Erlangen 2026 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
T: Fachverband Teilchenphysik
T 98: Higgs Physics X
T 98.2: Talk
Friday, March 20, 2026, 09:15–09:30, KH 02.013
Focusing the Inference Lens: Probing the top-Higgs CP Structure with Neural Simulation-Based Inference — •Stefan Katsarov1, Levi Evans1, Alexander Held2, Stephen Jiggins1, Judith Katzy1, Nino Kovacic3, Jay Sandesara2, and Chris Scheulen4 — 1DESY, Hamburg — 2University of Wisconsin-Madison — 3University of Zagreb — 4University of Geneva
Neural Simulation-Based Inference (NSBI) is an emerging statistical framework that leverages modern neural networks as powerful function approximators to achieve the statistical objective of accurately estimating probabilistic relationships between data and parameters of interest. This approach enables inference directly from the full dimensionality of reconstruction-level data.
We implement NSBI to measure the CP structure of the top-Higgs coupling in a ttH and tH enriched signal region. This measurement is directly sensitive to a CP-odd top-Higgs coupling, which has not yet been experimentally excluded. It could provide direct evidence of physics beyond the Standard Model, potentially hinting at an explanation for the observed matter-antimatter asymmetry in the Universe.
The ttH and tH processes exhibit interference effects, leading to degenerate structures in conventional observables. These otherwise prohibitive features lend themselves well to the high-dimensional approach of NSBI. I will outline the methodology used to construct our likelihood ratio function and compare the resulting fits with an analysis that used physics-motivated observables and a conventional classifier.
Keywords: Neural Simulation-Based Inference; CP-Structure; top-Higgs Coupling; Machine Learning; Physics Beyond the Standard Model