch, on your phone, wait for it to load, and load an event. Then press the button that looks like a viewer. At the bottom of the scene (you may have to scroll down) you will see a “ENTER VR” button. Press it and you will see the stereo view. The direction of travel of the camera follows your gaze so that if you look at the event it will appear to be moving closer towards you.

Key Points