Measurement of the top quark mass in final states with two leptons

We present measurements of the top quark mass ($m_t$) in $t\overline{t}$ candidate events with two final state leptons using $1{\mathrm{fb}}^{-1}$ of data collected by the D0 experiment. Our data sample is selected by requiring two fully identified leptons or by relaxing one lepton requirement to an isolated track if at least one jet is tagged as a $b$ jet. The top quark mass is extracted after reconstructing the event kinematics under the $t\overline{t}$ hypothesis using two methods. In the first method, we integrate over expected neutrino rapidity distributions, and in the second we calculate a weight for the possible top quark masses based on the observed particle momenta and the known parton distribution functions. We analyze 83 candidate events in the data and obtain $m_t=176.2\pm 4.8(\mathrm{stat})\pm 2.1(\mathrm{sys})\mathrm{GeV}$ and $m_t=173.2\pm 4.9(\mathrm{stat})\pm 2.0(\mathrm{sys})\mathrm{GeV}$ for the two methods, respectively. Accounting for correlations between the two methods, we combine the measurements to obtain $m_t=174.7\pm 4.4(\mathrm{stat})\pm 2.0(\mathrm{sys})\mathrm{GeV}$.

Figure 1

Comparison of the expected distributions from individual backgrounds and $t\overline{t}$ ($m_t=170\mathrm{GeV}$) in the $\ell +\mathrm{\text{track}}$ channels. (a) $M_{\ell \mathrm{t}}$ for the $e+\mathrm{\text{track}}$ channel without the requirement of the $b$ tag and with inverted ${\mathrm{E̸}}_T$ cuts. (b) ${\mathrm{E̸}}_T$ for the sum of both $\ell +\mathrm{\text{track}}$ channels, again without the $b$-tag requirement. We assume ${\sigma }_{t\overline{t}}=7.0\mathrm{pb}$.

Figure 2

(a) ${\mathrm{E̸}}_T$ and (b) leading lepton $p_T$ for $t\overline{t}$ ($m_t=170\mathrm{GeV}$) and background processes overlaid with those for observed events in all channels after the final event selection. We assume ${\sigma }_{t\overline{t}}=7.0\mathrm{pb}$.

Figure 3

Dependence of the resolution of unclustered ${\mathrm{E̸}}_T$ on the unclustered scalar transverse missing energy for $Z\to ee$ events with exactly two jets.

Figure 4

Example weight distributions for two different $t\overline{t}\to e\mu$ Monte Carlo events obtained with $\nu \mathrm{WT}$ and MWT methods. The generator level mass is $m_t=170\mathrm{GeV}$.

Figure 5

(a) Correlation between the mean of the $\nu \mathrm{WT}$ weight distribution and the input $m_t$. (b) Correlation between $\nu \mathrm{WT}$ ${\mu }_w$ and ${\sigma }_w$ for the $e\mu$ channel. Three test masses of 155 GeV, 180 GeV, and 200 GeV are shown.

Figure 6

An example of a template for the MWT method.

Figure 7

Slices of probability density histograms $h_s$ and fit functions $f_s$ for the $\nu \mathrm{WT}$ method in the $e\mu$ channel. Probability densities vs ${\sigma }_w$ and ${\mu }_w$ for $m_t=170\mathrm{GeV}$ are shown for (a) $h_s$ and (b) $f_s$. Probability densities vs $m_t$ and ${\mu }_w$ for ${\sigma }_w=30\mathrm{GeV}$ are shown for (c) $h_s$ and (d) $f_s$.

Figure 8

Probability density histogram $h_b$ and fit function $f_s$ for the $\nu \mathrm{WT}$ method in the $e\mu$ channel. Probability densities vs ${\sigma }_w$ and ${\mu }_w$ are shown for (a) $h_b$ and (b) $f_b$.

Figure 9

The combined calibration curves corresponding to the (a) $\nu {\mathrm{WT}}_h$, (b) $\nu {\mathrm{WT}}_f$, and (c) MWT methods. Overlaid is the result of the linear fit as defined in Eq. (20). The uncertainties are small and corresponding bars are hidden by the markers.

Figure 10

Distribution of statistical uncertainties for top quark mass measurements of pseudoexperiments for the combination of all channels for simulated events with $m_t=170\mathrm{GeV}$ for the (a) $\nu {\mathrm{WT}}_h$, (b) $\nu {\mathrm{WT}}_f$, and (c) MWT methods. The uncertainties are corrected by the calibration curve and for the pull width. The arrows indicate the statistical uncertainties for the measured top quark mass.

Figure 11

Negative log-likelihood $-\ln \mathcal{L}$ vs $m_t$ for the combination of all channels before calibration for the (a) $\nu {\mathrm{WT}}_h$, (b) $\nu {\mathrm{WT}}_f$, and (c) MWT methods.