DPG Phi
Verhandlungen
Verhandlungen
DPG

Karlsruhe 2024 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 99: Di-Higgs 3 (bbWW)

T 99.6: Vortrag

Donnerstag, 7. März 2024, 17:15–17:30, Geb. 30.41: HS 1

Separating tt and HH end states using neural networks — •Youn Jun Cho, Christoph Ames, Stephanie Goetz, Edis Hrustanbegovic, Lars Linden, Celine Stauch, Lukas von Stumpfeldt, and Otmar Biebel — Ludwig-Maximilian University of Munich

Immense research has been conducted on the Higgs Boson since its discovery. However, the sheer amount of background events in high energy particle colliders complicate the study of its physical properties. Research on the Higgs self interactions is no exception. The corresponding cross section is small compared to many competing processes with similar final states. For example, the Higgs self interactions and the top anti-top quark pair decays can have equivalent final states, e.g. bbW+W, but the cross sections are roughly 30 femtobarns and 1 microbarns, respectively, such that the top anti-top quark distributions reach into those of the Higgs pair production. To separate these two, Pytorch neural networks were applied onto the data simulated with the MCatNLO+Herwig event generator. Given a reasonable computation time, MCatNLO+Herwig could not generate enough top background events to train the neural network. Therefore, certain features of these data beyond kinematics were modified in order to generate sufficient training data. Overall, the neural network was effective in separating the two end states.

Keywords: Higgs boson self interactons; Top quark decays; Neural networks

100% | Mobil-Ansicht | English Version | Kontakt/Impressum/Datenschutz
DPG-Physik > DPG-Verhandlungen > 2024 > Karlsruhe