2017 Data/MC is not matching

[ram1123]

Describe the bug
Huge discripency in the data/mc of 2017.

To Reproduce
Steps to reproduce the behavior:

1. CMSSW version : 'CMSSW_9_4_13'
2. Use the git commit : ram1123@8e9e9ac

Additional context

To resolve this issue start with following:

• Make data/mc for just Z -> mu mu selection for 2017
• Compare the above with 2016.

[ram1123]

Data/MC for 2017 (Z -> mu mu)

CUT

the_cut = genWeight*
        ((type==0) && 
        (l_pt1>50 && (((type==0)&&(abs(l_eta1)<2.4)) )) && 
        (l_pt2>45 && (((type==0)&&(abs(l_eta2)<2.4)))))

Yield

Total MC	= 327873
data	= 2.44552e+06
data/MC	= 7.45875
VV	= 13197.5
W+jets	= 456.901
Z+jets	=  311963
top	= 1616.52
QCD	= 638.854

log plot	Linear plot

The above distributions are made using commit: ram1123@7557a2d

[ram1123]

@arapyan,

Data/MC for 2016 (Z -> mu mu)

CUT

the_cut = genWeight*
          ((type==0) && 
          (l_pt1>50 && (((type==0)&&(abs(l_eta1)<2.4)) )) && 
          (l_pt2>45 && (((type==0)&&(abs(l_eta2)<2.4)))))

Yield

Total MC	= 582625
data	= 10175
data/MC	= 0.0174641
W+jets	= 208.885
Z+jets	=  574765
top	= 7651.97

log plot	Linear plot

Even after applying the eff and id weights and including the opposite charge cut and mass window it does not effect significantly. In the code used there still MET part is there. So, just to confirm if its not doing anything strange, going to remove it and re-run the setup.

The above distributions are made using commit: ram1123@d605fce

[ram1123]

@arapyan

Data/MC for 2016 (After removing MET selection)

Below plots are made using code: ram1123@0a29e32 . Here we just selected the two leptons in pre-selection. Here only drell-yan background will matter and for Drell-yan sample we have following:

1. DY1JetsToLL
2. DY2JetsToLL
3. DY3JetsToLL
4. DY4JetsToLL
   We might be loosing the DY0-jet contribution. This might be the reason.

Cut

the_cut = genWeight*
         ((type==0) && 
         (l_pt1>50 && (((type==0)&&(abs(l_eta1)<2.4)) )) && 
         (l_pt2>45 && (((type==0)&&(abs(l_eta2)<2.4)))) && 
         ((l_charge1>0 && l_charge2<0) || (l_charge1<0 && l_charge2>0)) 

Yield

Total MC	= 587612
data	= 1.29447e+06
data/MC	= 2.20293
VV	= 4089.72
W+jets	=  157.58
Z+jets	=  574749
top	= 7251.45
QCD	= 1364.75

log plot	Linear plot

[ram1123]

Hi @arapyan,

For 2017, the situation remains same as in previous post, where data/mc is ~7. I think we don't have HTbin sample less than 100 GeV. It might be the main culprit for less MC. We can check this by using the inclusive Drell-yan sample. What do you say?

[ram1123]

Run2016 only

Hi @arapyan,

After using the amc@nlo sample of Z+jets the data/mc seems reasonable for 2016.

Using just genweights:

log plot	Linear plot

the_cut = genWeight*
        ((type==0) && 
       (l_pt1>50 && (((type==0)&&(abs(l_eta1)<2.4)) )) && 
       (l_pt2>45 && (((type==0)&&(abs(l_eta2)<2.4)))) && 
      ((l_charge1>0 && l_charge2<0) || (l_charge1<0 && l_charge2>0)))

Total MC	= 1.18622e+06
data	= 1.29447e+06
data/MC	= 1.09126
VV	= 4089.72
W+jets	=  157.58
Z+jets	= 1.17336e+06
top	= 7251.45
QCD	= 1364.75

Using all weights

log plot	Linear plot

the_cut = genWeight*trig_eff_Weight2*trig_eff_Weight*id_eff_Weight*id_eff_Weight2*pu_Weight*
        ((type==0) && 
        (l_pt1>50 && (((type==0)&&(abs(l_eta1)<2.4)) )) && 
        (l_pt2>45 && (((type==0)&&(abs(l_eta2)<2.4)))) && 
        ((l_charge1>0 && l_charge2<0) || (l_charge1<0 && l_charge2>0)))

Total MC	= 1.11166e+06
data	= 1.29447e+06
data/MC	= 1.16445
VV	= 3840.69
W+jets	= 144.615
Z+jets	= 1.09947e+06
top	= 6788.07
QCD	= 1414.83

With/Without pileup weight

wgt	log plot	Linear plot
No pileup weight
with pileup weight

With/Without Trigger weight

weight = trig_eff_Weight2*trig_eff_Weight*genWeight

wgt	log plot	Linear plot
No weight
With Weight

With/Without ID/ISO weight

weight = id_eff_Weight*id_eff_Weight2*genWeight

wgt	log plot	Linear plot
No weight
With Weight