Searching for Dark Matter using data from LHC Run 2 and Run 3 at the ATLAS experiment

汤姆博士项目的目的是进行世界领先的分析，寻找暗物质（DM）和新粒子介导其与标准模型（SM）相互作用的证据。大量的天体物理学证据表明DM的存在，但对它的粒子性质却一无所知：它不能在SM中被解释，并且仍然是物理学中最大的开放问题之一。SM的几个扩展假设稳定的、电中性的、弱相互作用的大质量粒子作为DM的候选者，它们可能在LHC的高能碰撞中产生。一旦产生，该DM将逃脱检测，从而在检测器的测量的横向动量（ETmiss）中产生不平衡。在LHC的假设过程中，一个或多个SM粒子被反推到DM上，导致“SM + ETmiss”特征。一个非常有趣和热门的特征是暗希格斯玻色子，它负责给暗物质和潜在的扩展暗扇区提供质量，并通过标量门户与SM物质相互作用。这些双介质暗物质模型没有得到很好的探索，并产生了一些有趣的新的和未被发现的签名。具体来说，对于质量超过250 GeV的暗希格斯粒子，衰变为两个希格斯玻色子是第二大贡献，其中高动量希格斯粒子的衰变和暗物质的ETmiss都可以很容易地单独重建。这种通过介质衰变为暗物质的共振双希格斯粒子产生显著ETmiss的特征没有专门的实验分析。学生的主要初始重点是分析这种最终状态，并为这些共振di-Higgs加上ETmiss签名开发第一个ATLAS分析，建立在现有的DM搜索单希格斯和超对称希格斯的基础上，其中两个希格斯玻色子在一个事件中独立产生。由于最终态的复杂性和信号的多重运动学处理，将研究利用机器学习技术，如分类和质量回归，以提供额外的灵敏度。对这一结果的解释将与其他暗希格斯最终态相同，如低质量的bb和高质量的WW，允许组合或汇总以给出ATLAS对这些模型的全部灵敏度。这个特征很重要，这是许多模型的一个普遍特征，其中扩展的希格斯扇区介导了与暗物质的相互作用。或者，他将有机会在他博士学位的后半部分继续这项分析工作，利用运行3的数据为旗舰暗物质搜索做出贡献，因为2023年和2024年运行3的高亮度和灵敏度可用于实验分析。最大化我们对包含希格斯衰变为b夸克对的最终状态的灵敏度是LHC的一个有趣问题。在高希格斯或玻色子动量下，两个b夸克靠得很近，可以被捕获在一个大喷流中。高动量b夸克喷流的识别是重要的，因为使用这些“希格斯喷流”中存在的全部信息-存在两个子喷流，每个子喷流被确定为可能包含b强子，喷流的整体结构及其运动学。ATLAS已经取得了实质性的进展，开发了一个“X->bb”标记器，并正在使用先进的机器学习算法进行工作，以提供一种识别和分类这些喷流的方法，有效地发现希格斯玻色子衰变的喷流，并大大改善了对顶夸克和轻夸克QCD背景的排斥。Tom将有机会成为首批使用这些算法进行分析的人员之一，并理解和贡献推动这些算法的机器学习算法。他将在CERN度过他的第二年，尽可能接近实验室，他的分析团队和Run-3数据采集，并在ATLAS合作中获得可见性。

The aim of Tom's PhD project is to make world-leading analyses searching for evidence of Dark Matter (DM) and new particles mediating its interactions with the Standard Model (SM). Overwhelming astrophysical evidence now suggests the existence of DM, yet nothing is known of its particle nature: it cannot be accounted in the SM and remains one of the largest open questions in physics.Several extensions of the SM postulate stable, electrically neutral, weakly interacting massive particles as DM candidates, which could be produced in the high energy collisions of the LHC. Once produced, this DM would escape detection, producing an imbalance in the measured transverse momentum (ETmiss) of the detector. A wide class of models probed at the LHC postulate processes wherein one or more SM particles are produced recoiling against DM, resulting in a "SM + ETmiss" signature.A highly interesting and topical signature is that of a dark Higgs boson, responsible for giving masses to dark matter and a potentially extended dark sector and interacting with SM matter via a scalar portal. These two-mediator dark matter models are not well explored, and yield some interesting new and uncovered signatures. Specifically, for dark Higgs masses above 250 GeV, the decay to two Higgs bosons represents the second largest contribution, and one in which both high-momentum Higgs decays and the ETmiss of the dark matter can be easily and separately reconstructed.This signature of a resonant di-Higgs production with significant ETmiss in the event through the mediator decay to dark matter has no dedicated experimental analyses. The main initial focus of the studentship is to analyse this final state, and developing a first ATLAS analysis for these resonant di-Higgs plus ETmiss signatures, building on the existing DM searches for mono-Higgs and supersymmetric Higgsinos in which two Higgs bosons are produced independently in an event. Due to the complicated final state and multiple kinematic handles on the signal, exploitation of machine learning techniques, such as classification and mass regression, will be investigated to provide additional sensitivity.The interpretation of this result will be in common with other dark Higgs final states, such as bb at low mass, and WW at high mass, allowing for a combination or summary to give the full sensitivity of ATLAS to these models. This signature is important, a general feature of many models with extended Higgs sectors mediating interactions with dark matter. Alternatively, he will have the opportunity to follow this analysis work with contributing to flagship dark matter searches using Run 3 data in the latter half of his PhD, as the high luminosities and sensitivities from Run 3 in 2023 and 2024 become available for experimental analysis.Maximising our sensitivity to final states containing Higgs decays to b-quark pairs is an interesting problem at the LHC. At high Higgs or boson momentum, the two b-quarks are close together, and can be captured in one large jet. The identification of high momentum b-quark jets is important, as is use of the full information present in these 'Higgs jets' - the presence of two sub-jets each identified as likely to contain a b-hadron, the overall structure of the jet and its kinematics. Substantial progress has been made in ATLAS with the development of an 'X->bb' tagger and ongoing work using advanced machine learning algorithms to provide a means of identifying and classifying these jets, finding jets from Higgs boson decays efficiently and with much improved rejection of top and light-quark QCD backgrounds. Tom will have the opportunity to be amongst the first analyses using these algorithms and to understand and contribute to the machine learning algorithms driving those.He will spend his second year at CERN to be as close as possible to the laboratory, his analysis team and Run-3 data-taking, and to gain visibility within the ATLAS collaboration.