Evidence for a mass dependent forward-backward asymmetry in top quark pair production

We present a new measurement of the inclusive forward-backward $t\overline{t}$ production asymmetry and its rapidity and mass dependence. The measurements are performed with data corresponding to an integrated luminosity of $5.3{\mathrm{fb}}^{-1}$ of $p\overline{p}$ collisions at $\sqrt{s}=1.96\mathrm{TeV}$, recorded with the CDF-II Detector at the Fermilab Tevatron. Significant inclusive asymmetries are observed in both the laboratory frame and the $t\overline{t}$ rest frame, and in both cases are found to be consistent with $CP$ conservation under interchange of $t$ and $\overline{t}$. In the $t\overline{t}$ rest frame, the asymmetry is observed to increase with the $t\overline{t}$ rapidity difference, $\Delta y$, and with the invariant mass $M_{t\overline{t}}$ of the $t\overline{t}$ system. Fully corrected parton-level asymmetries are derived in two regions of each variable, and the asymmetry is found to be most significant at large $\Delta y$ and $M_{t\overline{t}}$. For $M_{t\overline{t}}\ge 450\mathrm{GeV}/c^2$, the parton-level asymmetry in the $t\overline{t}$ rest frame is $A^{t\overline{t}}=0.475\pm 0.114$ compared to a next-to-leading order QCD prediction of $0.088\pm 0.013$.

Figure 1

Interfering $q\overline{q}\to t\overline{t}$ (above) and $q\overline{q}\to t\overline{t}j$ (below) amplitudes.

Figure 2

Rapidity distributions in data compared to predictions. The legend “$t\overline{t}+\mathrm{bkg}$” implies totals in those bins are the sum of the $t\overline{t}$ and background components. The asymmetries in the data and the predicted $t\overline{t}$ signal, background, and combination are shown in legends on top right of plots, using the conventions of Table 2.

Figure 3

Distributions of $\Delta y_{lh}=y_l\mathrm{\text{−}}{y}_h$ (top) and $y_h$ (bottom) for events with negative leptons (left) and positive leptons (right).

Figure 4

$\Delta y$ and $-q{y}_h=y_t^{p\overline{p}}$ distributions in data vs prediction.

Figure 5

Four-bin representation of rapidity distributions for all correction levels. Solid histogram is the pythia $t\overline{t}$ model.

Figure 6

$\Delta y$-dependence of $A^{t\overline{t}}$ according to mcfm.

Figure 7

Parton-level asymmetries at small and large $\Delta y$ compared to SM prediction of mcfm. The shaded bands represent the total uncertainty in each bin.

Figure 8

Parton-level $M_{t\overline{t}}$-dependence of $A^{t\overline{t}}$ according to mcfm.

Figure 9

Event distribution as a function of the total invariant mass $M_{t\overline{t}}$.

Figure 10

Left: The $\Delta y\mathrm{\text{−}}{M}_{t\overline{t}}$ plane. Each dot represents one event, while the intensity of the shading shows approximately the event probability in the standard pythia based prediction. Right: The $t\overline{t}$ frame asymmetry in the data in bins of invariant mass $M_{t\overline{t}}$, compared to the prediction of mc@nlo $t\overline{t}+\mathrm{\text{backgrounds}}$. The last bin includes all events with $M_{t\overline{t}}\ge 700\mathrm{GeV}/c^2$.

Figure 11

The $\Delta y_{lh}$ asymmetries in bins of invariant mass $M_{t\overline{t}}$ when the data is partitioned by lepton charge.

Figure 12

Top: The distribution of $\Delta y$ at low mass (left) and high-mass (right).

Figure 13

The distribution of $\Delta y_{lh}$ at high-mass for events with negative leptons (left) and positive leptons (right).

Figure 14

Parton-level asymmetry in $\Delta y$ at high and low mass compared to mcfm prediction. The shaded region represents the total uncertainty in each bin.

Figure 15

Distribution of $t\overline{t}$ reconstruction ${\chi }^2$. Black crosses are data. The histogram is $\mathrm{sig}+\mathrm{bkg}$ prediction: blue is background, green is pythia stacked on background. The last bin on the right contains all events with ${\chi }^2>100$.

Figure 16

The distributions of $M_{t\overline{t}}$ (left) and $\Delta y$ (right) in the OctetA sample compared to the $\mathrm{PYTHIA}+\mathrm{\text{background}}$ prediction.

Figure 17

The reconstructed $M_{t\overline{t}}$ dependence of $A^{t\overline{t}}$ in the color-octet models. Left: OctetA. Right: OctetB.

Figure 18

Top: The distribution of $\Delta y$ at low mass (left) and high-mass (right). The pythia baseline model is the filled histogram and the Octet samples are the points.