University

Since the beginning of my PhD, the main focus of my research has been the in-depth study of the internal structure of the proton. I have participated in several measurements of the Parton Distribution Functions (PDF) of the proton, both with and without ATLAS Collaboration, and have become an expert—and subsequently a developer—of the xFitter framework, an open-source tool commonly used in PDF studies. I was the convenor of the ATLAS PDF Fit Forum and the main contributor to the first PDF fit analysis performed by an LHC collaboration using multiple datasets (including those at 13 TeV, not yet used in the most recent global PDF analyses) and evaluating the impact of systematic uncertainty correlations between different datasets. This is something that global fit groups cannot do, as it requires in-depth knowledge of the experimental details of the input measurements. This work has provided crucial information to the entire PDF community, and has already been incorporated into the next version of the global fits.

In parallel with my studies on PDFs, I have been actively involved in analyses particularly relevant to the extraction of fundamental parameters of the Standard Model. I have worked on the measurement of the cross section of the low-mass Drell–Yan process up to mll = 7 GeV (the first analysis of this type in ATLAS). This analysis explores kinematic regions complementary to those accessible in the LHCb and HERA experiments, where new effects in both perturbative and non-perturbative QCD (such as logarithmic increases at small x or small pT) can become important.

I am currently convenor of the ATLAS W/Z Physics Group and responsible for the precision measurement of the W boson production cross section in association with a c quark, focusing on the muon decay channel. Since this channel is dominated by processes initiated by strange quarks, these data can help clarify the debate on the level of suppression of the strange component at low Bjorken x. In this analysis, the Soft Muon Tagging (SMT) technique is used to identify the hadronic jet that has a spatially associated muon. I am one of the main developers of this algorithm, which was also used in a recent measurement of the top quark mass, exploiting the semileptonic decays of b-containing hadrons produced in the top quark decay chain. The results of this study are characterized by reduced systematic uncertainties and are largely independent of conventional analyses of the top quark mass in the lepton + jets channel.

Furthermore, I am currently contributing to measurements of the tt/Z cross section ratio and inclusive W and Z cross sections with initial Run 3 data. In both analyses, I have generated advanced theoretical predictions at NNLO QCD + NLO EW order for different sets of PDFs. In addition, I am involved in the measurement of the Z boson mass, analyzing higher-order QED and EW corrections to the Z boson resonance line; I will soon also be involved in muon calibration studies, deriving an additional correction to the sagitta bias using the “pseudo-mass asymmetry” method.

I am also focusing on improving the modeling of pT(Z) and pT(W) using state-of-the-art N3LO + aN4LL calculations, performing fits to numerous ATLAS and non-ATLAS measurements to better constrain the non-perturbative corrections at very small transverse moments (pT < 5 GeV) introduced in the calculations. This work could lead to significant improvements in future determinations of the W mass. I am also interested in finalizing the measurement of the leptonically effective electroweak mixing angle via Z boson decays. In the past, I have generated theoretical predictions for the measurement of the W charge asymmetry in ATLAS at 8 TeV and studied the potential impact of this determination. Since PDF uncertainties are the main source of error in this measurement, further constraints could lead to a significant reduction in the uncertainty on sin²(θW).

During this period, I also had the opportunity to contribute to other tasks in ATLAS. I participated in the development and installation of the Phase 1 upgrade of the high-performance Resistive Plate Chambers organized in triplets, specifically designed for the barrel-forward transition region of the ATLAS detector (project BIS78).

In addition, over the past two years, I have been deeply involved in trigger operations, serving as an on-call expert for both the online trigger and the High Level Trigger (HLT) menu. Until recently, I was the ATLAS Express Stream Coordinator, responsible for defining the express stream menu that provides the data used in the initial data quality assessment, as well as monitoring the stream's bandwidth consumption to ensure that the set objectives are met.

I am also exploring the possibility of running ATLAS trigger software on GPUs or FPGAs. Developing HLT software for the HL-LHC on these new technologies could enable faster processing speeds than using CPUs and allow for further developments, such as applying machine learning to trigger-level tracking algorithms. These potential benefits are currently being investigated by the community. In this context, I analyzed the performance in terms of execution times of commercial accelerator cards in the context of the New Small Wheel HLT. I studied the performance obtained by running three machine learning models (a Deep Neural Network for cluster position reconstruction, a Convolutional Neural Network, and a Recurrent Neural Network for muon pattern recognition) on Xilinx Alveo U50, U250, and VCK5000 Versal cards. I compared the inference times obtained between CPUs, GPUs, and the aforementioned cards using development tools and commercial hardware.