Sensitivity to an invisibly decaying Higgs boson

This page contains approved plots and results in the order as they appear in the CSC note. Only the CSC note contains all the relevant information and should thus be consulted if one of the plots is used.

Figure 1: Feynman diagram of the VBF process. The V represents either a Z or W boson.

Figure 2: Comparison of tagged jet properties for signal events (m_H = 130 GeV) and the three major backgrounds. The upper left plot shows the p_T of the leading tagged jet, the upper right plot shows the p_T of the jet with the second highest p_T. The lower left plot shows the product of the directions of the two tagged jets in pseudo-rapidity (η_j1 η_j2 ), and the lower right plot shows the difference in η between the two tagged jets (Δη). The enhancement at Δη of 0.5 in the VBF signal results from a single high p_T jet being reconstructed as two jets. The filter cut described in Section 2.1 has been applied to the W+jet and Z+jet Monte-Carlo data, but no trigger cuts have been applied. The distributions are normalized to unity. The vertical dotted lines show the cut values used in the analysis.

Figure 3: The reconstructed invariant mass of the tagging jets (left) and the E_T^miss(right) for the invisible Higgs boson signal (m_H = 130 GeV) and the three main backgrounds. Single events in the high E_T tail of each individual sample (J0, J1 etc) can result in a spike with a large error. The filter cut described in Section 2.1 has been applied to the W+jet and Z+jet Monte-Carlo data, but no trigger cuts have been applied. The distributions are normalized to unity.

Figure 4: The distribution of the reconstructed E_T isolation variable (I) is shown in the right hand plot and the azimuthal angle between the tagging jets (φjj ) is shown in the right hand plot for the invisible Higgs boson signal (m_H=130 GeV) and the three main backgrounds. The filter cut described in Section 2.1 has been applied to the W+jet and Z+jet Monte-Carlo data, but no trigger cuts have been applied. The distributions are normalized to unity.

Figure 5: The φ_jj distribution for the signal and background in radians. The solid and dash-dotted lines represent the expected distribution for the Higgs boson signal with Higgs boson masses of m_H = 120 and m_H = 300 GeV respectively. The dotted and dashed lines represent the distributions expected from the backgrounds. The plot shows the distributions after VBF selection cuts have been made. Note that for the φ_jj plot shown earlier did not have these cuts applied, (Figure 4).

Figure 6: Sensitivity for an invisible Higgs boson at 95% C.L. via the VBF channel using shape analysis for an integrated luminosity of 30 fb^-1 with and without systematic uncertainties. The black triangles (circles) are the results from this analysis with (without) systematic uncertainties.

Figure 7: (Left): The Feynman diagram for Higgs boson associated production with a Z boson. (Right): A representation of the decay of a Higgs boson into two invisible neutralinos represented by χ⁰ recoiling against the two leptons coming from the Z decay.

Figure 8: Input variables used by the Boosted Decision Tree for the signal with m_H=130 GeV and the main backgrounds. Top left: Missing E_T. Top right: transverse mass, defined as the reconstructed mass in the transverse plane, namely m_T² = E_T² - p_T². Bottom left: cosine of the angle between the missing E_T and highest momentum lepton in the transverse plane. Bottom right: reconstructed Z mass. Each plot has been normalized to unity. The combined samples had first been scaled to the same luminosity.

Figure 9: Input variables used by the Boosted Decision Tree for the signal with m_H = 130 GeV and the main backgrounds. Top left: Transverse momentum of the most energetic lepton. Top right: Transverse momentum of the second lepton. Bottom left: Cosine of the angle between the two leptons in the transverse plane and, Bottom right: in 3-dimensions. Each plot has been normalized to unity. The combined samples had first been scaled to the same luminosity.

Figure 10: Input variables used by the Boosted Decision Tree for the signal with m_H = 130 GeV and the main backgrounds. Top left: The cosine of the angle between the direction of missing E_T and the reconstructed Z transverse momentum. Top right: The cosine of the angle between the most energetic jet and the direction of missing E_T. Bottom left: The energy contained in a cone of 0.10 rad around the most energetic lepton. Bottom right: The energy contained in a cone of 0.10 rad around the second lepton. Each plot has been normalized to unity. The combined samples had first been scaled to the same luminosity.

Figure 11: Input variables used by the Boosted Decision Tree for the signal with m_H = 130 GeV and the main backgrounds. Top left: The energy distribution for the most energetic jet; Top right: for the second and, Bottom left: third most energetic jets. Bottom right: the number of jets in the event. Each plot has been normalized to unity. The combined samples had first been scaled to the same luminosity.

Figure 12: The Boosted Decision Tree (BDT) output variables obtained after comparing half the signal events to five different backgrounds separately, namely, from top to bottom: tt->blνbl&nu, WW->lνlν, ZZ->llνν, ZW->lllν and Z->ll+jets. The BDT assigns values close to +1 for a signal-like event and -1 for background-like events. The distributions are shown for the signal and all types of background when using the other half of the events for the analysis. The Boosted Decision Trees trained against the ZW background offers the best separation power. The ξ² decreases further once additional cuts on the other BDT output variables are applied, namely the WW BDT output, then the ZZ->ττ BDT output, the (Z+jet) BDT output and finally the tt BDT output. All BDT output cut values are indicated by a vertical dashed line. Nothing is gained from training a BDT against the irreducible background, ZZ->llνν for all Higgs boson mass hypotheses, so it is not used.

Figure 13: Sensitivity to an invisible Higgs boson with ATLAS for both the VBF and ZH channels with 30 fb^-1 of data assuming only Standard Model backgrounds. The open crosses show the sensitivity for the ZH analysis and the solid triangles show the sensitivity for the VBF shape analysis for 95% CL. Both these results include systematic uncertainties.