Published Results of SECVTX + Kinematics Cross Section on September 2004
(Using lepton+jets channel)
Taka Maruyama and Mel Shochet
University of Chicago        

September 2004 Cross Section Results, using 161.6 pb^-1
6.0 ± 1.6(stat.) ± 1.2 (syst.) pb
(The signal fraction is 0.68 +0.14-0.16(stat.) in 57 events.)

The t-tbar cross section is extracted by fitting the leading jet ET spectrum to the sum of signal and background distributions. The various backgrounds have very similar shapes. The background ET distribution used for the final fit comes entirely from data: W+3,4 jet events without a b-tag, plus a non-W contribution from non-isolated lepton events.

  • Blessed plots and tables (on 2004-Sep-02: paper seminar).

    • Description gif eps
    Fit templates for fignal (left) and background (right).
    The signal template is come from ttbar MC using HERWIG (175GeV) and background uses W+3 or more jets light flavor jets data.     		eps(sig)

    eps(bkg)

    The leading jet Et (GeV) distribution with best fit curve in the signal region (W+3, or more jets):
    Two components fit using the shape of the ttbar signal (red) and that of the background (blue) provided 0.68+0.14-0.16 signal fraction using the leading jet Et distribution. The background consists of Wbbbar, Wccbar, Wc, mistags from b-tagging and non-W components, which have quite similar spectra. The background shape was extracted from W+light flavor data (W+jets with no jet b-tagged) and non isolated lepton sample.     		eps
    The size of the systematic uncertainties for the current analysis. eps
    Test of our background model using the data in W+1 and W+2 jets bin.
    Red: b-tagged data, black: background expectation from the data. Figure shows the merged jet Ets in W+1 and 2 jets bins. We assume that Et shape of W+heavy flavors is same as W+light flavor sample, so that W+light sample can be used as a mimic of W+heavy flavor shape. Shaded blue shows the non-W estimated from non-isolated sample.     		eps (W+1,2)

    The ttbar signal fraction measurement using other jet variables (the 2nd and the SumEt of 2 leading jets in W+3 or more jets region.)
     Left: the fit using the 2nd leading jet, Right: Sum of 2 leading jet Et. The best fit values for signal fraction are 0.75+0.11-0.13 (the 2nd leading jet Et), 0.65+0.14-0.16 (the Sum jet Et for the 2 leading jets)     		eps (2nd)

    eps (ET1+ET2)

  • Blessed plots and tables (on 04-Mar-18).
     ttbar cross section is 6.0 +1.5-1.8 (fit) +- 0.8 (acc. + lum.)
    signal fraction os 0.67 +0.13-0.16 (fit)

    • Description gif eps
    Sensitivity plot for the signal fraction using the current statistics in the signal region.
    Comparison among the leading, 2nd leading jets and ET1+ET2. Looks ET2 is the best variable among them, but there might be some wired things (NLO/LO effects or so). Now we use ET1 for our analysis.     		eps

    Sensitivity for the ttbar cross section measurement at the various integrated luminosty.
    Table shows the sensitivity at current, 320 pb-1 and 1600 pb-1 using ET1, ET2 and ET1+ET2.     		eps

    The shape comparison of non-W background (estimated by the data) and other dominant backgrounds in W+3 or more jets sample.
     Red crosses show the estimated non-W shape and black one is that of other dominant backgrounds.     		eps

    Fit templates for fignal (left) and background (right).
    The signal template is come from ttbar MC using HERWIG (175GeV) and background uses W+3 or more jets light flavor jets data and non-W components mentioned above.     		eps(sig)

    eps(bkg)

    The leading jet Et (GeV) distribution with best fit curve in the signal region (W+3, or more jets):
    Two components fit using the shape of the ttbar signal (red) and that of the background (blue) provided 0.67+0.13-0.16 signal fraction using the leading jet Et distribution. The background consists of Wbbbar, Wccbar, Wc, mistags from b-tagging and non-W components, which have quite similar spectra. The background shape was extracted from W+light flavor data (W+jets with no jet b-tagged)     		eps
    The size of the systematic uncertainties for the current analysis. eps
    Test of our background model using the data in W+1 and W+2 jets bin.
    Red: b-tagged data, black: background expectation from the data. Top-left: W+1 jet bin test, top-right: W+2 jets bin test (left: leading jet Et, right: 2nd leading jet Et) and bottom: merged jet Ets in W+1 and 2 jets bins. We assume that Et shape of W+heavy flavors is same as W+light flavor sample, so that W+light sample can be used as a mimic of W+heavy flavor shape.     		eps (W+1)

    eps (W+2)

    eps (W+1,2)

    Shape coparisons of various jet Et variables between the ttbar signal and typical background (Wbbbar) in W+3 or more jets region (MC).
     Red: ttbar and black: Wbbbar. Top-left: the leading jet Et, top-right: the 2nd leading jet Et and bottom: sum of the two highest jet ETs (GeV).     		eps
    The ttbar signal fraction measurement using other jet variables (the 2nd and the SumEt of 2 leading jets in W+3 or more jets region.)
     Left: the fit using the 2nd leading jet, Right: Sum of 2 leading jet Et. The best fit values for signal fraction are 0.73+0.12-0.14 (the 2nd leading jet Et), 0.63+0.15-0.18 (the Sum jet Et for the 2 leading jets)     		eps (2nd)

    eps (3rd)

    Summary of the best fit values using the various jet Et variables.
     The best fit values are mentioned above. All answers from fits are consistent each other within the statistical error. We use the value of the leading jet Et as a official value. Other results are for cross checks after considering statistical and systematic errors.     		eps

  • Questions at pre-bless and their answers.

    • Description gif eps
    (Q1) What's the typical pseudo-experiments' shapes for the center values and uncertainties.
    (A1) Left: Center value, right: one sigma error from loglikelihood. We used 1000 pseudo-experiments with 57 events. (the signal fraction is assumed to be 0.67). Arrows in right figure shows the real error of data.     		eps (center)

    eps (error)

    (Q2) Btag efficiency is coming from MC, but MC stat. is not so huge. How do you operate it?
     (A2) We have reasonably good MC stat. right now (left fig.: b-tag eff. as a function of the leading jet ET), and the MC stat. error of b-tag efficiency is relatively smaller than that of data which is used for the BKG template (right fig: comparison of error size: data stat. vs. btag.) It seems that btag eff. error is only less than 50% of the other one. Means that sqrt(1.0+0.5**2)=1.12, so that only 10% effect to total error. This should be small and within given sys. error.     		eps (center)

    eps (error)

  • Blessed Plots and tables. (on 2003-Oct)
    ttbar cross section is 6.9 +1.6-1.8 (fit) +- 0.9 (acc. + lum.)
    signal fraction os 0.88 +0.10-0.16

    • Description gif eps
    The shape comparison between the ttbar signal and the typical background (Wbbbar) on leading jet Et (GeV) in W+3 or more jets bin:
    Red one shows the typical background (Wbbbar), while black one shows that of ttbar signal shape.     		eps
    Sanity check of the fitting procedure using pseudo-experiments.
    Left figures show the input of the jet Et distribution with the different signal and background ratio (i.e. 25%, 50% and 75%). Right figures show the best fit results of 1000 pseudo-experiments corresponding to the left input values (each pseudo-experiment has 35 events in W+3 or more jets for likelihood fitting).     		eps (input)

    eps (output)

    The shape comparison among W+heavy flavors (i.e. Wbb, Wcc and Wc) and W+light flavor sample (MC) of the laeding jet (GeV) in W+3 or more jets sample:
    Left: Wbbbar vs W+light flavor, middle: Wc vs W+light flavor and right: Wccbar vs W+light flavor sample.
    (Note that there is b-tagger efficiency correction in W+light flavor.)     		eps
    The shape comparison of non-W background (estimated by the data) and other dominant backgrounds in W+3 or more jets sample.
     Red crosses show the estimated non-W shape and black one is that of other dominant backgrounds.     		eps
    Mistag (of b-tagging) component vs other backgrounds in W+3 or more jets sample.
    Mistag component are compared to W+heavy flavor components. Left: Wbbbar (red) vs mistag (black), middle: Wc (green) vs mistag, right: Wccbar(blue) vs mistag.     		eps
    The leading jet Et (GeV) distribution with best fit curve in the signal region (W+3, or more jets):
    Two components fit using the shape of the ttbar signal (red) and that of the background (blue) provided 0.88+0.10-0.16 signal fraction using the leading jet Et distribution. The background consists of Wbbbar, Wccbar, Wc, mistags from b-tagging and non-W components, which have quite similar spectra. The background shape was extracted from W+light flavor data (W+jets with no jet b-tagged)     		eps
    The size of the systematic uncertainties for the current signal fraction (0.88). eps
    Test of our background model using the data in W+1 and W+2 jets bin.
    Red: b-tagged data, black: background expectation from the data. Top-left: W+1 jet bin test, top-right: W+2 jets bin test (left: leading jet Et, right: 2nd leading jet Et) and bottom: merged jet Ets in W+1 and 2 jets bins. We assume that Et shape of W+heavy flavors is same as W+light flavor sample, so that W+light sample can be used as a mimic of W+heavy flavor shape.     		eps (W+1)

    eps (W+2)

    eps (W+1,2)

    Shape coparisons of various jet Et variables between the ttbar signal and typical background (Wbbbar) in W+3 or more jets region (MC).
     Black: ttbar and red: Wbbbar. Top-left: Sum of 3 leading jet Et, top-right: the leading jet Et, bottom-left: the 2nd and bottom-right: the 3rd.     		eps
    Sensitivities of fitting using various jet Et variables. We examined the fitting sensitivities using the four variables. We regarded the width ("sigma" in the table) of the 1000 pseudo-experiments as the statistical uncertainty for the fit. eps
    The ttbar signal fraction measurement using other jet variables (the 2nd and the 3rd leading jets, and the SumEt of 3 leading jets in W+3 or more jets region.)
     Top-left: the fit using the 2nd leading jet, Top-right: the 3rd leading jet, and bottom: Sum of 3 leading jet Et. The best fit values for signal fraction are 0.96+0.04-0.12 (the 2nd leading jet Et), 0.77+0.16-0.18 (the 3rd) and 0.91+0.08-0.12 (the Sum jet Et for the 3 leading jets)     		eps (2nd)

    eps (3rd)

    eps (SumEt)

    Summary of the best fit values using the various jet Et variables.
     The best fit values are mentioned above. All answers from fits are consistent each other within the statistical error. We use the value of the leading jet Et as a official value. Other results are for cross checks after considering statistical and systematic errors.     		eps