Top quark mass measurement in the missing ET + jets channel




Authors

G. Busetto, Giorgio Cortiana, J. Donini, T. Dorigo


INFN & University of Padova, Italy



Read the article (PhysRevD_75_111103, 2006)
Write to the authors

Summary



EPS 		We study events collected by a multi-jet trigger and selected using optimized kinematical criteria that contain jets identified in the Silicon VerteX detector as originating from b-quarks.

In a new approach, (see the cross section measurement public web page ) significant missing transverse energy and related geometrical properties, rather than lepton identification requirements, are used to discriminate the top quark pairs production from other background processes. Moreover, b-jet identification parametrized probabilities, measured directly from data, are used to predict the overall amount of background-produced b-jets in the kinematically selected sample.

A measurement of the top quark mass in the met+jets channel is performed exploiting the correlation between Mtop and HT which is defined as the sum of the missing transverse energy and all jet energies. Using 311 pb-1 of ppbar data collected by the CDF detector we perform un-binned likelihood fits of the observed HT variable to the sum of Monte Carlo ttbar and data-driven background templates. Pseudo-experiment techniques are used in order to check the method stability and reliability and to estimate the expected statistical error. We measure a top quark mass of :

mtop = 172.3 +10.8 -9.6 (stat.) +10.8 -10.8 (syst.) GeV/c2


Data Sample

Number of positive SecVtx tagged jets as a function of the jet multiplicity in the sample after the optimized kinematical selection. Data are represented by dots, the solid grey histogram shows the tagging matrix background prediction while the red and green lines represent the signal + background expectation for the inclusive and tau+jets ttbar decays respectively normalized to the theoretical cross section of 6.1 pb (i.e. assuming a top quark mass of 178 GeV/c2).
 EPS
The sample used to perfom the measurement of the top quark mass is the same as the one described for the cross section analysis in the missing ET+jets final state ( see here for more details ).

We select events having:
  • at least 4 jets with ET >= 15 GeV;
  • missing ET significance (missing ET / (sum ET)1/2) >= 4.0 GeV1/2;
  • mininimum Delta(phi) difference between the missing missing ET ad jets >= 0.4 rad.
The number of b-tagged jets in events with >=4 jets can be used for a top quark mass measurement.
Anyway, since in this analysis lepton informations are missing, and in particular we veto events with a good lepton (e/mu) candidate, the event kinematics can not be completely reconstructed, so that only global event variables correlated with mtop, such HT, can be used for a mass measurement.
Signal and background template parameterization

Background HT distribution in the final sample is derived directly from data uding a b-tagging rate parameterization. The template is corrected for the signal contamination and parameterized using a Gamma plus two Gausians functions.
EPS
The expected signal HT distributions are built as a function of the top quark mass. Analytical probability density functions (PDFs) of HT as a function of Mtop are extracted by fitting different signal templates generated with input top quark mass in the range [150, 200] GeV/c2. The PDFs for 5 value of M_top in the >=1 tag sample, along with the histograms used to derive the PDFs, are shown here.

EPS

The capability of the data-driven background parameterization to correctly predict the normalization and shape of observed HT is cross checked in signal-depleted data control region as defined in cross section web page . The overall background normalization is known to within 10% level, as shown by the ratio between observed to expected distribution in the bottom insets of the figure.
EPS

Sanity Checks

Various checks are performed to ensure that our fitting procedure is un-biased and perform as expected over the top mass range 150-200 GeV/c2.
We report the results of pseudo-experiment performed using a known input top quark masses. This plot shows the linearity between input and fitted top quark mass.
 EPS
This plot show the residuals, defined as Mtop fit - Mtopinput as a function of the input top quark mass.
 EPS





Pull mean distribution as a function of the input top quark mass.
 EPS
Pull sigma distribution as a function of the input top quark mass. Pulls sigma are slightly higher than one, indicating a general under-estimate of the statistical uncertainty associated to the measurement. For this reason in the final result the statistical uncertainty are increased by a factor 1.08.
 EPS

Fit to the data

This plots shows the final fit to the data. The inset shows the likelihood profile as a function of Mtop. The minimum is found for Mtop = 172.3 GeV/c2.
 EPS
Conparison between expected and measured statistical uncertainty on the top quark mass measurement, before the application of the correction factor of 1.08.
 EPS

Systematic uncetainties
Jet Energy Scale
9.6 GeV/c2
Generator
3.8 GeV/c2
Background shape
2.1 GeV/c2
Parton Distribution Functions
1.5 GeV/c2
Initial State Radiation
0.9 GeV/c2
Final State Radiation
0.9 GeV/c2
Background fraction
0.8 GeV/c2
b-jet energy scale
0.7 GeV/c2
Trigger efficiency
0.7 GeV/c2
Monte Carlo statistics
0.6 GeV/c2
b-tagging
0.5 GeV/c2
Total
10.8 GeV/c2


Last modified: Sat Jun 30 18:32:12 CDT 2007