The Higgs boson and the top quark in the ATLAS detector: gateway to new physics

为了通过宇宙的基本成分来理解宇宙，欧洲核子研究中心的大型强子对撞机（LHC）提供了重要的结果，证实了当前的粒子物理理论——标准模型（SM）。希格斯玻色子的发现是一项了不起的成就：它解释了基本粒子是如何获得质量的。尽管在量子尺度上对基本粒子物理定律进行了非常精确的描述，但SM无法解释我们宇宙中的主要宇宙现象。找到超越SM的“新物理学”的标志将是一个突破性的发现，并为更完整地描述自然提供指导。探测新物理学特征的一个有希望的途径是测量希格斯玻色子与最重的物质粒子——顶夸克的相互作用。相互作用强度，称为汤川耦合，对新物理非常敏感，任何与理论预测的偏差都会揭示非sm现象。我领导了对与顶夸克对（ttH）相关的希格斯玻色子产生的观测背景评估工作，并使用了大型强子对撞机的多用途ATLAS探测器。第t种生产模式是可以直接获得顶夸克汤川耦合的物理过程。然而，从LHC数据中提取ttH信号极具挑战性；主要的背景过程很难精确地建模。传统上使用蒙特卡罗（MC）技术进行的模拟由于理论限制而具有固有的偏差。我建议部署先进的机器学习工具来更精确地测量顶夸克汤川耦合，以充分利用即将到来的LHC运行3（2021 - 2023）的全部潜力。生成对抗网络（GANs）是一种有前途的深度学习模型，在高能物理中有相关的应用。我提出了一种渐进的方法来实现gan，以减轻较大的理论不确定性，并使ttH分析减少对MC模拟的依赖。这是为2026年开始的高亮度大型强子对撞机项目做准备，在这个项目中，具有高统计数据的MC的需求几乎增加了10倍。gan生成的碰撞比MC生成的碰撞要快得多，计算强度也小得多。准备高亮度大型强子对撞机还需要升级探测器，我建议利用我在粒子探测器表征方面的专业知识加入约克大学/多伦多集群的努力，该集群负责生产和测试未来ATLAS内部跟踪器的500个传感器单元（硅和电子）。约克大学最适合作为对未通过标准质量保证测试的部件进行详细评估和回收的专用站点。我建议设计和建设评价实验室，并监督生产过程中的再评价和回收工作。

In the quest to understand the universe through its elementary constituents, CERN's Large Hadron Collider (LHC) has delivered important results confirming the current theory of particle physics, the Standard Model (SM). The discovery of the Higgs boson was a remarkable achievement: it explained how elementary particles acquired their masses. Despite being a very accurate description of the fundamental particle physics laws at the quantum scale, the SM fails to explain major cosmological phenomena in our universe. Finding a sign of "new physics" extending beyond the SM would be a ground-breaking discovery, and a guide toward a more complete description of nature. A promising gateway to probe for signatures of new physics is the measurement of the Higgs boson's interaction with the heaviest particle of matter: the top quark. The interaction strength, called Yukawa coupling, is very sensitive to new physics, and any deviation with respect to the theoretical prediction would uncover non-SM phenomena. I led the effort on the background evaluation towards the observation of the Higgs boson production in association with a top quark pair, or ttH, with the multipurpose ATLAS detector at the LHC. The ttH production mode is the physics process that can give direct access to the top quark Yukawa coupling. However, extracting the ttH signal from the LHC data is extremely challenging; the main background processes are very difficult to accurately model. The simulations, traditionally performed using Monte Carlo (MC) techniques, are intrinsically biased due to theoretical limitations. I propose to deploy advanced machine learning tools to measure with better precision the top quark Yukawa coupling in order to exploit the full potential of the incoming LHC Run 3 (2021 - 2023). Generative Adversarial Networks (GANs) are promising deep learning models with relevant applications in high energy physics. I propose a gradual approach to implement GANs for mitigating the large theoretical uncertainties and make the ttH analysis less reliant on MC simulations. This is in preparation for the High-Luminosity LHC program starting 2026, where the needs of MC with high statistics is almost multiplied by ten. GAN-generated collisions would be faster to produce than MC and much less computationally intensive. Preparing for the High-Luminosity LHC also requires detector upgrades and I propose to use my expertise in particle detector characterization to join the effort of the York University/Toronto cluster, which is in charge of producing and testing 500 sensor units (silicon and electronics) of the future ATLAS inner tracker. York University is best suited to be the dedicated site for detailed evaluation and recovery of components failing standard quality assurance tests. I propose to design and construct the evaluation laboratory, and oversee the reevaluation and recovery efforts during production.