arXiv subjects

Timothée Pascal

Publications and source records attributed to Timothée Pascal.

4 recordsLinked to original sources

arXiv2024

SModelS v3: Going Beyond Z2 Topologies

SModelS is a public tool for fast reinterpretation of LHC searches for new physics based on a large database of simplified model results. While previous versions were limited to models with a Z2-type symmetry, such as R-parity conserving supersymmetry, version 3 can now handle arbitrary signal topologies. To this end, the tool was fully restructured and now relies on a graph-based description of simplified model topologies. In this work, we present the main conceptual changes and novel features of SModelS v3, together with the inclusion of new experimental searches for resonant production of spin-1 and spin-0 mediators with decays to quarks or to dark matter. Applying these results to a model containing two mediators, we discuss the interplay of resonance and missing energy searches, and the model's coverage by the currently available simplified model results.

hep-ph↗

arXiv2024

Trained Random Forests Completely Reveal your Dataset

We introduce an optimization-based reconstruction attack capable of completely or near-completely reconstructing a dataset utilized for training a random forest. Notably, our approach relies solely on information readily available in commonly used libraries such as scikit-learn. To achieve this, we formulate the reconstruction problem as a combinatorial problem under a maximum likelihood objective. We demonstrate that this problem is NP-hard, though solvable at scale using constraint programming -- an approach rooted in constraint propagation and solution-domain reduction. Through an extensive computational investigation, we demonstrate that random forests trained without bootstrap aggregation but with feature randomization are susceptible to a complete reconstruction. This holds true even with a small number of trees. Even with bootstrap aggregation, the majority of the data can also be reconstructed. These findings underscore a critical vulnerability inherent in widely adopted ensemble methods, warranting attention and mitigation. Although the potential for such reconstruction attacks has been discussed in privacy research, our study provides clear empirical evidence of their practicability.

cs.LG↗

arXiv2023

Global LHC constraints on electroweak-inos with SModelS v2.3

Electroweak-inos, superpartners of the electroweak gauge and Higgs bosons, play a special role in supersymmetric theories. Their intricate mixing into chargino and neutralino mass eigenstates leads to a rich phenomenology, which makes it difficult to derive generic limits from LHC data. In this paper, we present a global analysis of LHC constraints for promptly decaying electroweak-inos in the context of the minimal supersymmetric standard model, exploiting the SModelS software package. Combining up to 16 ATLAS and CMS searches for which electroweak-ino efficiency maps are available in SModelS, we study which combinations maximise the sensitivity in different regions of the parameter space, how fluctuations in the data in individual analyses influence the global likelihood, and what is the resulting exclusion power of the combination compared to the analysis-by-analysis approach.

hep-ph↗

arXiv2023

SModelS v2.3: enabling global likelihood analyses

We present version 2.3 of SModelS, a public tool for the fast reinterpretation of LHC searches for new physics on the basis of simplified-model results. The main new features are a database update with the latest available experimental results for full Run 2 luminosity, comprising in particular a large variety of electroweak-ino searches, and the ability to combine likelihoods from different analyses. This enables statistically more rigorous constraints and opens the way for global likelihood analyses for LHC searches. The physics impact is demonstrated for the electroweak-ino sector of the minimal supersymmetric standard model.

hep-ph↗