Fitting for a Z VBF signalChristian Gutschow, Kiran Joshi,

Emily Nurse, Andy Pilkington

SM EWK meeting, 10th August 2012

Select Z + 2 jet events Make VBF like cuts Fit to one discriminating variable for the number of signal and background events

VBFEmily Nurse 2

Datasets and cuts

• Z candidate • Two jets, pT

1 > 45 GeV, pT2 > 35 GeV

• Jet veto: No jets in between the two leading jets with pT > 25 GeV

• ΔΦ(j,j) > π/2• ΔΦ(Z,jj) > 2.9• Fit the mjj distribution

• 5.6 fb-1 of 8 TeV Zμμ and Zee candidates• Background: Sherpa QCD Z + jets samples

– Blah– Blah

• Signal: Currently using 7 TeV Sherpa EWK samples

VBFEmily Nurse 3

Fits to 8 TeV electron + muon

BackgroundBackground+Signal

Use signal template from 7 TeV

Signal events:Input = 71 events (assuming 30% larger cross-section at 8 TeV)Fit for 79 +59

-55

We should have 3 times this dataset for HCPMore events for W production

VBFEmily Nurse 4

1000 MC toys with 3 times the data (16.8 fb-1)

213 input events ~2σ significance

VBFEmily Nurse 5

mjj background shape

Can we define control regions to test the Sherpa model of the background shape?

• Z candidate Keep these cuts• Two jets, pT

1 > 45 GeV, pT2 > 35 GeV

• Jet veto: No jets in between the two leading jets Remove, relax or reverse• ΔΦ(j,j) > π/2• ΔΦ(Z,jj) > 2.9

signal region

VBFEmily Nurse 6

mjj in control regions

In events where the Z and di-jet system do not balance: MC overestimates mjj tails

These events generally either:• have additional jets, or • the di-jet system comes from DPI or pileup

Remove all cuts Remove all, reverse Φ(Z,jj)

VBFEmily Nurse 7

Look at events with Φ(Z,jj) < 2.5 and mjj > 750 GeV

VBFEmily Nurse 8

mjj in control regionsRemove all cuts Remove all, reverse Φ(Z,jj)

No

Z p

T c

ut

Z p

T >

20

GeV

VBFEmily Nurse 9

mjj in control regions, with Z pT > 20 GeV

Signal region Remove jet veto, relax Φ(Z,jj) > 2.7, reverse Φ(j,j)

Remove Φ(j,j), relax Φ(Z,jj) > 2.7, require extra jet Relax Φ(Z,jj) > 2.7, require extra jet OR reverse Φ(j,j)

VBFEmily Nurse 10

Conclusion on control regions

Within the current limited stats the control regions are well described

We will have three (six) times the data and much more MC We could use the W+jets events to check mjj in control regions

VBFEmily Nurse 11

JES systematic

Make mjj background templates with JES scaled up one sigma

Fit to a pol1

Within the current limited stats the JES systematic is small

We will have much more MC We could use the W+jets events to

check the JES effect on mjj

VBFEmily Nurse 12

JER systematic

Make mjj background templates with JER scaled up one sigma

Fit to a pol1

Within the current limited stats the JER systematic is small

We will have much more MC We could use the W+jets events to

check the JER effect on mjj

VBFEmily Nurse 13

How systematics effect mjj fits• So far not much sign of systematic shift within stats• Try weighting background mjj spectrum by a 1st order polynomial

• Do 1000 toy MC fits to shifted template

1 – (0.1/3000)×mjj

213 input events

1 – (0.2/3000)×mjj 1 – (0.3/3000)×mjj

213 input events 213 input events

Recall statistical uncertainty ~100 events ~2σ significance

± ~60 events ± ~90 events ± ~140 events