Erlangen 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 71: Data, AI, Computing, Electronics VII

T 71.6: Talk

Thursday, March 19, 2026, 17:30–17:45, KH 00.024

Symbolic Regression for the Extraction of Detector Response Formulas — •Johannes Merten and Johannes Erdmann — III. Physikalisches Institut A, RWTH Aachen University

Detector simulations are an essential part of high-energy physics research, enabling the interpretation of experimental data and the design of future experiments. Full-scale simulations, such as those based on GEANT4, provide high-fidelity representations of particle interactions within detectors but are computationally expensive. To facilitate large-scale analyses, fast simulation frameworks employ parameterizations to approximate detector responses. These parameterizations rely on simplified functional forms that may not fully capture the underlying complexities of the detector response, which may lead to systematic biases.

In this work a data-driven approach is proposed to derive these interpretable parametrizations directly from high-fidelity simulation data using Normalizing Flows and Kolmogorov-Arnold Networks.

Keywords: Detector response; XAI; GEANT4; Normalizing Flows; Kolmogorov-Arnold Networks