Periodic Reporting for period 2 - PBSP (Physics Beyond the Standard Proton)
Période du rapport: 2022-04-01 au 2023-09-30
The research efforts of the Physics Beyond the Standard Proton (PBSP) group focus on particle physics phenomenology, which is the field of theoretical particle physics that is mostly interested in the observable consequences of the theoretical understanding of the fundamental building blocks of Nature and of their mutual interactions. The objectives of our research are crucial in order to advance the interpretation of the experimental measurements of the ultra-relativist processes taking place at particle colliders, in particular at the Large Hadron Collider (LHC) at CERN.
The high-energy physics community finds itself at a particularly exciting crossroads, in which hints for new phenomena, such as a new kind of force that features different couplings with the different lepton families (dubbed "fifth force" by the newspapers), have been identified. Such hints have been suggested both by an experiment at Fermilab measuring the anomalous magnetic dipole moment of muons and by the precise flavour measurements performed by the LHCb detector at CERN. It is imperative to make the most of this window of opportunity and to statistically corroborate or falsify the presence of deviations from the Standard Model (SM) predictions coming from possible New Physics effects.
In order to maximise the chances of success, a robust determination of the significance of these signals is paramount. In particular, the outcome of this quest of the whole particle physics community requires the highest possible level of precision in three complementary components: the experimental data, the theoretical predictions and - crucially - a robust framework to globally interpret all subtle deviations from the SM predictions that might arise. Such a framework is crucial in order to confirm and strengthen or, alternatively, weaken and disprove the signals that have been observed so far.
The final outcome of PBSP's research project is the provision of two essential ingredient of such a much-needed statistical framework. On the one hand, PBSP will provide sets of Parton Distribution Functions (PDFs), i.e. the functions that parametrise the protons in terms of their elementary constituents (quarks and gluons), which are free of possible contaminations arising from New Physics. Given that PDFs are a crucial component of any theoretical predictions at the LHC, the delivery of such PDF set is crucial to achieve unbiased interpretation of the experimental observations at CERN. Second, PBSP will yield robust determination of Effective Field Theory (EFT) coefficients that parametrise the effect of heavy new particles at the energies probed by the LHC, in a model-independent way. The final goal of PBSP is to develop a new statistical methodology to devise global fits that are to simultaneously constrain both the proton structure and new physics dynamics from the data collected at the LHC, so that a global interpretation of the LHC data can be achieved.
The high-energy physics community finds itself at a particularly exciting crossroads, in which hints for new phenomena, such as a new kind of force that features different couplings with the different lepton families (dubbed "fifth force" by the newspapers), have been identified. Such hints have been suggested both by an experiment at Fermilab measuring the anomalous magnetic dipole moment of muons and by the precise flavour measurements performed by the LHCb detector at CERN. It is imperative to make the most of this window of opportunity and to statistically corroborate or falsify the presence of deviations from the Standard Model (SM) predictions coming from possible New Physics effects.
In order to maximise the chances of success, a robust determination of the significance of these signals is paramount. In particular, the outcome of this quest of the whole particle physics community requires the highest possible level of precision in three complementary components: the experimental data, the theoretical predictions and - crucially - a robust framework to globally interpret all subtle deviations from the SM predictions that might arise. Such a framework is crucial in order to confirm and strengthen or, alternatively, weaken and disprove the signals that have been observed so far.
The final outcome of PBSP's research project is the provision of two essential ingredient of such a much-needed statistical framework. On the one hand, PBSP will provide sets of Parton Distribution Functions (PDFs), i.e. the functions that parametrise the protons in terms of their elementary constituents (quarks and gluons), which are free of possible contaminations arising from New Physics. Given that PDFs are a crucial component of any theoretical predictions at the LHC, the delivery of such PDF set is crucial to achieve unbiased interpretation of the experimental observations at CERN. Second, PBSP will yield robust determination of Effective Field Theory (EFT) coefficients that parametrise the effect of heavy new particles at the energies probed by the LHC, in a model-independent way. The final goal of PBSP is to develop a new statistical methodology to devise global fits that are to simultaneously constrain both the proton structure and new physics dynamics from the data collected at the LHC, so that a global interpretation of the LHC data can be achieved.
In the first part of the project, we have achieved two crucial goals of the PBSP program.
On the one hand, we have identified a physics scenario in the High-Luminosity phase of the LHC, in which the interplay between new physics effects and the structure of the proton is strong and cannot be neglected if a solid interpretation of the experimental data is to be achieved. This result was crucial in order to motivate the whole effort of the PBSP collaboration. We showed the scientific community that the goals of PBSP are essential to achieve the correct interpretation of the LHC data and we discussed our findings at many international venues.
On the other hand, we have given an essential contribution to the delivery of the first set of Parton Distribution Functions (PDFs) based on a fully global dataset comprising nearly 5000 datapoints and on modern deep learning techniques, the NNPDF4.0 set. We also published and made available a software framework underlying such an analysis.
Building and expanding on the deep-learning techniques underlying the NNPDF4.0 analysis, PBSP built a completely new methodology to perform the simultaneous determination of the PDFs and of the Wilson coefficients of the SM Effective Field Theory (SMEFT) expansion. The novel methodology, dubbed simuNET, is based on modern machine learning techniques and on the usage of new statistical packages. We first tested it and successfully validated in the context of the fit of high-mass Drell-Yan data and proved that the methodology yields robust results, that can be validated against those obtained with less efficient techniques.
We then expanded the application of simuNET to the simultaneous fit of the proton’s PDFs and up to 25 SMEFT degrees of freedom in the top sector and achieved a state-of-the-art interpretations of top quark data based on the broadest LHC experimental dataset to date, therefore fully exploiting the information contained in the legacy Run II top quark measurements both in the context of PDF and SMEFT and for the first time in their joint interpretation.
Group members have also worked on related projects that advance our understanding of the interplay between the proton structure and New Physics, as well as on the possibility of exploring a large parameter space via a joint Mont Carlo sampling technique.
As far as the first area of investigation is concerned, we studied the sensitivity of the High-Luminosity LHC to a light baryonic dark photon, primarily coupled to quarks, as a constituent of the proton. This was achieved by allowing for a dark photon PDF and assessing how this assumption constrain the dark photon parameter space. As far as the second area of investigation is concerned, we built new methodology that exploits Monte Carlo sampling techniques for propagating experimental uncertainties into the PDF space to the space of theory parameters. A prior probability was assigned to each theoretical assumptions and a posterior probability is obtained by selecting replicas that satisfy fit-quality criteria.
Both achievements represent unforeseen but welcomed developments, which expand the prospects of discovery of the PBSP effort.
On the one hand, we have identified a physics scenario in the High-Luminosity phase of the LHC, in which the interplay between new physics effects and the structure of the proton is strong and cannot be neglected if a solid interpretation of the experimental data is to be achieved. This result was crucial in order to motivate the whole effort of the PBSP collaboration. We showed the scientific community that the goals of PBSP are essential to achieve the correct interpretation of the LHC data and we discussed our findings at many international venues.
On the other hand, we have given an essential contribution to the delivery of the first set of Parton Distribution Functions (PDFs) based on a fully global dataset comprising nearly 5000 datapoints and on modern deep learning techniques, the NNPDF4.0 set. We also published and made available a software framework underlying such an analysis.
Building and expanding on the deep-learning techniques underlying the NNPDF4.0 analysis, PBSP built a completely new methodology to perform the simultaneous determination of the PDFs and of the Wilson coefficients of the SM Effective Field Theory (SMEFT) expansion. The novel methodology, dubbed simuNET, is based on modern machine learning techniques and on the usage of new statistical packages. We first tested it and successfully validated in the context of the fit of high-mass Drell-Yan data and proved that the methodology yields robust results, that can be validated against those obtained with less efficient techniques.
We then expanded the application of simuNET to the simultaneous fit of the proton’s PDFs and up to 25 SMEFT degrees of freedom in the top sector and achieved a state-of-the-art interpretations of top quark data based on the broadest LHC experimental dataset to date, therefore fully exploiting the information contained in the legacy Run II top quark measurements both in the context of PDF and SMEFT and for the first time in their joint interpretation.
Group members have also worked on related projects that advance our understanding of the interplay between the proton structure and New Physics, as well as on the possibility of exploring a large parameter space via a joint Mont Carlo sampling technique.
As far as the first area of investigation is concerned, we studied the sensitivity of the High-Luminosity LHC to a light baryonic dark photon, primarily coupled to quarks, as a constituent of the proton. This was achieved by allowing for a dark photon PDF and assessing how this assumption constrain the dark photon parameter space. As far as the second area of investigation is concerned, we built new methodology that exploits Monte Carlo sampling techniques for propagating experimental uncertainties into the PDF space to the space of theory parameters. A prior probability was assigned to each theoretical assumptions and a posterior probability is obtained by selecting replicas that satisfy fit-quality criteria.
Both achievements represent unforeseen but welcomed developments, which expand the prospects of discovery of the PBSP effort.
The main achievements reported above go beyond the current state-of-the art. In particular, we have developed a novel framework that allows the simultaneous determination of the subnuclear proton structure in terms of its Parton Distribution Functions (PDFs) and any external parameters that enter theoretical predictions within the Standard Model (SM) and beyond it. We have proven that the interplay between the fit of PDFs of the fit of model-independent parametrization of New Physics effects within the SMEFT framework are intertwined and that their interplay will not be negligible in the high-luminosity phase of the LHC.
Looking forward we plan to (i) extend the novel simuNET methodology for simultaneous fits to be able to accommodate quadratic corrections in terms of the SMEFT effects; (ii) devise a methodology to spot and measure possible contamination from New Physics in the high energy data that are included in the fits of PDFs and formulate recommendations for PDF fitters based on our study; (iii) provide a set of public tools to assess the interplay and correlations between several parameters that enter theoretical predictions and must be determined from the data, starting from the PDF and SMEFT coefficients, and going beyond to include the SM precision parameters.
Looking forward we plan to (i) extend the novel simuNET methodology for simultaneous fits to be able to accommodate quadratic corrections in terms of the SMEFT effects; (ii) devise a methodology to spot and measure possible contamination from New Physics in the high energy data that are included in the fits of PDFs and formulate recommendations for PDF fitters based on our study; (iii) provide a set of public tools to assess the interplay and correlations between several parameters that enter theoretical predictions and must be determined from the data, starting from the PDF and SMEFT coefficients, and going beyond to include the SM precision parameters.