CERN Accelerating science

A diagram showing the \higgsd decay used as the benchmark model. The $s$ inherits Yukawa couplings to SM fermions from the $\Phi$, and therefore decays primarily to heavy quarks.

:

: : Trigger efficiency for simulated signal events as a function of the LLP \pt\ for (a) one of the low-\ET signal samples and (b) one of the high-\ET signal samples for HLT CalRatio triggers seeded by the high-\et L1 triggers with \ET thresholds of 60~\GeV and 100~\GeV and by the two versions of the low-\et L1 triggers. Only statistical uncertainties are shown.

:

: : Trigger efficiency for simulated signal events as a function of the LLP decay position in the $x$--$y$ plane ($L_{xy}$) for LLPs decaying in the barrel ($|\eta|<1.4$) for (a) one of the low-\ET signal samples and (b) one of the high-\ET signal samples for HLT CalRatio triggers seeded by the high-\et L1 triggers with \ET thresholds of 60~\GeV and 100~\GeV and by the two versions of the low-\et L1 triggers. Only statistical uncertainties are shown.

A simplified diagram of how the adversarial network is structured. The principal network which discriminates between signal, SM multijets and BIB is shown with the inputs moving through the network as black arrows. Dense layers refer to fully-connected layers, while LSTM refers to Long-Short Term Memory layers described in the text. The adversary which classifies jets as being MC simulation or data is shown with the inputs moving through the network as orange arrows. The input variables are labelled $n\times m$ where $n$ is the maximum of the number of tracks, topoclusters or muon segments per jet and $m$ is the number of distinct variables per input object.

:

:

: : The NN output scores in the dijet control region for (a) the low-\ET training with no adversary network and (b) with an adversary network included, (c) for the high-\ET training with no adversary network and (d) with an adversary network included. Statistical uncertainties are shown in all plots. In cases where training with adversary networks is considered, the systematic uncertainty related to modelling discrepancies is included as well.

: Caption not extracted

:

:

:

: : The NN output scores for (a,b) signal, (c,d) BIB and (e,f) SM multijets shown for low-\ET and high-\ET training. The jets used here are part of the validation set, and follow the selection used when choosing jets to train the NN in signal, multijets and BIB.

: Caption not extracted

: Caption not extracted

:

: : Distribution of the (a) low-\et per-event BDT and (b) high-\et per-event BDT outputs in main data, BIB data and some of the benchmark signal samples after preselection. Only statistical uncertainties are shown.

:

:

:

: : The distributions of $\sumMinDR$(jet, tracks) vs \evbdtLITau for (a) BIB events, (b) main data and (c) a signal sample after event cleaning for the high-\ET selection. Panels (d,e,f) show the equivalent distributions for the low-\ET selection. The signal sample with $\mPhi = 600 \GeV$ and $\mS = 150 \GeV$ is shown for the high-$\ET$ selection, while the sample with $\mPhi = 200 \GeV$ and $\mS = 50 \GeV$ is shown for the low-$\ET$ selection.

: Caption not extracted

: Caption not extracted

:

:

:

: : (a) The per-event BDT vs $\sumMinDR$(jet, tracks) distribution for the main dataset in the $\textrm{VRCD}_\textrm{high-\ET}$ validation region. Region A in the nominal high-\ET selection is shown for reference. Also shown is a comparison of the observed and expected events in region A of this VR. The $x$-axis shows the different (b) $\sumMinDR$(jet, tracks) and (c) BDT boundaries used to test the ABCD method. Statistical uncertainties are shown in (b) and (c). Panels (d,e,f) show the equivalent distributions in the $\textrm{VRCD}_\textrm{low-\ET}$ validation region.

: Caption not extracted

: Caption not extracted

:

: : (a) 95\% CL expected and observed limits on the BR of SM Higgs bosons to pairs of neutral LLPs ($B_{H\rightarrow ss}$), showing the $\pm 1 \sigma$ (green) and $\pm 2 \sigma$ (yellow) expected limit bands, as well as a comparison with the results from previous ATLAS searches~\cite{EXOT-2018-61, EXOT-2017-25}, and (b) summary of 95\% CL expected and observed limits on the BR of a SM Higgs boson mediator to pairs of neutral LLPs considered in this analysis. The cross-section for SM Higgs boson gluon--gluon fusion production is assumed to be $48.6$~pb.

:

:

: : Summary of 95\% CL expected and observed limits for the hidden sector (HS) models considered in this analysis, where the mediator is not the SM Higgs boson: (a) shows the case where $\mH=60\gev$; (b) shows intermediate masses of $\mH=200\gev$ and $\mH=400\gev$; (c) shows the case where $\mH=600\gev$; (d) shows the highest \mH hypothesis with $\mH=1\tev$.

: Caption not extracted