[행사/세미나] [Colloquium] Oct 11(Wed.) Hunt for the dark matter with the upgraded CMS detector and deep learning technique
- 물리학과
- 조회수1563
- 2023-10-05
이번 학기 여섯 번째 콜로퀴움은 경북대학교 물리학과 문창성 교수님을 모시고 콜로퀴움을 개최하오니, 학과 구성원 여러분의 많은 참여를 부탁 드립니다.
아 래
1. Title: Hunt for the dark matter with the upgraded CMS detector and deep learning technique
2. Speaker: 문 창 성 교수님 (경북대학교 물리학과)
3. Date & Time: Oct 11(Wed.) 2023. 4:30 PM
4. Place: Natural Science 1, Room No. 31214
5. Abstract: The searche for dark matter (DM) particles at the CERN Large Hadron Collider (LHC) is a thriving research field after the discovery of the Higgs boson as there is still no evidence for non-gravitational interactions between the DM and Standard Model (SM) particles. If such interactions exist, the DM particles could be produced at the LHC. Since the DM particles themselves do not produce any signal in the detector, one way to observe them is when they are produced in association with a visible SM particle X(=g, q, γ, Z, W, or h). In this talk, we present the main strategy for the search for DM with the upgraded CMS detector and the deep learning technique at the LHC.
The CERN has been preparing a major upgrade of the accelerators to increase the capability for new physics searches and high precision measurements, the so-called High-Luminosity LHC (HL-LHC). In order to fully exploit the scientific potential of the HL-LHC, the CMS collaboration is upgrading the detector and improving the ability of the apparatus to isolate and precisely measure the products of the most interesting collisions. Here, we discuss the Phase-2 Upgrade of the CMS MIP Timing Detector (MTD) for the HL-LHC to maintain the overall detector performance at the HL-LHC.
In addition, we perform the distributed training based on the Nurion supercomputer at KISTI to demonstrate that the deep learning model can be optimized and scaled effectively on a multi-node HPC system. The scalability of this deep learning model on the Nurion HPC architectures is reported.