Quark-gluon separation for forward jets with HGCAL information

You-Ying Li^1*, Zheng-Gang Chen¹, Kai-Feng Chen¹

¹Physics of Department, National Taiwan University, Taipei, Taiwan

* Presenter:You-Ying Li, email:yyaustinli@ntu.edu.tw

The technique of the quark-gluon separation plays an important role in the tagging of physical processes in hadron colliders, especially, vector boson scattering processes that contain two forward quark jets. Currently, deep learning in quark-gluon separation has been proven to have spectacular improvement with respect to classical methods such as likelihood, and BDT. Additionally, in the future CMS phase II upgrade, the HGCAL installed on the endcap region provides extra information in the radial direction which could help the quark-gluon discrimination. Therefore, this talk presents the quark-gluon separation in the CMS endcap region through state-of-the-art neural networks such as GNN and transformer. At the same time, the reconstructed objects given the HGCAL detector are also included to benefit the separation power.

Keywords: Quark-gluon separation, Forward jet, Deep learning, HGCAL