ATLAS at the LHC

Aurelio Juste

Since 1993, the IFAE-ATLAS group has made major contributions to the construction of the ATLAS detector and its trigger system, the reconstruction software, and preparatory physics studies. Using the data collected by the ATLAS experiment during Run 1 (2010-2012) and Run 2 (2015-2018) of the Large Hadron Collider (LHC), the IFAE group has carried out a strong physics program.


During 2020, the second year of the second long shutdown of the LHC (LS2), the group has been actively involved in the analysis of the full LHC Run-2 dataset, and the maintenance and commissioning of detector elements under IFAE responsibility in preparation for the start of the LHC Run 3 in 2021. The group has also maintained its visibility within the ATLAS Collaboration through a number of important management positions.


The year 2020 was characterized by an intense analysis effort using the full Run-2 dataset

The year 2020 was characterized by an intense analysis effort using the full Run-2 dataset, which corresponds to $139 fb^{-1}$ of proton-proton collisions at a center-of-mass energy of 13 TeV. The IFAE team released several results that were presented at International conferences on searches for dark matter, leptoquarks, and Higgs boson studies within the Standard Model (SM) and beyond (BSM). A summary of the main analysis activities is provided below.

Higgs Boson Physics

During Run 2 the IFAE team is participating strongly in measurements of $t\bar tH$ production, and searches for additional Higgs bosons.

Probing top-Higgs interactions

Direct tests of top-Higgs interactions are extremely interesting, since the top quark is the SM particle most strongly coupled to the Higgs sector. The most sensitive direct probe of the top-Higgs Yukawa coupling is provided by the measurement of the ttH cross section. During Run 1 and throughout Run 2 the IFAE team has actively participated in the $t\bar tH (\to b \bar b)$ search, resulting in several public results. The most recent result (ATLAS-CONF-2020-058) uses the full Run-2 dataset. In addition to the inclusive cross section measurement, this result includes a differential measurement in five bins of Higgs boson transverse momentum within the Simplified Template Cross Section (STXS) framework, including a boosted selection targeting Higgs boson transverse momentum above 300 GeV (see Fig. 1).

Figure 1: Signal-strength (μ) measurements in the individual STXS pTH bins, as well as the inclusive signal strength. From ATLAS-CONF-2020-058.

Since 2018 the IFAE team is also involved in the ttH search with $H \to WW^*/ZZ^*/\tau \bar\tau $decays, giving multilepton final states. Based on the findings from the most recent result (ATLAS-CONF-2019-04), which highlighted some issues with the modeling of the dominant ttW background, the IFAE team (N. Agaras, S. Epari, S. Kazakos, and A. Juste) is actively involved in the first measurements of differential cross sections for ttW production, using the full Run-2 dataset. Such measurement which will be part of S. Kazakos’ PhD Thesis.

Searches for additional Higgs bosons

During 2020 the IFAE team (M. Bosman, Ll.M. Mir, I. Riu, and A. Salvador) continued its strong involvement in the search for a heavy charged Higgs boson (H+) decaying to a top quark and a bottom quark. IFAE is the only institute in charge of the full Run-2 analysis in the standard lepton-plus-jets final state. A sophisticated approach using a neural network parameterized as a function of the H+ mass was developed to improve the search sensitivity. A preliminary result (ATLAS-CONF-2020-039) was presented at the ICHEP 2020 conference, achieving improved sensitivity with respected the previous partial Run 2 data analysis (see Fig. 2). Ll. M. Mir and I. Riu are corresponding editors of the paper, which was recently submitted for publication in JHEP (arXiv:2102.10076). Undergraduate students C. Soriano, G. Stefanescu and F. Recasens defended their Final Degree Projects in Physics working on several improvements for this search: the optimization of a kinematic discriminant used in the NN, the optimization of the NN architecture, and the implementation of a reweighting function to better simulate the data, respectively.

Figure 2: Observed and expected upper limits for the production of H+→tb in association with a top quark and a bottom quark as a function of the H+ mass. From arXiv:2102.10076.

The IFAE team (N. Agaras, S. Kazakos, A. Juste) developed a novel search for new heavy flavor-violating neutral Higgs bosons, mediating the production of two same-sign top quarks, three top quarks, or four top quarks, and resulting in multilepton final states. The results of this search are expected during 2021. The IFAE team has also continued to develop a broad program of searches for light scalars that are produced in the decay of Higgs bosons or top quarks, or in association with top quarks. Such scenarios are predicted in several extensions of the Higgs sector, and are poorly tested experimentally.

The IFAE team (A. Juste, I. Riu and P. Martínez) is involved in a search for $h \to a \bar a)$ decays, where “a” is a light pseudoscalar that decays dominantly into bb. For very light scalars, the two b-quarks are merged into a single fat jet, requiring the development of a dedicated tagging algorithm. The first result in this kinematic regime, using $36 fb^{-1}$ of Run 2 data, was published in Phys. Rev. D 102 (2020) 112006. A more sophisticated tagging algorithm has been developed for use in the full Run-2 dataset analysis, which is underway.

More advanced is the search for a light charged Higgs boson appearing in tt events, with one of the top quarks undergoing the decay $t \to H^+b$, and the H+ boson decaying into a bottom quark and a charm quark. The IFAE members participating are A. Juste and N. Orlando, who is the ATLAS analysis contact for this search. This search is expected to probe branching ratios as low as 0.1%, and will be published in 2021. The same analysis is being reoptimized to search, for the first time, for flavor-violating top-quark decays $t \to X(\to bb)c$, where “X” denotes a light pseudoscalar, a scalar, or an axion-like particle. The IFAE team (M. Bosman, A. Juste, Ll.M. Mir, N. Orlando, I. Riu, and A. Salvador) is leading this search, for which I. Riu and A. Salvador are the analysis contacts.

Searches for new phenomena in Jet+X

In 2020, the IFAE team (D. Bogavac, J.L. Muñoz, M. Martínez, R. Rosten, and S. González) maintained the leadership in the search for a jet plus large missing transverse momentum (ETmiss) using the full Run 2 dataset. Preliminary results (ATLAS-CONF-2020-048) were presented at the ICHEP 2020 conference, and the final results have been recently submitted for publication in Phys. Rev D (arXiv:2102.10874), with M. Martinez acting as corresponding editor. The data is in good agreement with the SM predictions. As illustrated in Fig. 3, the level of precision achieved in the analysis (below 2% at low ETmiss and about 4% at very high ETmiss) turns this result into a stringent test of the SM predictions leaving little room for new phenomena in this final state. The results have been interpreted in a number of models. In particular, IFAE has led the interpretations in terms of large extra spatial dimensions, supersymmetry in compressed scenarios, and axion-like particles. In addition, IFAE took the responsibility of extracting new bounds on invisibly decaying Higgs bosons.

Figure 3: Measured recoil transverse momentum distribution in the mono-jet final state compared to SM predictions and the predictions from different BSM models. From arXiv:2102.10874.

Part of the IFAE team (D. Bogavac, and S. González) were also deeply involved in the complementary mono-photon search, and played a central role in the final background determination and the interpretations related to axion-like particles. The analysis found good agreement between data and SM predictions and was recently published (JHEP 02 (2021) 226). The team is now focused on contributing with the mono-jet and mono-photon results to a number of ATLAS summary papers collecting all the Run 2 results on dark matter searches including additional interpretations. Finally, in 2020 the IFAE team continued the searches for mono-V (vector boson), as a natural complement of the mono-jet analysis, accessing new interpretations related to dark matter Higgs portal models and the measurement of the invisibly decaying Higgs branching fraction. D. Bogavac is the ATLAS analysis contact for this search. In this analysis, the reconstruction of boosted vector bosons at very high transverse momentum is required, for which both QCD-inspired sub-jet quantities in large-R jets and a deep learning approach are also employed in separating vector-boson decay products from gluon radiation. First preliminary results are expected by summer 2021.

Supersymmetry searches

One possible solution to the gauge hierarchy problem is provided by weak-scale supersymmetry, which extends the SM by introducing supersymmetric partners for all SM particles. Searches for gluinos, top/bottom squarks, and higgsinos are a high priority for the LHC Run 2 and beyond, and are areas of strong involvement by the IFAE team. During 2020, the IFAE team (A. Juste and C. Moreno) continued to participate in two high-profile supersymmetry searches featuring a high multiplicity of jets originating from the hadronisation of b-quarks (b-jets) and large missing transverse momentum. The first search is focused on the strong production of a pair of gluinos, with each gluino decaying into a neutralino and a tt pair or a bb pair. The second search targets the pair production of higgsinos, with each higgsino decaying into a gravitino and a Higgs boson, which in turn is required to decay into a bb pair. Both searches are based on the full Run 2 dataset and, for the first time, employ state-of-the-art machine-learning techniques, resulting in significant improvements in sensitivity compared to previous analyses. The final results are expected during 2021 and will be part of C. Moreno’s PhD Thesis.

Searches for compositeness

Models of partial compositeness represent another solution to the gauge hierarchy problem, predicting heavy vector-like quarks and new strong interactions resulting in a significant increase of the four top-quark production rate. During 2020 the IFAE team (A. Juste, N. Orlando, and T. Van Daalen) continued to lead the first ATLAS search for single production of a vector-like top quark (T), with the T-quark decaying into a top quark and a Higgs boson or a Z boson. Due to the large mass of the T-quark considered (above 1 TeV), the signal features boosted hadronically decaying SM resonances (W, Z and Higgs bosons, as well as top quarks), which are identified and used to discriminate it against the large background from top-quark pair production in association with jets (tt+jets). The results for this search, which is based on the full Run 2 dataset, are expected early in 2021, and will be part of T. Van Daalen’s PhD thesis. A. Juste is the ATLAS analysis contact for this search.

Figure 4: Comparison between data and prediction after the fit (Post-Fit') for the distribution of the BDT score in the signal region (SR’). The band includes the total uncertainty of the post-fit computation. The ratio of the data to the total post-fit computation is shown in the lower panel. From Eur. Phys. J. C 80 (2020) 1085.
In addition, the IFAE team (A. Juste, N. Orlando, and A. Sonay) continued to participate in the search for four-top-quark (tttt, denoted 4-top) production using the full Run-2 dataset. This analysis will be used to measure the SM 4-top production cross-section, and to probe BSM 4-top production, e.g. via a contact interaction due to an ultra-heavy new mediator particle. The IFAE team is focused on the analysis of events with exactly one lepton, or two opposite-charge leptons, and many jets, at least three of which are b-tagged (referred to as the “1LOS channel”). The team has made critical contributions to the search, such as the development of a novel strategy to improve the simulation of the dominant tt+jets background by using data-based corrections, or the development of a multivariate discriminant based on a boosted-decision-tree (BDT) that has led to a factor of two improvement in the sensitivity for this search, compared to previous analyses. The results for this search are expected during early 2021, and will be part of A. Sonay’s PhD thesis. These results will be combined with those from another published SM 4-top search (Eur. Phys. J. C 80 (2020) 1085) focused on same-charge leptons and trilepton events, which recently reported evidence for this process (see Fig. 4), and to which the IFAE team has also contributed. The IFAE team is also playing a leading role in a search for BSM 4-top production mediated by a new, heavy scalar resonance, again in the 1LOS channel, where it has developed a new approach to improve the modeling of the tt+jets background by the simulation by applying a multidimensional correction derived using a deep Neural Network. This technique appears very promising and will likely be adopted by other physics analyses in ATLAS. The results of this search are expected during 2021 and will also be part of A. Sonay’s PhD thesis.

Searches for leptoquarks

Over the last few years, results from the B-factories and the LHCb experiment show intriguing deviations of ~2–3σ in the ratios $R_K^{(*)}$ and $R_D^{(*)}$, where accurate tests of Lepton Flavor Universality can be performed. Currently, the favored BSM explanation is a leptoquark (LQ) with a mass in the TeV scale, and preferentially coupled to 3rd-generation quarks and 2nd and 3rd generation leptons. During 2019 the IFAE team (S. Kazakos and A. Juste) continued to participate in the first ATLAS search for pair production of a scalar leptoquark decaying into LQ→tτ, where they are leading the analysis of final states with multiple leptons, including τ-leptons. This search, which uses the full Run 2 dataset, will be published during 2020 and will be part of S. Kazakos’ PhD Thesis.

Figure 5: Observed (solid line) and expected (dashed line) 95% CL upper limits on the LQ pair production cross section as a function of the LQ mass resulting from the combination of all analysis channels, assuming 100% branching ratio for LQ🡪tτ decay. From arXiv:2101.11582.


In 2020, members of the IFAE group contributed strongly to the ATLAS Tile calorimeter (TileCal) maintenance during the LS2, to the calorimeter calibration, to the measurement of the ATLAS luminosity, and to the preparation of the detector upgrades for the high-luminosity LHC (HL-LHC) operation.

D. Bogavac took TileCal Run Coordination duties that include maintenance and DQ tests of the read-out, calibrations and management of the Tile operations during the integration ATLAS M-weeks. She had also personally converted the 2018 measurements from slow current integrators, components developed and maintained exclusively by IFAE, to the final corrections of the TleCal energy scale for the ATLAS Run 2 data reprocessing. Danijela’s work was recognized by an ATLAS Outstanding Achievement Award 2020. She continues as Tile Trigger Coordinator in 2021, another role of significant importance for the detector operation. In 2020, R. Rosten had concluded studies to correct for the TileCal phototube non-linearity vs mu for the case of the active HV dividers designed for the HL-LHC, but also deployed in the TileCal E-counters during Run 2. The obtained average corrections are critical for the analysis of the ATLAS luminosity measurements. They are also being used by S. González, who had extended his analysis of the Run 2 luminosity measurement with TileCal (see Fig. 6) to include the 2017 data set. In November 2020, N. Agaras had started feasibility studies on phototube non-linearity in the case of the passive HV dividers and the possibility of extending the luminosity measurement with the TileCal to the A-sample of the calorimeter.

In 2020, following a successful pre-production, the IFAE mechanical workshop entered in the production phase of the mini-drawers for the TileCal HL-LHC upgrade. In particular, all the small components for the production, assembly and tests of IFAE’s share of 104 Tile modules were purchased and procured. All the contracts with providers of the pre-manufactured components were signed. The new CNC milling machine was bought, installed, tested and properly programmed. The mini-drawer production and the modules assembly will take place in 2021-2022.
Figure 6: Ratios of the instantaneous luminosity measured by the Tile E-cell scintillator families to that from track-counting used to evaluate the ATLAS systematic uncertainty on the luminosity measurement in 2018. From ATLAS-COM-CONF-2020-024.


Since the first long shutdown of the LHC and during the whole Run 2, the IFAE ATLAS group has been involved in the Level-1 (L1) topological trigger system (L1Topo), consisting of two electronic boards with FPGA processors programmed to perform real-time event selection based on topological event variables. The group wrote the simulation of the topological trigger algorithms, was responsible of its evolution and provided diagnostic tools to identify sources of discrepancies or hardware malfunctioning.

In preparation for Run 3, the IFAE group took the responsibility of coding the simulation of the newly proposed topological algorithms, and N. Orlando took over the ATLAS-wide coordination of the L1Topo commissioning group. C. Moreno implemented the multiplicity triggers that count L1 trigger objects and will run in one of the three L1Topo boards. P. Martínez and A. Sonay started to update the simulation code to appropriately decode the new L1Muon and L1Calo trigger objects, respectively. In parallel, N. Orlando and I. Riu have been co-editing the paper describing the L1Topo performance in Run 2, which is close to being published.

In parallel, A. Salvador continued to be involved in the tracking algorithm of the tau High Level Trigger signature. He designed a Boosted Decision Tree (BDT) to select the best track to seed the precision tracking algorithm of the tau trigger. He used a large sample of simulated Z🡪ττ events and trained a BDT to distinguish fake from truth tau tracks. Information about energies reconstructed in the various calorimeters, the transverse momentum (pT) and quality of the tracks, and their distance to the calorimeter clusters were used. A. Huanay de Dios, defended his Final Degree Project in Physics on the optimization of this BDT using different signal definitions, input variables and classifiers. Figure 7 shows the tau track trigger efficiency as a function of the pT of the leading offline tau corresponding to various trainings with the AdaBoost classifier. Small differences among the various BDT trainings are observed, giving all an improved efficiency with respect to the method previously used.
Figure 7: Tau tracking trigger efficiency as a function of the transverse momentum (pT) of the leading offline tau, for various trainings with the AdaBoost classifier. The “leading good quality” method was used in Run 2 while the BDT method is proposed for Run 3.