Performance of the ATLAS trigger system in 2015

Abstract

During 2015 the ATLAS experiment recorded [Formula: see text] of proton-proton collision data at a centre-of-mass energy of [Formula: see text]. The ATLAS trigger system is a crucial component of the experiment, responsible for selecting events of interest at a recording rate of approximately 1 kHz from up to 40 MHz of collisions. This paper presents a short overview of the changes to the trigger and data acquisition systems during the first long shutdown of the LHC and shows the performance of the trigger system and its components based on the 2015 proton-proton collision data.

Fig. 1

The ATLAS TDAQ system in Run 2 with emphasis on the components relevant for triggering. L1Topo and FTK were being commissioned during 2015 and not used for the results shown here

Fig. 2

Schematic view of the trigger towers used as input to the L1Calo trigger algorithms

Fig. 3

The per-bunch trigger rate for the L1 missing transverse momentum trigger with a threshold of 50 $\text{GeV}$ (L1_XE50) as a function of the instantaneous luminosity per bunch. The rates are shown with and without pedestal correction applied

Fig. 4

A schematic view of the muon spectrometer with lines indicating various pseudorapidity regions. The curved arrow shows an example of a trajectory from slow particles generated at the beam pipe around $z\sim 10\text{m}.$ Triggers due to events of this type are mitigated by requiring an additional coincidence with the TGC-FI chambers in the region $1.3<|\mathit{\eta }|<1.9$

Fig. 5

a Number of events with an L1 muon trigger with transverse momentum ($p_{\text{T}}$) above 15 $\text{GeV}$ (L1_MU15) as a function of the muon trigger $\mathit{\eta }$ coordinate, requiring a coincidence with the TGC-FI chambers (open histogram) and not requiring it (cross-hatched histogram), together with the fractional event rate reduction in the bottom plot. The event rate reduction in the regions with no TGC-FI chambers is consistent with zero within the uncertainty. b Efficiency of L1_MU15 in the end-cap region, as a function of the $p_{\text{T}}$ of the offline muon measured via a tag-and-probe method (see Sect. 6) using $Z\to \mathit{\mu }\mathit{\mu }$ events with (open dots) and without (filled dots) the TGC-FI coincidence, together with the ratio in the bottom panel

Fig. 6

a L1 and b HLT trigger rates grouped by trigger signature during an LHC fill in October 2015 with a peak luminosity of $4.5\times {10}^{33}$ cm${}^{-2}$ s${}^{-1}$. Due to overlaps the sum of the individual groups is higher than the a L1 total rate and b Main physics stream rate, which are shown as black lines. Multi-object triggers are included in the b-jets and tau groups. The rate increase around luminosity block 400 is due to the removal of prescaling of the B-physics triggers. The combined group includes multiple triggers combining different trigger signatures such as electrons with muons, taus, jets or $E_{\text{T}}^{\text{miss}}$

Fig. 7

a HLT stream rates and b bandwidth during an LHC fill in October 2015 with a peak luminosity of $4.5\times {10}^{33}$ cm${}^{-2}$ s${}^{-1}$. Partial Event Building (partial EB) streams only store relevant subdetector data and thus have smaller event sizes. The other physics-related streams contain events with special readout settings and are used to overlay with MC events to simulate pile-up

Fig. 8

a HLT processing time distribution per event for an instantaneous luminosity of $5.2\times {10}^{33}$ cm${}^{-2}$ s${}^{-1}$ and average pile-up $⟨\mathit{\mu }⟩=15$ and b mean HLT processing time as a function of the instantaneous luminosity

Fig. 9

Simulated limits on the L1 trigger rate due to the IBL fixed frequency veto for two different filling schemes and the expected maximum L1 rate from rate predictions. The steps in the latter indicate a change in the prescale strategy. The simulated rate limit is confirmed with experimental tests. The rate limit is higher for the 72-bunch train configuration since the bunches are more equally spread across the LHC ring. The rate limitation was only crucial for the low luminosity phase, where the required physics L1 rate was higher than the limit imposed by the IBL veto. The maximum number of colliding bunches in 2015 was 2232

Fig. 10

The ID tracking efficiency for the 24 $\text{GeV}$ electron trigger is shown as a function of the a $\mathit{\eta }$ and b $p_{\text{T}}$ of the track of the offline electron candidate. Uncertainties based on Bayesian statistics are shown

Fig. 11

The ID tracking performance for the 6 $\text{GeV}$ muon trigger; a efficiency as a function of the offline reconstructed muon $p_{\text{T}}$, b the resolution of the transverse impact parameter, $d_0$ as a function of the offline reconstructed muon $p_{\text{T}}$. Uncertainties based on Bayesian statistics are shown

Fig. 12

A schematic illustrating the RoIs from the single-stage and two-stage tau lepton trigger tracking, shown in plan view (x–z plane) along the transverse direction and in perspective view. The z-axis is along the beam line. The combined tracking volume of the 1st and 2nd stage RoI in the two-stage tracking approach is significantly smaller than the RoI in the one-stage tracking scheme

Fig. 13

The ID trigger tau tracking performance with respect to offline tracks from very loose tau candidates with $p_{\text{T}}>1\text{GeV}$ from the 25 $\text{GeV}$ tau trigger; a the efficiency as a function of the offline reconstructed tau track $p_{\text{T}}$, b the resolution of the transverse impact parameter, $d_0$ as a function of the offline reconstructed tau track $p_{\text{T}}$. The offline reconstructed tau daughter tracks are required to have $p_{\text{T}}>1\text{GeV}$, lie in the region $|\mathit{\eta }|<2.5$ and have at least two pixel clusters and at least six SCT clusters. The closest matching trigger track within a cone of size $\mathrm{\Delta }R=0.05$ of the offline track is selected as the matching trigger track

Fig. 14

The trigger performance for primary vertices in the b-jet signatures for 55, 110 and 260$\text{GeV}$jet triggers; a the vertexing efficiency as a function of the number of offline tracks within the jets used for the vertex tracking, b the resolution in z of the vertex with respect to the offline vertex position as a function of the number of offline tracks from the offline vertex

Fig. 15

The CPU processing time for the fast and precision tracking per electron RoI for the 24 $\text{GeV}$ electron trigger. The precision tracking is seeded by the tracks found in the fast tracking stage and hence requires less CPU time

Fig. 16

The ID trigger tau tracking processing time for a the fast track finder and b the precision tracking comparing the single-stage and two-stage tracking approach

Fig. 17

The relative differences between the online and offline $E_{\text{T}}$ for a sliding-window clusters and b topo-clusters. Online and offline clusters are matched within $\mathrm{\Delta }R$ < 0.001. The distribution for the topo-clusters was obtained from the RoI-based topo-clustering algorithm that is used for online tau reconstruction

Fig. 18

The distributions of processing times for the topo-clustering algorithm executed a within an RoI and b on the full calorimeter. The processing times within an RoI are obtained from tau RoIs with a size of $\mathrm{\Delta }\mathit{\eta }\times \mathrm{\Delta }\mathit{\varphi }=0.8\times 0.8$

Fig. 19

Width of the residuals for inverse-$p_{\text{T}}$ as a function of offline muon $p_{\text{T}}$ for the precision MS-only and combined algorithms in the barrel ($|\mathit{\eta }|<1.05$) and end-caps ($1.0<|\mathit{\eta }|<2.4$)

Fig. 20

Width of the residuals as a function of the offline muon $p_{\text{T}}$ for a $\mathit{\eta }$ and b $\mathit{\varphi }$ for the precision MS-only and combined algorithms in the barrel ($|\mathit{\eta }|<1.05$) and end-caps ($1.0<|\mathit{\eta }|<2.4$)

Fig. 21

Processing times per RoI for the a fast MS-only and fast combined algorithms and b precision muon-finding algorithm. The time for the combined algorithm includes only the time for the ID–MS combination and not the tracking itself. The mean time of each algorithm is indicated in the legend. The large number of entries in the first bin in b is due to algorithm caching [29]

Fig. 22

Efficiency of L1_MBTS_1, L1_MBTS_2 and L1_MBTS_1_1 triggers as a function of the number tracks compatible with the beam line for two different transverse momentum requirements a $p_{\text{T}}>100\text{MeV}$ and b $p_{\text{T}}>500\text{MeV}$. The bands denote the total uncertainty

Fig. 23

Efficiency of high-multiplicity triggers as a function of the number of tracks compatible with the primary vertex for two different offline transverse momentum requirements a $p_{\text{T}}>400\text{MeV}$ and b $p_{\text{T}}>500\text{MeV}$. The curves represent three different selections on the minimum number of SCT space-points and reconstructed tracks (900/60, 1000/70 and 1400/80)

Fig. 24

L1 trigger rates as a function of the instantaneous luminosity for selected single- and multi-object triggers

Fig. 25

HLT trigger rates for a electron and b photon triggers as a function of the instantaneous luminosity for selected single- and multi-object triggers

Fig. 26

Efficiency of the L1_EM20VH trigger and the logical ‘or’ of the e24_lhmedium_L1EM20VH and e120_lhloose triggers as a function of a the probe electron transverse energy $E_{\text{T}}$ and b pseudorapidity $\mathit{\eta }$. The offline reconstructed electron candidate is required to have an $E_{\text{T}}$ value at least 1 $\text{GeV}$ above the trigger threshold

Fig. 27

Efficiency of HLT photon triggers g20_tight, g25_loose, g35_loose, and g120_loose relative to a looser HLT photon trigger as a function of a the transverse energy $E_{\text{T}}$ and b pseudorapidity $\mathit{\eta }$ of the photon candidates reconstructed offline and satisfying the tight identification and isolation requirements. The offline reconstructed photon candidate is required to have an $E_{\text{T}}$ value at least 5 $\text{GeV}$ above the trigger threshold. The transition region between the barrel and end-cap calorimeter ($1.37<|\mathit{\eta }|<1.52$) is excluded

Fig. 28

a L1 and b HLT muon trigger rates as a function of the instantaneous luminosity for primary single and dimuon triggers

Fig. 29

Efficiency of the L1 muon trigger L1_MU15 and the combination of the HLT muon triggers mu20_iloose_L1MU15 and mu50 as a function of the probe muon $p_{\mathrm{T}}$, separately for a the barrel and b the end-cap regions

Fig. 30

Efficiency of the L1 muon trigger L1_MU15 and the combination of the HLT muon triggers mu20_iloose_L1MU15 and mu50 as a function of the probe muon $\mathit{\varphi }$, separately for a the barrel and b the end-cap regions

Fig. 31

Efficiency of single-jet triggers as a function of offline jet $p_{\text{T}}$ for a L1 in the central region, b L1 in the forward region, c HLT in the central region, and d HLT in the forward region

Fig. 32

Comparison between the jet energy scales of trigger and offline jets. The black points represent the mean of the distribution at a given $p_{\text{T}}$ value. The 2% shift is due to differences in the jet calibration applied online and offline

Fig. 33

Efficiency of multi-jet a L1 and b HLT triggers as a function of offline jet $p_{\text{T}}$

Fig. 34

Efficiency of a $H_{\text{T}}$ triggers as a function of offline $H_{\text{T}}$ and b large-R ($R=1.0$) single-jet triggers as a function of offline $p_{\text{T}}$ . $H_{\text{T}}$ is defined as the summed transverse energy of all jets that are reconstructed above a transverse energy threshold of 50 $\text{GeV}$

Fig. 35

Jet $p_{\text{T}}$ spectrum after the basic kinematic selection for the TLA trigger jets (black) compared to trigger jets recorded by all single-jet triggers (blue)

Fig. 36

Trigger rates as a function of instantaneous luminosity for several a L1 and b HLT tau triggers

Fig. 37

Distributions of the HLT tau candidates passing the tau25_medium trigger: a transverse momentum, b pseudorapidity, c track multiplicity distributions of the core tracks $\mathrm{\Delta }R<0.2$ of the tau-axis and d online BDT identification score. The HLT tau candidates are matched to offline tau candidates with transverse momentum above 25 $\text{GeV}$, with one or three tracks and satisfying the offline medium tau identification criterion. Only statistical uncertainties are shown, and the last bin in a contains overflow events. The ratio of the observed data to the expected signal and background events is also shown, where the red band shows the statistical uncertainty of the total prediction

Fig. 38

Efficiency of the tau25_medium trigger measured in data as a function of the offline tau $p_{\text{T}}$ for offline tau candidates with $p_{\text{T}}$ above 25 $\text{GeV}$, one or three tracks and satisfying the offline medium identification requirement. The expected background contribution has been subtracted from the data. a Efficiencies after the L1 (red) and L1+HLT (blue) selections are shown separately with only statistical uncertainties. b Comparison of the measured efficiency after L1+HLT to simulation. Statistical uncertainties associated with data and simulation and the systematic uncertainty associated with the background subtraction procedure in data are shown

Fig. 39

Comparison of the different $E_{\text{T}}^{\text{miss}}$ distributions for events accepted by the HLT into the Main physics stream. The algorithms consist of a cell-based $E_{\text{T}}^{\text{miss}}$ (xe) and different topo-cluster-based algorithms described in the text. The zero entries of the xe_tc_pufit algorithm, which occur when no tower is above threshold, have been suppressed. The steps in the distributions are caused by the various trigger thresholds

Fig. 40

$E_{\text{T}}^{\text{miss}}$ trigger rates (a) at L1 and (b) for various HLT algorithms operating with nominal thresholds of 70 $\text{GeV}$. The HLT algorithms are each seeded by L1_XE50. Rates are shown as a function of instantaneous luminosity from various runs taken in 2015 excluding periods with atypically high or low rates arising from different pile-up conditions for the same instantaneous luminosity

Fig. 41

$E_{\text{T}}^{\text{miss}}$ trigger efficiency curves with respect to the $E_{\text{T}}^{\text{miss}}$ reconstructed offline without muon corrections for all events passing the a $W\to e\mathit{\nu }$ or b $W\to \mathit{\mu }\mathit{\nu }$ selections. The different efficiencies were measured for L1, and for the combination of L1 with each of the HLT $E_{\text{T}}^{\text{miss}}$ algorithms. The thresholds for the different algorithms correspond to an approximately equal trigger rate

Fig. 42

a $E_{\text{T}}^{\text{miss}}$ trigger linearity with respect to the $E_{\text{T}}^{\text{miss}}$ reconstructed offline without muon corrections and b $E_{\text{T}}^{\text{miss}}$ trigger resolution with respect to the $\mathrm{\Sigma }{E}_{\text{T}}$ reconstructed offline without muon corrections, for all events passing $W\to \mathit{\mu }\mathit{\nu }$ selections for L1 and for each HLT $E_{\text{T}}^{\text{miss}}$ algorithm. Linearity and resolution are defined in the text

Fig. 43

$E_{\text{T}}^{\text{miss}}$ trigger efficiency curves with respect to the $E_{\text{T}}^{\text{miss}}$ reconstructed offline without muon corrections for the $W\to \mathit{\mu }\mathit{\nu }$ selection. The different efficiencies were obtained for different pile-up conditions expressed in terms of various ranges of the average number of reconstructed vertices per bunch-crossing (denoted here as $N_{\mathrm{vtx}}$). The efficiency of the L1 algorithm is included

Fig. 44

a Transverse and b longitudinal impact parameter significance for tracks associated with light-flavour (black) and b-quark (red) jets measured in a sample of simulated $t\overline{t}$ events. The solid lines show the distribution for the offline tracks. The points show the corresponding distribution for tracks used in the b-jet trigger. The impact parameter significance is defined as the impact parameter divided by the associated uncertainty. The impact parameters are signed such that track displacements in the direction of the jet have positive values, while tracks with displacements opposite of the jet direction are negative

Fig. 45

The expected performance of the MV2c20 trigger tagger (solid black line) in terms of light-jet rejection is shown together with the expected performance of the IP3D+SV1 trigger tagger in Run 2 (dashed blue line) and its actual performance achieved during Run 1 (red stars)

Fig. 46

b-jet trigger efficiency as a function of jet $p_{\text{T}}$ for the loose (blue), medium (black) and tight (red) operating points. The open squares show the trigger efficiency using the IP3D+SV1 b-tagging algorithm. The closed circles show the trigger efficiency for the corresponding MV2 b-tagging algorithm working points. The efficiencies are measured in a pure sample of b-jets selected in $t\overline{t}$ events and are computed with respect to jets identified by the 70% working point of the offline MV2c20 b-tagging algorithm

Fig. 47

Rates of b-jet triggers as a function of the instantaneous luminosity

Fig. 48

Trigger rates for a low-$p_{\text{T}}$ dimuon L1 triggers with various muon $p_{\text{T}}$ thresholds and b primary HLT B-physics triggers as a function of instantaneous luminosity. b Shows triggers requiring two muons to pass various $p_{\text{T}}$ thresholds, to have an invariant mass within the $J/\mathit{\psi }$ mass window, and to form a good vertex (full markers); also shown are triggers requiring two muons with $p_{\text{T}}>6$ and 4 $\text{GeV}$ and either having an invariant mass in a different window ($B_{(s)}^0$, $\mathrm{{\rm Y}}(1,2,3\mathrm{S})$) or forming a $B\to \mathit{\mu }\mathit{\mu }X$ candidate after combination with additional tracks found in ID (open markers). As L1_2MU4 was prescaled at luminosities above $4\times {10}^{33}{\text{cm}}^{-2}{\text{s}}^{-1}$, the rate of 2mu4_bJpsimumu seeded from this L1 trigger drops above that luminosity

Fig. 49

Invariant mass distribution of offline-selected dimuon candidates passing the lowest thresholds of dimuon B-physics triggers. Triggers targeting different invariant mass ranges are illustrated with different colours, and the differing thresholds are shown with different shadings. No accounting for overlaps between triggers is made, and the distributions are shown overlaid, and not stacked. For comparison, the number of candidates passing the lowest unprescaled single-muon trigger and supporting dimuon trigger is also shown

Fig. 50

The efficiency of the opposite-sign muon requirement and vertex quality selection applied for dimuon B-physics triggers as a function of $p_{\text{T}}(\mathit{\mu }\mathit{\mu })$ for three rapidity regions. Supporting dimuon triggers with and without the selection criteria applied are used to determine the efficiency. The integrated luminosity shown takes into account the high prescale factors applied to the supporting triggers