simulated missing ET:

• myGenMET defined as the x-y component of the 3-vector obtained summing up all the neutrinos in the event (genParticle collection)

As expected, the p_T of the vector sum of all stable particles is equal to zero for both Zmumu (left) and TTbar (right) samples.

There is a spread of about 3 GeV around zero when only the stable particles with |eta| < 5.0 are summed. Again Zmumu (left) and TTbar (right).

The spread is almost equal to 4 GeV when an additional selection is applied to the set of generator particles: |eta|<5.0 and p_T > 0.5 GeV.

Transverse momentum of the vector sum of all neutrinos' momenta: Zmumu (left) and TTbar (right) samples. The Zmumu has rarely some low energy neutrinos. The TTbar sample can have high energy neutrino when the Ws decay leptonically. Other neutrinos can be produced from the b-jets.

reconstructed missing ET:

study on Zmumu sample:

In the following plots, the resolutions are defined as the RMSs of the distributions (myGenMET- recoMET).T, where T is the coordinate X or Y.

• recoMET IC5: Raw Missing Transverse Energy calculated using IC5 CaloJets.

Resolution and bias are comparable for X and Y. The peak at zero is due to event without generated neutrinos and reconstructed MET equal to zero. The latter happens when there aren't jets above the minimum threshold of energy.

The resolutions of the MET obtained by the other types of jets are very similar:

• recoMET KT4: Raw Missing Transverse Energy calculated using FastKt4 CaloJets.
• recoMET KT6: Raw Missing Transverse Energy calculated using FastKt6 CaloJets.
• recoMET SC5: Raw Missing Transverse Energy calculated using SisCone5 CaloJets .
• recoMET SC7: Raw Missing Transverse Energy calculated using SisCone7 CaloJets .

Instead of using the jets as input, the missing ET can be evaluated using the Calo Clusters.

• CaloMET met: Raw Missing Transverse Energy calculated using Scheme B CaloTowers (beyond 217 w/o HO).

Resolutions are similar to those obtained using jets as input. However the residual plots don't show the peak at zero because the MET reconstructed from clusters is very rarely exactly zero: there is ~always at least a cluster above threshold.

• recoMET tcMet: Track Corrected MET using "met", "muons", "pixelMatchGsfElectrons" and "tracks".

This reconstructed MET was not available yet when the Summer08 samples (release 21X and 22X) were produced. Can we evaluated it in 22X + tags ??

• PFMET pfMet: MET calculated from Particle Flow Candidates.

Surprisingly the resolution for ParticleFlow met is the worst one on Zmumu events. The PFMET is supposed to be corrected by the muon energy. There is something wrong here??

study on TTbar sample:

• recoMET IC5:

• CaloMET met:

• PFMET pfMet:

Comments and observations:

1. pfMet is better than calo clusters MET, which is better than MET from jets.
2. The resolution on MET is almost a factor 2 worse than the resolution obtained on Zmumu sample.
3. Should we evaluate the resolution on MET vs scalar_ET? vs MET??

-- BorisMangano - 2009/04/28