Measurements of σ(V+D*)/σ(V) in 9.7 fb-1 of data at CDF Run II

Keith Matera, Kevin Pitts
University of Illinois at Urbana-Champaign
Contact authors

CDF Public Note 11087

Download figures in png/eps/ps format by clicking on the appropriate link beneath the image.
The tex link beneath each image leads to an appropriate caption in tex form.
Tables can be downloaded in eps or tex format.


This analysis presents measurements of σ(W+D*)/σ(W) and σ(Z+D*)/σ(Z) in the W/Z leptonic decay channels using full D* reconstruction. In a sample of W and Z events skimmed from 9.7 fb-1 of high-pT electron and muon data in pp collisions at √s=1.96 GeV at the CDF , we identify charm by fully reconstructing D*(2010)→ D0(→ Kπ)πs decays at the track level. Using a binned fit of Δm = m(Kππs) - m(Kπ) to count reconstructed D* candidates, we then unfold these raw counts with acceptance values derived from Monte Carlo. All measurements are found to be in agreement with Pythia Monte Carlo predictions. This analysis includes the first measurement of W/Z+D* production with pT(c) < 15 GeV at the Tevatron.

For more information, see CDF Public Note 11087

 Counting W/Z+D* events

  • We search for W and Z decays in the electron and muon channels using standard cuts. We then search all tracks with |η|<1.1 in the vicinity of the high-momentum leptons from W/Z decay for evidence of decay D*(2010)→ D0(→ Kπ)πs. Candidate events must pass a set of flat cuts, as well as a cut on the output score of a trained neural network. The difference in invariant mass of the reconstructed D* and D0 vertices, Δm = m(Kπ)-m(Kππs), is binned for all candidates. The resulting plot is fit to a signal-plus-background hypothesis to count W/Z+D* events.

  • Fitted plots of Δm = m(Kπ)-m(Kππs) for all W/Z+D* candidates which pass our cuts:

         .png     .eps     .ps     .tex       .png     .eps     .ps     .tex 
 Acceptance rates

  • The acceptance rates of our W/Z and W/Z+D* tagging algorithms are determined by running on simulated events using Pythia 6.2 Monte Carlo. The acceptance rate for W/Z+D* events drops to zero for D* with a transverse momentum below 3 GeV. We report ``inclusive'' acceptance rates for all events with D* transverse momentum greater than 3 GeV:

         .png     .tex 

  • We also display below the differential acceptance rate as a function of D* transverse momentum, for the example case of W(→μν)+D* events:
         .png     .eps     .ps     .tex 


  • Statistical uncertainties (and uncertainties on the whole) are dominated by the statistics of our Δm = m(Kπ)-m(Kππs) signal peaks.

  • The method by which we account for backgrounds when determining the rates (W+D*)tagged/Wtagged (where ``tagged'' refers to the count of tagged events of that type, minus any background contributions), is responsible for the majority of our systematic uncertainty. We refer to this as Backgrounds systematic uncertainty.

  • Our signal modeling systematic uncertainty comes from a finite sample size of simulated events when determining acceptance rates. Our acceptance rates are also affected by a PDF uncertainty.

  • The contribution from each of these sources of uncertainty to the uncertainty in our final results, is summarized in the tables below.

         .png     .tex 

         .png     .tex 

         .png     .tex 
 Measurements of σ(V+D*)/σ(V)

  • We apply acceptance values to tagged rates (W+D*)tagged/Wtagged and unfold to measurements of σ(W/Z+D*)/σ(W/Z), both for the inclusive sample (D* transverse momentum > 3 GeV), and differentially as a function of the D* transverse momentum. Results from the electron and muon decay channels are combined with a best linear uncertainty estimate, assuming that systematic uncertainties are fully correlated.

  • We present here the inclusive measurements:

         .png     .tex 

  • And differential measurements as a function of the D* transverse momentum:

         .png     .tex 

  • As well as plots of the differential measurements as a function of the D* transverse momentum. Statistical uncertainty in these plots is shown with error bars, while statistical plus systematic uncertainty is displayed as a yellow error band. Uncertainty in the theoretical predictions is displayed as red lines on either side of the dotted red line representing the central theoretical value.
         .png     .eps     .ps     .tex       .png     .eps     .ps     .tex       .png     .eps     .ps     .tex 
 Signal fraction by production process

  • We also determine the fraction of our W+D* signal that comes from each of the three lowest-order W+D* production processes. The best measurement of the pp→W+c fraction is obtained by subtracting the number of ``same sign'' (SS) W+D* signal events (the W and D* have the sign sign) from the number of ``opposite sign'' (OS) W+D* signal events. The best measurement of the pp→W+bb fraction (B→D*+X) is found using a two-layer system of neural networks, with each neural network trained to recognize one type of process over another. The remaining fraction is attributed to pp→W+cc production, assuming that the uncertainties in the other fractions are completely uncorrelated (as they were derived by completely different means).

         .png     .tex