CERN Accelerating science

{\bf Left}: A schematic view of the muon spectrometer with lines indicating various pseudo rapidity regions~\cite{bib:atltriperf2015}. The {\it curved arrows} shows an example of a trajectory from slow particles generated at the beam pipe around $z\sim 10$m. Triggers due to events of this type are mitigated by requiring an additional coincidence with the TGC-FI chambers in the region $1.3<|\eta|<1.9$*. {\bf Right}: Number of events with an L1 muon trigger with transverse momentum ($p_T$) above 15 GeV (L1\_MU15) as a function of the muon $\eta$ coordinate, when a coincidence with the TGC-FI chambers is required (upper histogram) or no requirement is applied (lower histogram)~\cite{bib:atltriperf2015}*.

Luminosity-weighted distribution of the mean number of interactions per bunch crossing for the 2015-2018 pp collision data at 13 TeV centre-of-mass energy. The mean number of interactions per bunch crossing corresponds to the mean of the poisson distribution of the number of interactions per crossing calculated for each bunch. It is calculated from the instantaneous per bunch luminosity as $\mu=L_{bunch}\times\sigma_{intel} / f_r$, where $L_{bunch}$ is the per bunch instantaneous luminosity, $\sigma_{inel}$ is the inelastic cross section which was taken to be 80 mb for 13 TeV collisions, and $f_r$ is the LHC revolution frequency~\cite{bib:trigpub}*.

{\bf Left}: A schematic view of the muon spectrometer with lines indicating various pseudo rapidity regions~\cite{bib:atltriperf2015}. The {\it curved arrows} shows an example of a trajectory from slow particles generated at the beam pipe around $z\sim 10$m. Triggers due to events of this type are mitigated by requiring an additional coincidence with the TGC-FI chambers in the region $1.3<|\eta|<1.9$*. {\bf Right}: Number of events with an L1 muon trigger with transverse momentum ($p_T$) above 15 GeV (L1\_MU15) as a function of the muon $\eta$ coordinate, when a coincidence with the TGC-FI chambers is required (upper histogram) or no requirement is applied (lower histogram)~\cite{bib:atltriperf2015}*.

The track finding efficiency of the Inner Detector (ID) trigger for tracks with $p_T > 1$ GeV within jets shown as a function of the multiplicity of the mean number of pileup interactions in the event. For the jet and Bjet triggers the reconstruction in the ID trigger runs in three stages~\cite{bib:trigpub}.*

{\bf Left}: A schematic view of the muon spectrometer with lines indicating various pseudo rapidity regions~\cite{bib:atltriperf2015}. The {\it curved arrows} shows an example of a trajectory from slow particles generated at the beam pipe around $z\sim 10$m. Triggers due to events of this type are mitigated by requiring an additional coincidence with the TGC-FI chambers in the region $1.3<|\eta|<1.9$*. {\bf Right}: Number of events with an L1 muon trigger with transverse momentum ($p_T$) above 15 GeV (L1\_MU15) as a function of the muon $\eta$ coordinate, when a coincidence with the TGC-FI chambers is required (upper histogram) or no requirement is applied (lower histogram)~\cite{bib:atltriperf2015}*.

The per-bunch trigger rate for the L1 missing transverse momentum trigger with a threshold of 50 GeV (L1\_XE50) as a function of the instantaneous luminosity per bunch~\cite{bib:atltriperf2015}. The rates are shown with and without the pedestal correction applied*.

The trigger track transverse impact parameter resolution of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the offline muon $p_T$. This resolution is evaluated for both a 10 GeV and a 24 GeV muon triggers running in a mode where the trigger decision is made based on early muon candidates reconstructed from the Muon Spectrometer information only and so can contain candidates where the full offline reconstructed muons have a $p_T$ lower than the trigger threshold. The ID trigger first runs a Fast Track Finder stage followed by a detailed Precision Tracking stage to refine the track candidates identified in the first stage and improve their quality~\cite{bib:trigpub}*.

HLT trigger rates grouped by trigger signature during an LHC fill in July 2016 with a peak luminosity of $1.02\times10^{34}$ cm$^{-2}$ s$^{-1}$~\cite{bib:trigpub}. Due to overlaps the sum of the individual groups is higher than the Main physics stream rate, which is shown as a black line. Multi-object triggers are included in the b-jets and tau groups. The B-physics triggers are mainly muon-based triggers. The combined group includes multiple triggers combining different trigger signatures such as electrons with muons, taus, jets or $E_T^{miss}$. Common features to all rates are their exponential decay with decreasing luminosity during an LHC fill. The rates periodically increase due to change of prescales to optimize the bandwidth usage, dips are due to deadtime, and spikes are caused by detector noise*.

The ATLAS TDAQ system in Run-2 with emphasis on the components relevant for triggering~\cite{bib:atltriperf2015}*.

{\bf Top}: Output rates of the single-electron and di-electron primary triggers as a function of the un-calibrated instantaneous luminosity measured online during the 2016 proton-proton data taking at a center-of-mass energy of 13 TeV~\cite{bib:trigpub}*. {\bf Bottom}: Rate (in Hz) of the isolated single electron trigger as a function of the $E_T$ threshold at the high-level trigger (HLT) in the [26,72] GeV range, for the same likelihood-based tight identification and Level-1 selections. The rate is measured in a dataset collected at a constant instantaneous luminosity of $8\times10^{33}$ cm$^{-2}$ s$^{-1}$ at $\sqrt{s} = 13$ TeV, while the contribution from W, Z and multi-jet production is estimated with Monte Carlo. The dominant uncertainty on the multi-jet rate is evaluated with a data-driven technique~\cite{bib:trigpub}*.

Luminosity-weighted distribution of the mean number of interactions per bunch crossing for the 2015-2018 pp collision data at 13 TeV centre-of-mass energy. The mean number of interactions per bunch crossing corresponds to the mean of the poisson distribution of the number of interactions per crossing calculated for each bunch. It is calculated from the instantaneous per bunch luminosity as $\mu=L_{bunch}\times\sigma_{intel} / f_r$, where $L_{bunch}$ is the per bunch instantaneous luminosity, $\sigma_{inel}$ is the inelastic cross section which was taken to be 80 mb for 13 TeV collisions, and $f_r$ is the LHC revolution frequency~\cite{bib:trigpub}*.

{\bf Top}: Output rates of the single-electron and di-electron primary triggers as a function of the un-calibrated instantaneous luminosity measured online during the 2016 proton-proton data taking at a center-of-mass energy of 13 TeV~\cite{bib:trigpub}*. {\bf Bottom}: Rate (in Hz) of the isolated single electron trigger as a function of the $E_T$ threshold at the high-level trigger (HLT) in the [26,72] GeV range, for the same likelihood-based tight identification and Level-1 selections. The rate is measured in a dataset collected at a constant instantaneous luminosity of $8\times10^{33}$ cm$^{-2}$ s$^{-1}$ at $\sqrt{s} = 13$ TeV, while the contribution from W, Z and multi-jet production is estimated with Monte Carlo. The dominant uncertainty on the multi-jet rate is evaluated with a data-driven technique~\cite{bib:trigpub}*.

{\bf Left}: Mean HLT processing time as a function of the instantaneous luminosity. The peak at low luminosity is due to special B-physics triggers that are activated when the luminosity drops at the end of a fill~\cite{bib:atltriperf2015}*. {\bf Right}: Distribution of an HLT processing time per event for an instantaneous luminosity of $5.2\times10^{33}$ cm$^{-2}$ s$^{-1}$ and average pile-up $<\mu>=15$~\cite{bib:atltriperf2015}*.

Comparison of the likelihood-base and the cut-base HLT electron triggers efficiency as a function of the offline electron candidate's transverse energy $E_T$ with respect to true reconstructed electrons in $Z\rightarrow ee$ simulation. The HLT\_e24\_medium\_iloose\_L1EM18VH trigger is the Run-1 algorithm requiring an electron candidate with $E_T > 24$ GeV satisfying the cut-based medium identification, while HLT\_e24\_lhmedium\_iloose\_L1EM18VH corresponds to the Run-2 algorithm using the likelihood-based lhmedium electron identification. Both trigger chains also require the same track isolation selection and are seeded by the same level-1 trigger (L1\_EM18VH)~\cite{bib:trigpub}*.

{\bf Left}: Mean HLT processing time as a function of the instantaneous luminosity. The peak at low luminosity is due to special B-physics triggers that are activated when the luminosity drops at the end of a fill~\cite{bib:atltriperf2015}*. {\bf Right}: Distribution of an HLT processing time per event for an instantaneous luminosity of $5.2\times10^{33}$ cm$^{-2}$ s$^{-1}$ and average pile-up $<\mu>=15$~\cite{bib:atltriperf2015}*.

{\bf Left}: Mean HLT processing time as a function of the instantaneous luminosity. The peak at low luminosity is due to special B-physics triggers that are activated when the luminosity drops at the end of a fill~\cite{bib:atltriperf2015}*. {\bf Right}: Distribution of an HLT processing time per event for an instantaneous luminosity of $5.2\times10^{33}$ cm$^{-2}$ s$^{-1}$ and average pile-up $<\mu>=15$~\cite{bib:atltriperf2015}*.

The track finding efficiency of the Inner Detector (ID) trigger for tracks with $p_T > 1$ GeV within jets shown as a function of: {\bf Left} the offline track $p_T$; and {\bf Right} the multiplicity of the mean number of pileup interactions in the event. For the jet and Bjet triggers the reconstruction in the ID trigger runs in three stages~\cite{bib:trigpub}.*

Efficiency of the L1\_EM20VHI trigger (circles) as well as the combined L1\_EM20VHI and HLT\_e24\_lhtight\_nod0\_ivarloose trigger (blue triangles) as a function of the offline electron candidate's transverse energy ($E_T$). A variable-size cone isolation criteria is applied ("ivarloose"). The HLT trigger requires an electron candidate with $E_T > 24$ GeV satisfying the likelihood-based tight identification. The offline reconstructed electron is required to pass a likelihood-based tight identification~\cite{bib:trigpub}*.

{\bf Left}: Mean HLT processing time as a function of the instantaneous luminosity. The peak at low luminosity is due to special B-physics triggers that are activated when the luminosity drops at the end of a fill~\cite{bib:atltriperf2015}*. {\bf Right}: Distribution of an HLT processing time per event for an instantaneous luminosity of $5.2\times10^{33}$ cm$^{-2}$ s$^{-1}$ and average pile-up $<\mu>=15$~\cite{bib:atltriperf2015}*.

The track finding efficiency of the Inner Detector (ID) trigger for tracks with $p_T > 1$ GeV within jets shown as a function of: {\bf Left} the offline track $p_T$; and {\bf Right} the multiplicity of the mean number of pileup interactions in the event. For the jet and Bjet triggers the reconstruction in the ID trigger runs in three stages~\cite{bib:trigpub}.*

Efficiency of the HLT likelihood-base electron trigger as a function of the offline electron candidate's transverse energy ($E_T$) as measured with the tab and probe method on a sample of 2017 ATLAS data as well as on a $Z\rightarrow ee$ Monte Carlo sample~\cite{bib:trigpub}*. The data-to-MC agreement is very good. The efficiency is calculated for two different trigger chain: an electron candidate with $E_T > 24$ GeV satisfying the likelihood-based very loose identification and seeded by a 20 GeV electron L1 trigger ({\bf Top}); and an electron candidate with $E_T > 26$ GeV satisfying the likelihood-based tight identification and seeded by a 22 GeV electron L1 trigger ({\bf Bottom}).

Efficiency of the HLT likelihood-base electron trigger as a function of the offline electron candidate's transverse energy ($E_T$) as measured with the tab and probe method on a sample of 2017 ATLAS data as well as on a $Z\rightarrow ee$ Monte Carlo sample~\cite{bib:trigpub}*. The data-to-MC agreement is very good. The efficiency is calculated for two different trigger chain: an electron candidate with $E_T > 24$ GeV satisfying the likelihood-based very loose identification and seeded by a 20 GeV electron L1 trigger ({\bf Top}); and an electron candidate with $E_T > 26$ GeV satisfying the likelihood-based tight identification and seeded by a 22 GeV electron L1 trigger ({\bf Bottom}).

{\bf Left}: The track finding efficiency of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the mean number of interactions per bunch crossing*. {\bf Right}: The trigger track transverse impact parameter resolution of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the offline muon $p_T$. Both the efficiency and the resolution are evaluated for a 10 GeV and a 24 GeV muon triggers running in a mode where the trigger decision is made based on early muon candidates reconstructed from the Muon Spectrometer information only and so can contain candidates where the full offline reconstructed muons have a $p_T$ lower than the trigger threshold. The ID trigger first runs a Fast Track Finder stage followed by a detailed Precision Tracking stage to refine the track candidates identified in the first stage and improve their quality~\cite{bib:trigpub}*.

{\bf Left}: The track finding efficiency of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the mean number of interactions per bunch crossing*. {\bf Right}: The trigger track transverse impact parameter resolution of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the offline muon $p_T$. Both the efficiency and the resolution are evaluated for a 10 GeV and a 24 GeV muon triggers running in a mode where the trigger decision is made based on early muon candidates reconstructed from the Muon Spectrometer information only and so can contain candidates where the full offline reconstructed muons have a $p_T$ lower than the trigger threshold. The ID trigger first runs a Fast Track Finder stage followed by a detailed Precision Tracking stage to refine the track candidates identified in the first stage and improve their quality~\cite{bib:trigpub}*.

Absolute efficiency of Level 1 (L1) MU20 trigger and absolute and relative efficiencies of the OR of mu26\_ivarmedium with mu50 High Level Triggers (HLT) plotted as a function of $\phi$ of offline muon candidates in the barrel detector region. The efficiency is computed with respect to offline isolated muon candidates which are reconstructed using standard ATLAS software and are required to pass "Medium" quality requirement. The selection is restricted to the plateau region with $p_T > 27$ GeV~\cite{bib:trigpub}*.

{\bf Left}: The distributions of processing times for the topo-clustering algorithm executed on the full calorimeter~\cite{bib:atltriperf2015}*. {\bf Right}: The transverse energy $E_T$ resolution for online (HLT) topo-clusters with respect to offline topo-clusters with $E_T > 3$ GeV~\cite{bib:trigpub}. The online and offline clusters are both hadronically calibrated. The improvement in the $E_T$ resolution for online topo-clusters with respect to offline topo-clusters in 2016 is due to the introduction cell-level, BCID/$<\mu>$ based energy corrections. These corrections account for pedestal changes that arise due to out-of-time pile-up and particularly affect the first bunch crossings in each bunch train. Similar corrections were already being applied offline*.

Transverse momentum distribution for the positron coming from inclusive $W^+$ decays. The data are compared to the simulation including signal and background contributions. Detector calibration and physics-modelling corrections are applied to the simulated events.The lower panels show the data-to-prediction ratios, the error bars show the statistical uncertainty, and the band shows the systematic uncertainty of the prediction. The $\chi^2$ values displayed in each figure account for all sources of uncertainty and include the effects of bin-to-bin correlations induced by the systematic uncertainties~\cite{bib:wmass}*.

Absolute efficiency of Level 1 (L1) MU20 trigger and absolute and relative efficiencies of the OR of mu26\_ivarmedium with mu50 High Level Triggers (HLT) plotted as a function of $p_T$ of offline muon candidates in the barrel detector region ({\bf Top} ), and the endcap detector region ({\bf Bottom} )~\cite{bib:trigpub}*. The efficiency is computed exactly like described in the caption of Fig.~\ref{fig:muonphi}.

{\bf Left}: The distributions of processing times for the topo-clustering algorithm executed on the full calorimeter~\cite{bib:atltriperf2015}*. {\bf Right}: The transverse energy $E_T$ resolution for online (HLT) topo-clusters with respect to offline topo-clusters with $E_T > 3$ GeV~\cite{bib:trigpub}. The online and offline clusters are both hadronically calibrated. The improvement in the $E_T$ resolution for online topo-clusters with respect to offline topo-clusters in 2016 is due to the introduction cell-level, BCID/$<\mu>$ based energy corrections. These corrections account for pedestal changes that arise due to out-of-time pile-up and particularly affect the first bunch crossings in each bunch train. Similar corrections were already being applied offline*.

Absolute efficiency of Level 1 (L1) MU20 trigger and absolute and relative efficiencies of the OR of mu26\_ivarmedium with mu50 High Level Triggers (HLT) plotted as a function of $p_T$ of offline muon candidates in the barrel detector region ({\bf Top} ), and the endcap detector region ({\bf Bottom} )~\cite{bib:trigpub}*. The efficiency is computed exactly like described in the caption of Fig.~\ref{fig:muonphi}.

HLT trigger rates grouped by trigger signature during an LHC fill in July 2016 with a peak luminosity of $1.02\times10^{34}$ cm$^{-2}$ s$^{-1}$~\cite{bib:trigpub}. Due to overlaps the sum of the individual groups is higher than the Main physics stream rate, which is shown as a black line. Multi-object triggers are included in the b-jets and tau groups. The B-physics triggers are mainly muon-based triggers. The combined group includes multiple triggers combining different trigger signatures such as electrons with muons, taus, jets or $E_T^{miss}$. Common features to all rates are their exponential decay with decreasing luminosity during an LHC fill. The rates periodically increase due to change of prescales to optimize the bandwidth usage, dips are due to deadtime, and spikes are caused by detector noise*.

Efficiencies are shown for a single-jet trigger with three different calibrations~\cite{bib:jetcalib} applied to jets in the ATLAS high-level trigger (HLT)~\cite{bib:trigpub}*.

{\bf Top}: Efficiencies for HLT single-jet triggers as a function of leading offline jet $p_{T}$. Triggers denoted HLT\_jX accept an event if a jet is reconstructed at HLT with $E_T > X$ GeV*. The unprescaled trigger with the lowest threshold requires a jet with $E_\mathrm{T} > 380$ GeV~\cite{bib:trigpub}. {\bf Bottom}: Efficiencies for HLT large-R single-jet triggers as a function of the leading offline trimmed~\cite{bib:trim} jet $p_\mathrm{T}$. Blue circles represent a trimmed large-R jet trigger with a $p_\mathrm{T}$ threshold of 420 GeV. Adding an additional 30 GeV cut on the jet mass of the selected trimmed trigger jet is shown in green triangles. The mass cut significantly suppresses the QCD di-jet background, allowing a lower $p_\mathrm{T}$ threshold of 390 GeV, while retaining nearly all signal-like jets with a mass of above 50 GeV*.

Jet $p_T$ spectrum after the basic kinematic selection for the TLA trigger jets compared to trigger jets recorded by all single-jet triggers~\cite{bib:atltriperf2015}*.

The track finding efficiency of the Inner Detector (ID) trigger for tracks with $p_T > 1$ GeV within jets shown as a function of: {\bf Left} the offline track $p_T$; and {\bf Right} the multiplicity of the mean number of pileup interactions in the event. For the jet and Bjet triggers the reconstruction in the ID trigger runs in three stages~\cite{bib:trigpub}.*

{\bf Left}: Output rates of the single-electron and di-electron primary triggers as a function of the un-calibrated instantaneous luminosity measured online during the 2016 proton-proton data taking at a center-of-mass energy of 13 TeV~\cite{bib:trigpub}*. {\bf Right}: Rate (in Hz) of the isolated single electron trigger as a function of the $E_T$ threshold at the high-level trigger (HLT) in the [26,72] GeV range, for the same likelihood-based tight identification and Level-1 selections. The rate is measured in a dataset collected at a constant instantaneous luminosity of $8\times10^{33}$ cm$^{-2}$ s$^{-1}$ at $\sqrt{s} = 13$ TeV, while the contribution from W, Z and multi-jet production is estimated with Monte Carlo. The dominant uncertainty on the multi-jet rate is evaluated with a data-driven technique~\cite{bib:trigpub}*.

{\bf Top}: Efficiencies for HLT single-jet triggers as a function of leading offline jet $p_{T}$. Triggers denoted HLT\_jX accept an event if a jet is reconstructed at HLT with $E_T > X$ GeV*. The unprescaled trigger with the lowest threshold requires a jet with $E_\mathrm{T} > 380$ GeV~\cite{bib:trigpub}. {\bf Bottom}: Efficiencies for HLT large-R single-jet triggers as a function of the leading offline trimmed~\cite{bib:trim} jet $p_\mathrm{T}$. Blue circles represent a trimmed large-R jet trigger with a $p_\mathrm{T}$ threshold of 420 GeV. Adding an additional 30 GeV cut on the jet mass of the selected trimmed trigger jet is shown in green triangles. The mass cut significantly suppresses the QCD di-jet background, allowing a lower $p_\mathrm{T}$ threshold of 390 GeV, while retaining nearly all signal-like jets with a mass of above 50 GeV*.

{\bf Left}: Output rates of the single-electron and di-electron primary triggers as a function of the un-calibrated instantaneous luminosity measured online during the 2016 proton-proton data taking at a center-of-mass energy of 13 TeV~\cite{bib:trigpub}*. {\bf Right}: Rate (in Hz) of the isolated single electron trigger as a function of the $E_T$ threshold at the high-level trigger (HLT) in the [26,72] GeV range, for the same likelihood-based tight identification and Level-1 selections. The rate is measured in a dataset collected at a constant instantaneous luminosity of $8\times10^{33}$ cm$^{-2}$ s$^{-1}$ at $\sqrt{s} = 13$ TeV, while the contribution from W, Z and multi-jet production is estimated with Monte Carlo. The dominant uncertainty on the multi-jet rate is evaluated with a data-driven technique~\cite{bib:trigpub}*.

{\bf Left}: The track finding efficiency of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the mean number of interactions per bunch crossing*. {\bf Right}: The trigger track transverse impact parameter resolution of the Inner Detector (ID) trigger for muons with $p_T > 4$ GeV from medium quality offline muon candidates, shown as a function of the offline muon $p_T$. Both the efficiency and the resolution are evaluated for a 10 GeV and a 24 GeV muon triggers running in a mode where the trigger decision is made based on early muon candidates reconstructed from the Muon Spectrometer information only and so can contain candidates where the full offline reconstructed muons have a $p_T$ lower than the trigger threshold. The ID trigger first runs a Fast Track Finder stage followed by a detailed Precision Tracking stage to refine the track candidates identified in the first stage and improve their quality~\cite{bib:trigpub}*.

{\bf Top}: Tau trigger efficiency measured in data and compared to simulation, with respect to offline reconstructed tau candidate with one or three tracks and passing the offline medium identification criteria, as function of the offline transverse momentum. The trigger efficiency is measured in a tag and probe analysis with $Z\rightarrow\tau\tau\rightarrow\mu\tau_{had}$ event from the 2016 dataset in 13TeV collision (8.0 fb$^{-1}$)*. {\bf Bottom} Comparison of this HLT tau trigger efficiency with the L1 tau trigger efficiency~\cite{bib:trigpub}*.

{\bf Top}: Tau trigger efficiency measured in data and compared to simulation, with respect to offline reconstructed tau candidate with one or three tracks and passing the offline medium identification criteria, as function of the offline transverse momentum. The trigger efficiency is measured in a tag and probe analysis with $Z\rightarrow\tau\tau\rightarrow\mu\tau_{had}$ event from the 2016 dataset in 13TeV collision (8.0 fb$^{-1}$)*. {\bf Bottom} Comparison of this HLT tau trigger efficiency with the L1 tau trigger efficiency~\cite{bib:trigpub}*.

The per-bunch trigger rate for the L1 missing transverse momentum trigger with a threshold of 50 GeV (L1\_XE50) as a function of the instantaneous luminosity per bunch~\cite{bib:atltriperf2015}. The rates are shown with and without the pedestal correction applied*.

Comparison of the likelihood-base and the cut-base HLT electron triggers efficiency as a function of the offline electron candidate's transverse energy $E_T$ with respect to true reconstructed electrons in $Z\rightarrow ee$ simulation. The HLT\_e24\_medium\_iloose\_L1EM18VH trigger is the Run-1 algorithm requiring an electron candidate with $E_T > 24$ GeV satisfying the cut-based medium identification, while HLT\_e24\_lhmedium\_iloose\_L1EM18VH corresponds to the Run-2 algorithm using the likelihood-based lhmedium electron identification. Both trigger chains also require the same track isolation selection and are seeded by the same level-1 trigger (L1\_EM18VH)~\cite{bib:trigpub}*.

The trigger cross-section as measured by using online rate and luminosity is compared for the main trigger $E_T^{miss}$ reconstruction algorithms used in 2016 ("mht") and 2017 ("pufit") as a function of the mean number of simultaneous interactions per proton-proton bunch crossing averaged over all bunches circulating in the LHC~\cite{bib:trigpub}*.

Efficiency of the L1\_EM20VHI trigger (circles) as well as the combined L1\_EM20VHI and HLT\_e24\_lhtight\_nod0\_ivarloose trigger (blue triangles) as a function of the offline electron candidate's transverse energy ($E_T$). A variable-size cone isolation criteria is applied ("ivarloose"). The HLT trigger requires an electron candidate with $E_T > 24$ GeV satisfying the likelihood-based tight identification. The offline reconstructed electron is required to pass a likelihood-based tight identification~\cite{bib:trigpub}*.

The combined L1 and HLT efficiency of the missing transverse energy triggers HLT\_xe110\_pufit\_L1XE50 and HLT\_xe110\_mht\_L1XE50 as well as the efficiency of the corresponding L1 trigger (L1\_XE50) are shown as a function of the reconstructed $E_T^{miss}$ (modified to count muons as invisible)~\cite{bib:trigpub}. The events shown are taken from data with a $W\rightarrow\mu\nu$ selection to provide a sample enriched in real $E_T^{miss}$*.

Efficiency of the HLT likelihood-base electron trigger as a function of the offline electron candidate's transverse energy ($E_T$) as measured with the tab and probe method on a sample of 2017 ATLAS data as well as on a $Z\rightarrow ee$ Monte Carlo sample~\cite{bib:trigpub}*. The data-to-MC agreement is very good. The efficiency is calculated for two different trigger chain: an electron candidate with $E_T > 24$ GeV satisfying the likelihood-based very loose identification and seeded by a 20 GeV electron L1 trigger ({\bf Left}); and an electron candidate with $E_T > 26$ GeV satisfying the likelihood-based tight identification and seeded by a 22 GeV electron L1 trigger ({\bf Right}).

{\bf Left}: The distributions of processing times for the topo-clustering algorithm executed on the full calorimeter~\cite{bib:atltriperf2015}*. {\bf Right}: The transverse energy $E_T$ resolution for online (HLT) topo-clusters with respect to offline topo-clusters with $E_T > 3$ GeV~\cite{bib:trigpub}. The online and offline clusters are both hadronically calibrated. The improvement in the $E_T$ resolution for online topo-clusters with respect to offline topo-clusters in 2016 is due to the introduction cell-level, BCID/$<\mu>$ based energy corrections. These corrections account for pedestal changes that arise due to out-of-time pile-up and particularly affect the first bunch crossings in each bunch train. Similar corrections were already being applied offline*.

Efficiency of the HLT likelihood-base electron trigger as a function of the offline electron candidate's transverse energy ($E_T$) as measured with the tab and probe method on a sample of 2017 ATLAS data as well as on a $Z\rightarrow ee$ Monte Carlo sample~\cite{bib:trigpub}*. The data-to-MC agreement is very good. The efficiency is calculated for two different trigger chain: an electron candidate with $E_T > 24$ GeV satisfying the likelihood-based very loose identification and seeded by a 20 GeV electron L1 trigger ({\bf Left}); and an electron candidate with $E_T > 26$ GeV satisfying the likelihood-based tight identification and seeded by a 22 GeV electron L1 trigger ({\bf Right}).

Transverse momentum distribution for the positron coming from inclusive $W^+$ decays. The data are compared to the simulation including signal and background contributions. Detector calibration and physics-modelling corrections are applied to the simulated events.The lower panels show the data-to-prediction ratios, the error bars show the statistical uncertainty, and the band shows the systematic uncertainty of the prediction. The $\chi^2$ values displayed in each figure account for all sources of uncertainty and include the effects of bin-to-bin correlations induced by the systematic uncertainties~\cite{bib:wmass}*.

{\bf Left}: The distributions of processing times for the topo-clustering algorithm executed on the full calorimeter~\cite{bib:atltriperf2015}*. {\bf Right}: The transverse energy $E_T$ resolution for online (HLT) topo-clusters with respect to offline topo-clusters with $E_T > 3$ GeV~\cite{bib:trigpub}. The online and offline clusters are both hadronically calibrated. The improvement in the $E_T$ resolution for online topo-clusters with respect to offline topo-clusters in 2016 is due to the introduction cell-level, BCID/$<\mu>$ based energy corrections. These corrections account for pedestal changes that arise due to out-of-time pile-up and particularly affect the first bunch crossings in each bunch train. Similar corrections were already being applied offline*.

Absolute efficiency of Level 1 (L1) MU20 trigger and absolute and relative efficiencies of the OR of mu26\_ivarmedium with mu50 High Level Triggers (HLT) plotted as a function of $\phi$ of offline muon candidates in the barrel detector region. The efficiency is computed with respect to offline isolated muon candidates which are reconstructed using standard ATLAS software and are required to pass "Medium" quality requirement. The selection is restricted to the plateau region with $p_T > 27$ GeV~\cite{bib:trigpub}*.

The ATLAS TDAQ system in Run-2 with emphasis on the components relevant for triggering~\cite{bib:atltriperf2015}*.

Jet $p_T$ spectrum after the basic kinematic selection for the TLA trigger jets compared to trigger jets recorded by all single-jet triggers~\cite{bib:atltriperf2015}*.

Absolute efficiency of Level 1 (L1) MU20 trigger and absolute and relative efficiencies of the OR of mu26\_ivarmedium with mu50 High Level Triggers (HLT) plotted as a function of $p_T$ of offline muon candidates in the barrel detector region ({\bf Left} ), and the endcap detector region ({\bf Right} )~\cite{bib:trigpub}*. The efficiency is computed exactly like described in the caption of Fig.~\ref{fig:muonphi}.

Absolute efficiency of Level 1 (L1) MU20 trigger and absolute and relative efficiencies of the OR of mu26\_ivarmedium with mu50 High Level Triggers (HLT) plotted as a function of $p_T$ of offline muon candidates in the barrel detector region ({\bf Left} ), and the endcap detector region ({\bf Right} )~\cite{bib:trigpub}*. The efficiency is computed exactly like described in the caption of Fig.~\ref{fig:muonphi}.

Efficiencies are shown for a single-jet trigger with three different calibrations~\cite{bib:jetcalib} applied to jets in the ATLAS high-level trigger (HLT)~\cite{bib:trigpub}*.

Distributions of the HLT tau candidates passing the tau25\_medium trigger: ({\bf Left} ) transverse momentum, ({\bf Right} online BDT identification score. The HLT tau candidates are matched to offline tau candidates with transverse momentum above 25 GeV, with one or three tracks and satisfying the offline medium tau identification criterion~\cite{bib:atltriperf2015}*.

{\bf Left}: Efficiencies for HLT single-jet triggers as a function of leading offline jet $p_{T}$. Triggers denoted HLT\_jX accept an event if a jet is reconstructed at HLT with $E_T > X$ GeV*. The unprescaled trigger with the lowest threshold requires a jet with $E_\mathrm{T} > 380$ GeV~\cite{bib:trigpub}. {\bf Right}: Efficiencies for HLT large-R single-jet triggers as a function of the leading offline trimmed~\cite{bib:trim} jet $p_\mathrm{T}$. Blue circles represent a trimmed large-R jet trigger with a $p_\mathrm{T}$ threshold of 420 GeV. Adding an additional 30 GeV cut on the jet mass of the selected trimmed trigger jet is shown in green triangles. The mass cut significantly suppresses the QCD di-jet background, allowing a lower $p_\mathrm{T}$ threshold of 390 GeV, while retaining nearly all signal-like jets with a mass of above 50 GeV*.

{\bf Left}: A schematic view of the muon spectrometer with lines indicating various pseudo rapidity regions~\cite{bib:atltriperf2015}. The {\it curved arrows} shows an example of a trajectory from slow particles generated at the beam pipe around $z\sim 10$m. Triggers due to events of this type are mitigated by requiring an additional coincidence with the TGC-FI chambers in the region $1.3<|\eta|<1.9$*. {\bf Right}: Number of events with an L1 muon trigger with transverse momentum ($p_T$) above 15 GeV (L1\_MU15) as a function of the muon $\eta$ coordinate, when a coincidence with the TGC-FI chambers is required (upper histogram) or no requirement is applied (lower histogram)~\cite{bib:atltriperf2015}*.

{\bf Left}: Efficiencies for HLT single-jet triggers as a function of leading offline jet $p_{T}$. Triggers denoted HLT\_jX accept an event if a jet is reconstructed at HLT with $E_T > X$ GeV*. The unprescaled trigger with the lowest threshold requires a jet with $E_\mathrm{T} > 380$ GeV~\cite{bib:trigpub}. {\bf Right}: Efficiencies for HLT large-R single-jet triggers as a function of the leading offline trimmed~\cite{bib:trim} jet $p_\mathrm{T}$. Blue circles represent a trimmed large-R jet trigger with a $p_\mathrm{T}$ threshold of 420 GeV. Adding an additional 30 GeV cut on the jet mass of the selected trimmed trigger jet is shown in green triangles. The mass cut significantly suppresses the QCD di-jet background, allowing a lower $p_\mathrm{T}$ threshold of 390 GeV, while retaining nearly all signal-like jets with a mass of above 50 GeV*.

Distributions of the HLT tau candidates passing the tau25\_medium trigger: ({\bf Left} ) transverse momentum, ({\bf Right} online BDT identification score. The HLT tau candidates are matched to offline tau candidates with transverse momentum above 25 GeV, with one or three tracks and satisfying the offline medium tau identification criterion~\cite{bib:atltriperf2015}*.

Distributions of the HLT tau candidates passing the tau25\_medium trigger: ({\bf Left} ) transverse momentum, ({\bf Right} online BDT identification score. The HLT tau candidates are matched to offline tau candidates with transverse momentum above 25 GeV, with one or three tracks and satisfying the offline medium tau identification criterion~\cite{bib:atltriperf2015}*.

{\bf Left}: Tau trigger efficiency measured in data and compared to simulation, with respect to offline reconstructed tau candidate with one or three tracks and passing the offline medium identification criteria, as function of the offline transverse momentum. The trigger efficiency is measured in a tag and probe analysis with $Z\rightarrow\tau\tau\rightarrow\mu\tau_{had}$ event from the 2016 dataset in 13TeV collision (8.0 fb$^{-1}$)*. {\bf Right} Comparison of this HLT tau trigger efficiency with the L1 tau trigger efficiency~\cite{bib:trigpub}*.

{\bf Left}: Tau trigger efficiency measured in data and compared to simulation, with respect to offline reconstructed tau candidate with one or three tracks and passing the offline medium identification criteria, as function of the offline transverse momentum. The trigger efficiency is measured in a tag and probe analysis with $Z\rightarrow\tau\tau\rightarrow\mu\tau_{had}$ event from the 2016 dataset in 13TeV collision (8.0 fb$^{-1}$)*. {\bf Right} Comparison of this HLT tau trigger efficiency with the L1 tau trigger efficiency~\cite{bib:trigpub}*.

Distributions of the HLT tau candidates passing the tau25\_medium trigger: ({\bf Left} ) transverse momentum, ({\bf Right} online BDT identification score. The HLT tau candidates are matched to offline tau candidates with transverse momentum above 25 GeV, with one or three tracks and satisfying the offline medium tau identification criterion~\cite{bib:atltriperf2015}*.

The trigger cross-section as measured by using online rate and luminosity is compared for the main trigger $E_T^{miss}$ reconstruction algorithms used in 2016 ("mht") and 2017 ("pufit") as a function of the mean number of simultaneous interactions per proton-proton bunch crossing averaged over all bunches circulating in the LHC~\cite{bib:trigpub}*.

The combined L1 and HLT efficiency of the missing transverse energy triggers HLT\_xe110\_pufit\_L1XE50 and HLT\_xe110\_mht\_L1XE50 as well as the efficiency of the corresponding L1 trigger (L1\_XE50) are shown as a function of the reconstructed $E_T^{miss}$ (modified to count muons as invisible)~\cite{bib:trigpub}. The events shown are taken from data with a $W\rightarrow\mu\nu$ selection to provide a sample enriched in real $E_T^{miss}$*.