Measurement of the Forward-Backward Asymmetry
in Top Pair Production in 3.2/fb of ppbar collisions at sqrt(s)=1.96 TeV

Glenn L Strycker, Dan Amidei, Monica Tecchio
University of Michigan

Thomas A Schwarz, Robin Erbacher, John Conway
University of California-Davis

Documentation
Public Note

Send email to authors



Abstract


We measure the forward-backward asymmetry of pair produced top quarks using 776 semi-leptonic b-tagged ttbar events reconstructed with a chi^2 based kinematic fitter and correct the value for experimental effects. This is an update to the measurement published in PRL 101, 202001, 2009. We have increased the dataset from 1.9/fb up to 3.2/fb. We study the rapidity y_had of the hadronically-decaying top (or anti-top) system, tagging the charge with the lepton sign Q_l from the leptonically decaying system. We find the forward-backward asymmetry in the ppbar lab frame to be Afb = 0.193 +- 0.065 (stat) +- 0.024 (syst), consistent with previous results. An independent cross-check uses a likelihood fit to templates derived from linear asymmetry in the top production angle in the ttbar rest frame (1+A*cos(alpha)) and finds asymmetry consistent with our unfold method and our 1+A*cos(alpha) hypothesis. These results should compared with the small ppbar lab frame charge asymmetry expected in QCD at NLO, Afb = 0.05 +- 0.015.

Please see our public note for a complete description of our method. Below we show our main analysis plots.





This is our main plot -- if anything is to be presented, it is this! Here we show the raw distribution of -Q*yhad.
NOTE THAT THE SECOND PLOT USES CTOP75 FOR THE SIGNAL MC




This is a plot of -Q*cos(theta), which we used in our previous analysis and update here with data through 3.2/fb.
(this plot still uses PYTHIA for the SIGNAL shape)




Here are our smear, acceptance, and unfold matrices




Our systematic uncertainties




The final measured Afb values. The template method value was obtained by finding the minimum of a -log(likelihood) plot for template fits as explained in the note.




Here are the Afb values for the background components... the nJet >=4 values are also on the following plot




This plot shows the various contributions to the background with their individual Afbs.





Because the antitagged (nTag=0) sample has a much larger background component, this plot of -Q*yhad for the antitagged sample shows that our background MC is well modeled and agrees with the data.
NOTE THAT THE SECOND PLOT USES CTOP75 FOR THE SIGNAL MC




We explain this plot in the public note. It is a check of our unfold method, showing that the measured Afb value from MC events after event selection and reconstructed matches the Afb in the truth information before cuts.




This plot demonstrates how our -Q*yhad distribution changes during our correction procedure. We begin with the data (black), subtract off the background (blue) arriving at the data-bkg shape (green). After unfolding we arrive at the final shape (red). Binning has been optimized as explained in our note.




This set of plots demonstrates our reweight procedure. We begin with Pythia MC with no asymmetry (the far left plot, black shape). We reweight by 1+A*cos(theta) in the ttbar rest frame (red shape). See the note for details on how we propagate this reweight to the lab frame, change variables to rapidity, and reweight a post-selection shape.




Here are several functions we used for our shape systematic. We begin with cos(theta) in the ttbar-frame, reweight by different functions, and propagate the reweight into the final ppbar-frame rapidity variable. This plot demonstrates that very different shapes in ppbar cos(th) end up looking similar in ppbar-frame rapidity. The final Afbs are the same for all functions.




Here we show several of our reweighted template shapes for different values of ttbar-frame asymmetry.




Several templates along with data. Note that the 17% lab-frame asymmetry sample agrees with the data quite well, and better than the other templates.




Here is a template with a 19% Afb (which corresponds to our Afb measurement using the unfold method).




Hadronically-decaying Top quark, nominal sample plot and the signal shape reweighted to have a 19% asymmetry in the ppbar frame.




Hadronically-decaying W Boson, nominal sample plot and the signal shape reweighted to have a 19% asymmetry in the ppbar frame.




Hadronically-decaying B quark, nominal sample plot and the signal shape reweighted to have a 19% asymmetry in the ppbar frame.



Links

Previous public page for 1.9/fb Tevatron data
PRL paper
Arxiv paper (hep-ex/0806.2472)
APS April Meeting Abstract
APS April Meeting Slides
Glenn Strycker's Google Profile