Report on 2003.07868v2

Waleed Abdallah, Shehu AbdusSalam, A. I. Ahmadov, Amine Ahriche, M Schirmer, B. C. Allanach, Jack Y. Araz, Alexandre Arbey, Chiara Arina, Peter Athron, Emanuele Bagnaschi, Y. Bai, Michael J. Baker, Csaba Balázs, Daniele Barducci, P. Bechtle, Aoife Bharucha, A. G. Buckley, J. M. Butterworth, Haiying Cai, C. Campagnari, Cari Cesarotti, M. Chrząszcz, A. Coccaro, E. Conte, Jonathan M. Cornell, L. D. Corpe, M. Danninger, Luc Darmé, Aldo Deandrea, Nishita Desai, Barry M. Dillon, C. Doglioni, Matthew J. Dolan, Juhi Dutta, John Ellis, Sebastian A. R. Ellis, M. Feickert, Nicolas Fernandez, S. Fichet, Thomas Flacke, Benjamin Fuks, A. Geiser, M. H. Genest, Akshay Ghalsasi, Tomás E. Gonzalo, Mark D. Goodsell, Stefania Gori, P. Gras, Admir Greljo, Diego Guadagnoli, S. Heinemeyer, L. Heinrich, Jan Heisig, Deog Ki Hong, T. Hryn’ova, Katri Huitu, P. Ilten, Ahmed Ismail, Adil Jueid, Felix Kahlhoefer, J. Kalinowski, Jernej F. Kamenik, D. Kar, Yevgeny Kats, Charanjit K. Khosa, V.a. Khoze, T. Klingl, Pyungwon Ko, Kyoungchul Kong, Wojciech Kotlarski, Michael Krämer, Sabine Kraml, Suchita Kulkarni, Anders Kvellestad, C. Lange, K. Lassila-Perini, Seung Lee, André Lessa, Zhen Liu, L. Lloret Iglesias, J. Lorenz, D. M. Mac Donell, F. Mahmoudi, J. Mamuzic, Andrea Carlo Marini, P. Markowitz, Pablo Martinez Ruiz, Del Arbol, D. W. Miller, V. A. Mitsou, Stefano Moretti, Marco Nardecchia, S. Neshatpour, Dao Thi Nhung, P. Osland, P. Owen, Orlando Panella, А. А. Панков, Myeonghun Park, Torre, R., C. Collard, A. Colaleo, Florian Goertz, Joachim Kopp, Manuel E. Krauss, M. TytgatOpenalex Unknown 2020 · DOI · arXiv · OpenAlex

Abstract

This report summarizes the status and recommendations of the LHC Reinterpretation Forum regarding the preservation and reuse of experimental data from the Large Hadron Collider (LHC). As the particle physics community seeks to maximize the scientific return on the LHC investment, ensuring that experimental results can be tested against new theoretical models—a process known as reinterpretation or recasting—is essential. We review current methodologies, technical implementations, and the infrastructure supporting data preservation, while providing a framework for future experimental publications.

Introduction: The Challenge of Data Longevity

LHC experiments like ATLAS and CMS are designed to search for physics beyond the Standard Model (BSM). However, experimentalists typically test only a limited set of theoretical benchmarks. Because the theoretical landscape is constantly evolving, many models remain untested against existing data.

Without detailed auxiliary information, experimental results risk becoming "black boxes" whose utility expires shortly after publication. The LHC Reinterpretation Forum addresses this by establishing standards that allow the broader community to apply existing data to new theoretical contexts, effectively extending the "scientific shelf life" of LHC results.

Methodologies for Reinterpretation

The report identifies two primary workflows for recasting LHC data:

  1. The Simplified Model Approach: This method utilizes published cross-section limits and efficiency maps. It is computationally efficient and allows for the rapid testing of new models that share topological similarities with the original experimental search. Tools like SModelS facilitate this by comparing new theories against a database of simplified results.
  2. Full Monte Carlo (MC) Simulation (Recasting): This involves reproducing the entire experimental analysis logic within public software frameworks such as Rivet, MadAnalysis 5, or CheckMATE. These frameworks run on simulated events from a new theory to determine if the model is excluded by the data.

A significant advancement highlighted in the report is the shift toward publishing Full Likelihoods. By using tools like pyhf to release JSON-serialized likelihoods, experiments allow external researchers to perform statistical inference with a level of precision previously reserved for internal experimental collaborations.

Infrastructure and Data Preservation

Reinterpretation relies on a robust ecosystem of data repositories and simulation tools:

  • HEPData: The primary repository for the numerical results of experimental papers.
  • CERN Open Data Portal: This platform provides access to petabytes of raw and processed collision data. Notably, the CMS collaboration has released over 2PB of data to the public.
  • Standard Model Backgrounds: The authors advocate for the public sharing of massive simulated background samples. Sharing these samples reduces the immense computational burden on theorists who would otherwise need to reproduce these simulations from scratch.

Key Insights for Future Analyses

To ensure results remain interpretable, the Forum emphasizes several critical technical requirements:

  • Correlation and Covariance: It is impossible to accurately combine results from different signal regions without knowing their uncertainties' correlations. The report pushes for the publication of full covariance matrices or orthogonal error sources.
  • The Machine Learning Gap: Modern analyses increasingly rely on Deep Learning for signal selection. However, if the model weights and architectures (e.g., in ONNX format) are not published, the analysis becomes nearly impossible to reinterpret. Establishing standards for sharing ML models is a high priority for the post-Run 2 era.
  • Long-Lived Particles (LLPs): Particles that decay at a distance from the primary interaction point present unique challenges. Their detection efficiency depends heavily on specific detector geometry, which is rarely fully public, making these searches some of the most difficult to recast.

Global Analyses and Software Tools

The report catalogs mature software tools that have enabled hundreds of new studies. Beyond recasting individual searches, frameworks like GAMBIT and MasterCode perform "Global Fits," combining data from the LHC, dark matter experiments, and cosmology to identify the most viable regions of BSM parameter space. Additionally, the Contur tool utilizes model-independent Standard Model measurements to look for deviations that might signal new physics.

Conclusion and Recommendations

The Forum concludes with ten core recommendations for the particle physics community. Central to these is the mandate that numerical data and executable "pseudocode" for every analysis be made available at the time of publication. By adopting these standards, the community ensures that the LHC's legacy will continue to drive discovery for decades, regardless of how theoretical fashions evolve.

Summary updated

Abstract

We report on the status of efforts to improve the reinterpretation of searches and measurements at the LHC in terms of models for new physics, in the context of the LHC Reinterpretation Forum. We detail current experimental offerings in direct searches for new particles, measurements, technical implementations and Open Data, and provide a set of recommendations for further improving the presentation of LHC results in order to better enable reinterpretation in the future. We also provide a brief description of existing software reinterpretation frameworks and recent global analyses of new physics that make use of the current data.
BibTeX
@inproceedings{abdallah2020report,
  author = {Waleed Abdallah and Shehu AbdusSalam and A. I. Ahmadov and Amine Ahriche and M Schirmer and B. C. Allanach and Jack Y. Araz and Alexandre Arbey and Chiara Arina and Peter Athron and Emanuele Bagnaschi and Y. Bai and Michael J. Baker and Csaba Balázs and Daniele Barducci and P. Bechtle and Aoife Bharucha and A. G. Buckley and J. M. Butterworth and Haiying Cai and C. Campagnari and Cari Cesarotti and M. Chrząszcz and A. Coccaro and E. Conte and Jonathan M. Cornell and L. D. Corpe and M. Danninger and Luc Darmé and Aldo Deandrea and Nishita Desai and Barry M. Dillon and C. Doglioni and Matthew J. Dolan and Juhi Dutta and John Ellis and Sebastian A. R. Ellis and M. Feickert and Nicolas Fernandez and S. Fichet and Thomas Flacke and Benjamin Fuks and A. Geiser and M. H. Genest and Akshay Ghalsasi and Tomás E. Gonzalo and Mark D. Goodsell and Stefania Gori and P. Gras and Admir Greljo and Diego Guadagnoli and S. Heinemeyer and L. Heinrich and Jan Heisig and Deog Ki Hong and T. Hryn’ova and Katri Huitu and P. Ilten and Ahmed Ismail and Adil Jueid and Felix Kahlhoefer and J. Kalinowski and Jernej F. Kamenik and D. Kar and Yevgeny Kats and Charanjit K. Khosa and V.A. Khoze and T. Klingl and Pyungwon Ko and Kyoungchul Kong and Wojciech Kotlarski and Michael Krämer and Sabine Kraml and Suchita Kulkarni and Anders Kvellestad and C. Lange and K. Lassila-Perini and Seung Lee and André Lessa and Zhen Liu and L. Lloret Iglesias and J. Lorenz and D. M. Mac Donell and F. Mahmoudi and J. Mamuzic and Andrea Carlo Marini and P. Markowitz and Pablo Martinez Ruiz and Del Arbol and D. W. Miller and V. A. Mitsou and Stefano Moretti and Marco Nardecchia and S. Neshatpour and Dao Thi Nhung and P. Osland and P. Owen and Orlando Panella and А. А. Панков and Myeonghun Park and Torre, R. and C. Collard and A. Colaleo and Florian Goertz and Joachim Kopp and Manuel E. Krauss and Jeanette Miriam Lorenz and M. Tytgat},
  title = {{Report on 2003.07868v2}},
  booktitle = {Openalex Unknown},
  year = {2020},
  doi = {10.21468/scipost.report.1652},
}

Loading related content...

Figure