LLP2025: Fifteenth workshop of the Long-Lived Particle Community

Using BDTs as Surrogate Models for BSM searches in ATLAS

5 Jun 2025, 12:20

17m

València - ADEIT

Theory and pheno V

Speaker

Abdelhamid Haddad (Laboratoire de Physique de Clermont-Auvergne (LPCA))

Description

In this talk, I will present a novel approach for the reinterpretation of LHC searches for long-lived particles (LLPs) using Boosted Decision Trees (BDTs). This method allows us to estimate the likelihood that events from a new physics model would have been selected in a prior analysis, based solely on truth-level information. I will discuss the development and validation of method, including its integration within the HackAnalysis framework. Furthermore, I will explore its application to an extended version of the HAHM model with asymmetric LLPs (in mass and lifetime), and highlight the promising results obtained.

Presentation materials

BDT as a Surrogate Model _ Valencia.pdf