1.Inclusive and Exclusive Jets

Exclusive Jets:

1. Find the smallest of d_ij and d_iB below some d_cut.
2. If d_ij -> replace ij with a new particle. If d_iB -> remove i and take it as part of the beam jets.
3. Finally one will get Jet1,Jet2,Jet3…Beam Jet

Inclusive Jets:

1. Find the smallest of d_ij and d_iB without d_cut
2. If d_ij ->replace ij with a new particle. If d_iB -> remove i and take it as an inclusive jet.
3. Finally one will get Jet1,Jet2,Jet3…

3. Jet Clustering

Jet Finding (Vertex-Assisted)

​

3. Results in an increased chance of correct jet separation. This effect is pronounced in final states with many b jets.

Vertex/Jet association are further refined after the jets are identified.

​

1. Difficult to separate two b-jets which are close. Ordinary kt algorithm tends to merge them.

2. To overcome this, find secondary vertices first using all tracks in the event, and use them as seeds for jet finding.

3. Jet Clustering

LCFIPlus Algorithms

​

3. LCFIPlus Algorithms - Jet Clustering

​

1. The vertices and leptons are treated as jet cores. If the number of jet cores is larger than the required number of jets, the nearest jet cores are combined until the required number is reached
2. Combine the remaining tracks and neutral clusters to one of the jet cores in two steps
1. Cone jet clustering: particle fall within 0.2 rad of the jet core are merged with that jet core
2. The remaining particles (tracks or clusters) are combined to the jet cores based on the modified Durham distance

arXiv:1506.08371

4. Reco and Analysis Code

H->bb/cc/gg Motivation Part

The masses of the fermions M_f in the SM are proportional to their Yukawa couplings h to the Higgs field: M = vh/sqrt(2).Thus measuring the Yukawa couplings between the Higgs boson and the SM fermions is essential to understand the origin of the fermions' masses.

Deviation of these couplings from SM prediction would indicate new physics.

https://iopscience.iop.org/article/10.1088/1674-1137/43/4/043002/pdf

Analysis goal : Measurement of H->bb H->gg H->cc branching fraction in the CEPC experiment.

Chinese Physics C Vol. 44, No. 1 (2020) 013001

H->bb/cc/gg Paper Event Selection Part 1

Signal : 2 SFOS leptons + 2 jets (llH with H->bb/cc/gg)

Background :

1.Irreducible : Semileptonic ZZ process (Z1->ll,Z2->qq)

2.Reducible : All the other backgrounds with different final states (hadronic/leptonic WW,hadronic ZZ,llH with other final states, other higgs production process,lepton pair or quark pair)

Cuts:

1.lep_energy > 20 GeV, extra isolated lepton veto √

2.Polar angle of lepton pair system : |Cos(𝜃)|<0.71(0.81) for ee(𝜇𝜇) √

3.Angle between the two isolation tracks : Cosφ>-0.93(-0.74) for ee(𝜇𝜇) √

4.M_ll in Z mass window (77.5~104.5) GeV √

5.Polar angle |Cos(𝜃)|<0.96 for jets,jet contains 20 particles at least, each with energy no less than 0.4 GeV (against fake jets from photons or leptons).Mjj should be (75~150) to reject irreducible backgrounds. √

6.Higgs mass window is defined by requiring M_ll,recoil between (124~140) GeV √

H->bb/cc/gg Paper Selection Part 2

Event yield of cutflow

H->bb/cc/gg Paper Fit Part 1

A unbinned likelihood fit is performed on three variables : M_ll,recoil,X_B,X_C simultaneously.

​

With the overall likelihood function constructed as :

​

​

L the likeness of two individual jets

The RooFit package is implemented to perform an unbinned likelihood fit to the weighted events

H->bb/cc/gg Paper Fit Part 2

Projection on M_ll,recoil, X_B, X_C .

The model describes the data very well

H->bb/cc/gg Paper Uncertainty Part 1

Statistical: 10000 toy MC according to Poisson distribution.

​

Modelling of M_ll distribution: Fit to signal and background dataset respectively.

​

Fixed background: Set H->WW* H->ZZ* yields 5% higher and lower. Vary non-ZZ backgrounds by ±100%.

​

Event selection: Extra lepton veto efficiency, jet mass pair resolution

H->bb/cc/gg Paper Uncertainty Part 2

Uncertainty Summary

Previous

Modified Yu Bai’s reco-xml to our case, 4jet, 2lep

Electron channel: 4-fermion bkg reco done, Higgs bkg reco done, signal reco done.

Muon channel: None

Could start to analyze the electron channel once all done

Chinese Physics C Vol. 44, No. 1 (2020) 013001

Topic for January 4th

Sample Production (reco-level sample production)

​

Analysis ntuple production (key variables and normalization information added to the ntuple)

​

Kinetic distribution and cutflow table

Sample Production

Electron channel:

Bkg all done: 2-fermion, 4-fermion, ffH.

Signal all done: mass from 5GeV to 60 GeV according to

​

Muon channel:

Bkg all done: 2-fermion, 4-fermion, ffH.

Signal need to be done: Waiting for Xuliang

arXiv:1911.10210v1

Analysis Ntuple Production

Variables stored in the ntuple:

4 jet px/py/pz/e/pt, 4 jet b tag value, n_elec/n_mu, ll_costheta/ll_cosphi, lep_pt, m_ll, mrecoil_ll, 4j_inv_mass, 4j_recoil_mass, weight=xsection*Lumi/n_of_events.

(need to add opening angle of jets from a, lep_Pxyze, energy of a.)

​

Automatic production macro are prepared. Analysis ntuples could be found at ihep’s server at : /publicfs/atlas/atlasnew/SM/VBS/zhenw/cepc/exotic/sample_backup/e_bkg_ntuple and /publicfs/atlas/atlasnew/SM/VBS/zhenw/cepc/exotic/sample_backup/e_sig_ntuple

Xsection reference: http://cepcsoft.ihep.ac.cn/guides/Generation/docs/ExistingSamples/#4-fermion-backgrounds-1

To do last time

Need to add opening angle of jets from a, lep_Pxyze, energy of a.Explain opening angle at CEPC (Gang)

​

Make plots of opening angles.

​

Verify the signal sample (Added the BR of a->bb)

​

Strategy for combination of bjets (difference in a mass, …)

Variable Distribution

Variable Distribution

Variable Distribution

Variable Distribution

Kinetic Distribution and Cutflow

Origin : After reco-xml from CEPC framework

​

n_elec : Extra lepton veto

​

ll_costheta : Polar angle of ll system

​

ll_cosphi : Separation angle between two tracks

​

mll : 77.5~104.5

​

mrecoil_ll : 124~140

​

b_likeness : L_b1*L_b2*L_b3*L_b4 / (L_b1*L_b2*L_b3*L_b4 + (1-L_b1)*(1-L_b2)*(1-L_b3)*(1-L_b4))

Kinetic Distribution and Cutflow

npfo4j : number of particles in the 4 jets each with energy larger than 0.4 GeV

​

jet_pt : ljpt, sljpt, ssljpt, sssljpt

​

amass_dif : mass difference between the two reconstructed exotic particle a

​

open_angle : opening angle between the two jets of the exotic paritcle a

Kinetic Distribution and Cutflow

Distribution of 4j invariant mass after all the cuts.

​

Dominant background is from ZH_bb process

TRExFitter as Limit Setting Framework

Settings :

Variable : m_recoil_ll

Signal Sample : Exotic decay signal at different mass point after selection

Bkg Sample : All the background event after selection

Data : Asimov data with all bkg samples

Limits at different mass points

Uncertainty due to Event Selection

Randomly changed cuts, need to classify those uncertainties and do detail study.

BDT Implementation

Using variables beyond the baseline cuts.

Trained bkg and signal events with basic cuts (b-likeness>0.01) applied.

BDT Implementation

Classifier output distribution

BDT Implementation

Limit setting with BDT cut at 0.05

BDT gives better limits than cut based results.

Current Status

Complete the uncertainty study of electron channel:

Statistical, fixed background, event selection, …

​

Try pyhf to do limit setting or sensitivity study ?

​

​

Plans

Extend the range for the m_recoil_ll

​

Optimization of cuts targeting different mass point signals

​

BDT optimization targeting different mass point signals (parametrized training)

H->bb/cc/gg(Bai Yu) : Chinese Physics C Vol. 44, No. 1 (2020) 013001

Train Sample size (befor loose cut) :

Signal = 100000, Background = 30000

​

​

Test Sample size:

Signal = 196562, Background = 1628

​

Energy Resolution (Z->veve, H->bb)

ee->ZH, Z->veve, H->bb

Leading jet

Didn’t use FSClasser

Fullsim->LCFIplus -> processor

Sub leading jet

Energy Resolution (Z->veve, H->bb)

Energy Resolution (Z->veve H->aa->4b)

Z->veve, H->aa->4b

5E4 events

m1 = 50 GeV

leading jet

softest jet

Energy Resolution (Z->veve H->aa->4b)

​

Z->veve, H->aa->4b

5E4 events

m1 = 50 GeV

Energy resolution (Z->veve H->aa->4b)