Adding fragments for HH_WWZZ analysis #3822

asculac · 2024-12-18T15:50:32Z

This PR contains new fragments needed for HH_WWZZ analysis for both Run2 and Run3.
They are needed for a new diHiggs analysis which is considering HH decaying into 4V (2 W bosons and 2 Z bosons), non-resonant, and focusing on leptonic and semi-leptonic final states, with Run2 and Run3 data. You can see the latest analysis status reported in the last HZZ meeting [1]

For signal yield estimate we are using already existing “GluGlutoHHto4V” signal samples, which are including all 4V decays. This sample is appropriate for our analysis, as it contains events where HH decays to WWZZ. But, since our decay mode is 4 times more rare than the 4W case, this leaves us with a few % of the inclusive sample once sub-selecting WWZZ in 3l and 4l. Therefore we are in need of more statistic for our BDT model trainings.
Additionally, we are adding fragments for qq->WWZZ, as this is one of our backgrounds and the sample is not existing at the moment.

4 new fragments are added per Run (4 for Run2 + 4 for Run3):

HHtoWWZZ_3l (HH->WWZZ->exactly 3 leptons)
HHtoWWZZ_4lplus (HH->WWZZ->at least 4 leptons)
qqWWZZ_3l (qq->WWZZ->exactly 3 leptons)
qqWWZZ_4l_plus (qq->WWZZ->at least 4 leptons)

Details about the request and testing fragments are given in the notebook

lviliani

Related to the few comments inline, I also have some general questions:

What are the Powheg processes used for the gridpacks? Especially in the background case.
You are not including taus in the ResonanceDecayFilter, so you will not have leptonic tau decays that could be relevant in your analysis. Is this done on purpose?
Related to the comment above, I have some doubts about the 4lplus scenario. IIUC in this case you must have at least 2 leptonic Zs or 2 leptonic Ws and 1 leptonic Z. The remaining Ws or Zs can decay to hadrons or leptons, therefore also to taus, which on the other hand you are excluding in the filter. So the setup does not look fully consistent to me.

Thanks and sorry if I misunderstood something.

lviliani · 2025-01-09T12:29:46Z

genfragments/ThirteenPointSixTeV/qqWWZZ_3l_TuneCP5_13p6TeV_powheg-pythia8.py

+              '23:mMin = 0.05',
+              '24:mMin = 0.05',
+              '23:onMode = off', # disable all Z decay modes
+              '23:onIfAny = 1 2 3 4 5 6 7 8 11 13 15', # enable only Z->ll and Z->qq


Do you really need also 6, 7, and 8?

Oh not really, we just included all by default, can be removed if necessary

lviliani · 2025-01-09T12:31:37Z

genfragments/ThirteenPointSixTeV/qqWWZZ_3l_TuneCP5_13p6TeV_powheg-pythia8.py

+        pythia8PSweightsSettingsBlock,
+        pythia8PowhegEmissionVetoSettingsBlock,
+        processParameters = cms.vstring(
+          'POWHEG:nFinal = 4',   ## Number of final state particles


Which Powheg process are you using to generate the gridpack?

lviliani · 2025-01-09T12:39:49Z

genfragments/ThirteenPointSixTeV/qqWWZZ_3l_TuneCP5_13p6TeV_powheg-pythia8.py

+            'ResonanceDecayFilter:exclusive = on', #off: require at least the specified number of daughters, on: require exactly the specified number of daughters
+             'ResonanceDecayFilter:mothers = 24,23', #? TBC! list of mothers not specified -> count all particles in hard process+resonance decays 
+            'ResonanceDecayFilter:eMuAsEquivalent = on', #on: treat electrons and muons as equivalent
+            'ResonanceDecayFilter:daughters = 11,11,11',


taus are not included here, is it intentional?

asculac · 2025-01-10T10:25:21Z

Thank you for your followup, I hope I understood well your questions and you can find answers below:

Related to the few comments inline, I also have some general questions:

What are the Powheg processes used for the gridpacks? Especially in the background case.

For the HH case, it is a glugluToHH cental gridpack. For the qq_WWZZ It is a private gridpack, producing q q > w w z z

You are not including taus in the ResonanceDecayFilter, so you will not have leptonic tau decays that could be relevant in your analysis. Is this done on purpose?

We are not including taus in the ResonanceDecayFilter as we don't want our Higgs to decay to taus (we are targeting only HH_WWZZ), but we do allow to have them in the final state (coming from W or a Z). This you can see form the notebook linked in our PR, where we showcase the output of our production, see example for HH_WWZZ_3l:

Processed 999 event(s) from file /data_CMS/cms/asculac/MCProduction/OutputSamples/2024_27_04/ggHH_3l_part2/Step_4/Ntuple_1.root
HHWWZZ:0 ele 3 mu 4 Q 1 Nu 0 Tau -> 223 (22.3%)
HHWWZZ:1 ele 2 mu 4 Q 1 Nu 0 Tau -> 202 (20.2%)
HHWWZZ:3 ele 0 mu 4 Q 1 Nu 0 Tau -> 201 (20.1%)
HHWWZZ:2 ele 1 mu 4 Q 1 Nu 0 Tau -> 198 (19.8%)
HHWWZZ:0 ele 3 mu 2 Q 2 Nu 1 Tau -> 39 (3.9%)
HHWWZZ:2 ele 1 mu 2 Q 2 Nu 1 Tau -> 35 (3.5%)
HHWWZZ:3 ele 0 mu 2 Q 2 Nu 1 Tau -> 31 (3.1%)
HHWWZZ:1 ele 2 mu 2 Q 2 Nu 1 Tau -> 31 (3.1%)
HHWWZZ:1 ele 2 mu 2 Q 1 Nu 2 Tau -> 11 (1.1%)
HHWWZZ:3 ele 0 mu 2 Q 1 Nu 2 Tau -> 9 (0.9%)
HHWWZZ:2 ele 1 mu 2 Q 1 Nu 2 Tau -> 8 (0.8%)
HHWWZZ:0 ele 3 mu 2 Q 1 Nu 2 Tau -> 6 (0.6%)
HHWWZZ:3 ele 0 mu 0 Q 2 Nu 3 Tau -> 3 (0.3%)
HHWWZZ:1 ele 2 mu 0 Q 2 Nu 3 Tau -> 1 (0.1%)
HHWWZZ:0 ele 3 mu 0 Q 2 Nu 3 Tau -> 1 (0.1%)

Related to the comment above, I have some doubts about the 4lplus scenario. IIUC in this case you must have at least 2 leptonic Zs or 2 leptonic Ws and 1 leptonic Z. The remaining Ws or Zs can decay to hadrons or leptons, therefore also to taus, which on the other hand you are excluding in the filter. So the setup does not look fully consistent to me.

Again, similar answer as to the previous questions, we are allowing the taus to be there in the final state. Here is the 4plus output example for qq_WWZZ_4lplus:

Processed 500 event(s) from file 4l_plus.root
Total number of events in file: 500
Number of Higgs bosons: 0
Number of WWZZ from qq: 500
Number of WWZZ from HH: 0

:3 ele 1 mu 2 Q 2 Nu 0 Tau —> 68 (13.60%)
:1 ele 3 mu 2 Q 2 Nu 0 Tau —> 67 (13.40%)
12 ele 2 mu 4 Q 0 Nu 0 Tau -> 62 (12.40%)
:2 ele 2 mu 2 Q 2 Nu 0 Tau -> 52 (10.40%)
:4 ele 0 mu 2 Q 2 Nu 0 Tau —> 27 (5.40%)
:3 ele 2 mu 2 Q 1 Nu 0 Tau —> 26 (5.20%)
:0 ele 4 mu 2 Q 2 Nu 0 Tau —> 24 (4.80%)
:4 ele 0 mu 4 Q 0 Nu 0 Tau -> 22 (4.40%)
:2 ele 3 mu 2 Q 1 Nu 0 Tau —> 21 (4.20%)
:0 ele 4 mu 4 Q 0 Nu 0 Tau —> 19 (3.80%)
:0 ele 5 mu 2 Q 1 Nu 0 Tau —> 18 (3.60%)
:2 ele 2 mu 2 Q 1 Nu 1 Tau —> 17 (3.40%)
:1 ele 4 mu 2 Q 1 Nu 0 Tau —> 12 (2.40%)
:4 ele 1 mu 2 Q 1 Nu 0 Tau —> 11 (2.20%)
:5 ele 0 mu 2 Q 1 Nu 0 Tau —> 8 (1.60%)
:0 ele 4 mu 2 Q 1 Nu 1 Tau —> 7 (1.40%)
:3 ele 3 mu 0 Q 2 Nu 0 Tau —> 5 (1.00%)
:0 ele 5 mu 0 Q 2 Nu 1 Tau —> 5 (1.00%)
:2 ele 2 mu 0 Q 2 Nu 2 Tau —> 5 (1.00%)
:4 ele 0 mu 2 Q 1 Nu 1 Tau —> 3 (0.60%)
:3 ele 2 mu 0 Q 2 Nu 1 Tau —> 3 (0.60%)
:2 ele 3 mu 0 Q 2 Nu 1 Tau —> 3 (0.60%)

Thanks and sorry if I misunderstood something.

Hope this clears it up!

lviliani · 2025-01-10T12:48:39Z

Thanks for the quick reply.

For the HH case, it is a glugluToHH cental gridpack. For the qq_WWZZ It is a private gridpack, producing q q > w w z z

Regarding this, I was asking about the specific Powheg process used for generating the gridpack, because I'm not aware of any Powheg code capable to produce qq->WWZZ (i.e. I don't see it in this list, but maybe I'm just overlooking).

Regarding the taus, and focusing on the 3l case as an example, my comment is that you are keeping only the events in which one Z and one W decay as Z->ee/mm , W->ev/mv and the other two as Z->qq/tautau W->qq/tauv.
But you are not keeping other configurations such as (Z->tautau W->tauv Z->qq W->qq) that can still lead to a final state with 3 electrons or muons considering leptonic tau decays.
Depending on the analysis cuts, these final states can have an impact.

Also, note that you are not including channels with Z to neutrinos, which might also be relevant,

asculac · 2025-01-14T13:43:56Z

Thanks for the quick reply.

For the HH case, it is a glugluToHH cental gridpack. For the qq_WWZZ It is a private gridpack, producing q q > w w z z

Regarding this, I was asking about the specific Powheg process used for generating the gridpack, because I'm not aware of any Powheg code capable to produce qq->WWZZ (i.e. I don't see it in this list, but maybe I'm just overlooking).

Oh thanks a lot for pointing that out. The gridpack for qq_WWZZ was done with MADGRAPH, not POWHEG. And then I see now that in fragment it is wrong (also in the naming) and that we left the same settings as for HH_WWZZ where the central gridpack is POWHEG. This needs to be fixed in the provided qq fragments, sorry for the confusion. I am a bit surprised that this fragment in the end didn't cause problems and that we managed to run it and produce samples without problems, but I guess then that this part of the fragment is just ignored?

Regarding the taus, and focusing on the 3l case as an example, my comment is that you are keeping only the events in which one Z and one W decay as Z->ee/mm , W->ev/mv and the other two as Z->qq/tautau W->qq/tauv. But you are not keeping other configurations such as (Z->tautau W->tauv Z->qq W->qq) that can still lead to a final state with 3 electrons or muons considering leptonic tau decays. Depending on the analysis cuts, these final states can have an impact.

Correct, we do not consider leptons coming from taus in our analysis. These samples are used for BDT training, used to target the leptonic decays from HH-WWZZ coming directly from W and Z bosons.

Also, note that you are not including channels with Z to neutrinos, which might also be relevant,

We are also not considering final states with Z->neutrinos, same argument as the previous one.

lviliani · 2025-01-14T16:05:01Z

Thanks! Please fix the fragment then.

About taus, if you are sure you don't have any contamination from leptonic tau decays in the analysis, then I guess it's ok.

asculac · 2025-02-14T12:41:01Z

Thanks! Please fix the fragment then.

About taus, if you are sure you don't have any contamination from leptonic tau decays in the analysis, then I guess it's ok.

I am so sorry for the delay. I was convinced that I had pushed the changes to the branch. Let me know if anything else is needed and thank you!

asculac added 13 commits December 18, 2024 12:40

Create HHToWWZZ_3l.py

e05dcec

Create HHToWWZZ_4lplus.py

a7f8989

Create qqWWZZ_3l.py

4004c35

Create qqWWZZ_4lplus.py

3c2c238

Create HHToWWZZ_3l_TuneCP5_13p6TeV_powheg-pythia8.py.py

d9281ca

Edit name

b5a1a3d

Create HHToWWZZ_4lplus_TuneCP5_13p6TeV_powheg-pythia8.py

ce85095

small edit

e7f04bd

small edit

1a69251

small edit

fbcbd2f

small edit

fb173fc

Create qqWWZZ_3l_TuneCP5_13p6TeV_powheg-pythia8.py

1fbce1d

Create qqWWZZ_4lplus_TuneCP5_13p6TeV_powheg-pythia8.py

4643078

lviliani reviewed Jan 9, 2025

View reviewed changes

asculac added 6 commits January 31, 2025 14:34

Merge branch 'cms-sw:master' into HH_WWZZ

e43e114

Merge branch 'cms-sw:master' into HH_WWZZ

8879cd0

fix qqWWZZ_3l_13TeV

6dac54e

fix qqWWZZ4lplus_13TeV

ed4fce6

fix qqWWZZ_3l_13p6TeV

43c61b9

fix qqWWZZ4lplus_13p6TeV

dc15858

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding fragments for HH_WWZZ analysis #3822

Adding fragments for HH_WWZZ analysis #3822

asculac commented Dec 18, 2024

lviliani left a comment

lviliani Jan 9, 2025

asculac Jan 10, 2025

lviliani Jan 9, 2025

lviliani Jan 9, 2025

asculac commented Jan 10, 2025

lviliani commented Jan 10, 2025

asculac commented Jan 14, 2025

lviliani commented Jan 14, 2025

asculac commented Feb 14, 2025

Adding fragments for HH_WWZZ analysis #3822

Are you sure you want to change the base?

Adding fragments for HH_WWZZ analysis #3822

Conversation

asculac commented Dec 18, 2024

lviliani left a comment

Choose a reason for hiding this comment

lviliani Jan 9, 2025

Choose a reason for hiding this comment

asculac Jan 10, 2025

Choose a reason for hiding this comment

lviliani Jan 9, 2025

Choose a reason for hiding this comment

lviliani Jan 9, 2025

Choose a reason for hiding this comment

asculac commented Jan 10, 2025

lviliani commented Jan 10, 2025

asculac commented Jan 14, 2025

lviliani commented Jan 14, 2025

asculac commented Feb 14, 2025