Measurement of the Top-Quark Mass with Dilepton Events Selected Using Neuroevolution at CDF

We report a measurement of the top-quark mass $M_t$ in the dilepton decay channel $t\overline{t}\to b{l}^{\prime +}{\nu }_l^{\prime }\overline{b}{l}^{-}{\overline{\nu }}_l$. Events are selected with a neural network which has been directly optimized for statistical precision in top-quark mass using neuroevolution, a technique modeled on biological evolution. The top-quark mass is extracted from per-event probability densities that are formed by the convolution of leading order matrix elements and detector resolution functions. The joint probability is the product of the probability densities from 344 candidate events in $2.0{\mathrm{fb}}^{-1}$ of $p\overline{p}$ collisions collected with the CDF II detector, yielding a measurement of $M_t=171.2\pm 2.7(\mathrm{stat})\pm 2.9(\mathrm{syst})\mathrm{GeV}/c^2$.

Figure 1

(a) Expected statistical uncertainty for the best network in each successive generation of network evaluation. The points show the average performance for each generation; the error bars show the variation due to the randomly generated networks in generation 0. (b) Expected statistical uncertainty on $M_t$ versus signal fraction after neural-network selection, for all evaluated networks. The selection [7] used in previous measurements is shown ($\star$) for comparison. The arrows show the expected statistical uncertainty and signal fraction corresponding to the network used in the analysis.

Figure 2

The output of the final network evaluated on the collected data (black triangles), with expected signal and background contributions (stacked solid histograms). The data show events passing the preselection. The evolution of the optimum selection network is performed with an a priori threshold set at 0.5 for candidate selection. Of the 642 preselected events shown, 344 events pass this threshold and constitute the final candidate sample for mass fitting.

Figure 3

(a) Mean measured $M_t$ in simulated experiments versus top-quark masses. The solid line is a linear fit to the points. (b) Pull widths from simulated experiments versus top-quark masses. The solid line is the average over all points.