Search for Anomalous Production of Di-photon+MET Events in 2 fb-1 of Data.

Authors:  Sasha Pronko (Fermilab); Ray Culbertson (Fermilab); Max Goncharov (MIT); Shin Shan Yu (Fermilab).

Results for di-photon signal region: two tight central photons with ET>13 GeV. The baseline sample is ~25% true di-photon events.
There are 31,116 events in the baseline sample before MET requirements.

Non-collision backgrounds: cosmic ray and beam halo muons showering in the detector material.
EWK backgrounds: mostly W+pho (e/tau->fake pho)  or W+jet (e/tau/jet->fake pho), with smaller contributions from Z(mumu)+2pho and Z(nunu)+2pho events
Fake MET: regular di-photon events and QCD pho-jet or jet-jet events where one or both photons are faked by a jet. Fake MET comes from energy mis-measurement
of jets or photons.


MetSig>3.0 MetSig>4.0 MetSig>5.0
Observed
 82
 31
 23
EWK with real MET
 35.4±2.2 29.9±2.0 25.9±1.9
Fake Met
 35.4±7.2 8.3±2.3 3.7±1.5
Non-collision
 0.9±0.3 0.8±0.3 0.8±0.3
Total predicted
background
 71.7±7.5 39.0±3.1 30.4±2.4


MET distribution in all selected events. The background predictions are shown by stacked histograms (eps, gif).


MET-significance distribution for all selected events. The background predictions are shown by stacked histograms (eps,gif).

MET distribution for events with MET-significance>3, 5. The background predictions are shown by stacked histograms (sig>3: eps,gif, sig>5: eps,gif).


Diphoton invariant mass distribution for events with MET-significance>3, 5. The background predictions are shown by stacked histograms (sig>3: eps,gif, sig>5: eps,gif).


ET of leading photon for events with MET-significance>3, 5. The background predictions are shown by stacked histograms (sig>3: eps,gif, sig>5: eps,gif).


HT (scalar sum of ET of all identified objects and MET) distribution for events with MET-significance>3, 5. The background predictions are shown by stacked histograms (sig>3: eps,gif, sig>5: eps,gif).


Number of jets with ET >15 GeV in events with MET-significance>3, 5. The background predictions are shown by stacked histograms (sig>3: eps,gif, sig>5: eps,gif).

ET of leading jet in events with MET-significance>3. The background predictions are shown by stacked histograms (eps,gif).


ET of identified electrons in events with MET-significance>3. The background predictions are shown by stacked histograms (eps,gif).

Support plots to illustrate METMODEL performance:

MET distribution and efficiency of MET-significance cuts for W(enu)+pho events from Baur MC (eps-a, eps-b)


MET  distribution in Pythia di-photon and Z->ee events (no real MET) in comparison with METMODEL prediction for fake MET (eps-a,eps-b)


Example of METMODEL predictions for possible fake MET and MET-significance distributions for a data event (eps-a,eps-b)


Example of jet energy resolution (JER) fits for two different jet energies (eps-a,eps-b).


Example of fake MET parameterization due to unclustered energy:
a) double Gaussian fit to METY distribution in on one the sqrt(SumET) bins (eps);
b) dependence of Gaussian width from Fig-a on sqrt(SumET) (eps). 

Created by Sasha Pronko (pronko@fnal.gov)