在大型強子對撞機 (LHC) 實驗中，經由質子對撞所產生的頂夸克對具有非常複 雜的過程以及產物，至今仍無法被非常正確的判別以及重建。在本研究中，我 們提出了一個利用新穎的機器學習方法來對雙頂夸克全強子衰變過程進行重建。 此方法基於 Attention mechanism，我們稱之為 Symmetry Preserving Attention Networks(SPA-Net)。這個模型架構可以在避免組合性爆炸的前提下對所有的衰 變產物進行辨識以及重建。此方法對比於傳統的 χ2 重建方式，表現出了非常巨 大的差異。本方法可以在一、存在 6 jets 條件下正確的重建 93% 的事件;二、 存在 7 jets 條件下正確的重建 87% 的事件;三、存在大於 8 jets 條件下正確的 重建 82.6% 的事件。此架構的出色表現指明其具有更大的潛能能以更有效的手 段進行事件重建。對於本架構而言，只要是一個在最終產物具有置換對稱性的 物理過程，就可以是一個適合且有潛力的套用對象。

The top quarks produced by proton-proton (pp) collisions in the Large Hadron Collider (LHC) undergo a very complicated process of formation. To date, the decay process of top quarks has not yet been well-classified, because of its complicated topology and large background events. In this project, we present a novel approach to the “all hadronic decay” process of top quarks based on the neural networks with attention mechanism, we refer to as the “Symmetry Preserving Attention Networks” (SPA-Net). These networks identify the decay products of each quark unambiguously and without combinatorial explosion (i.e. the rapid growth of the complexity). This approach performs outstandingly compared to the generally accepted methods. Our networks can correctly assign all hadronic decay in 93.0% of 6 jets, 87.8% of 7 jets, and 82.6% of ≥ 8 jets event respectively. The outstanding performance of this structure points the way to a more efficient solution to parton-jet assignment problem. A physical process that contains a permutation symmetry in final states is a well-suited item to study with SPA-NET.

[1] ATLAS Collaboration, “Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHC,” Phys. Lett. B 716, 1-29 (2012) [arXiv:1207.7214 [hep-ex]]. https://arxiv.org/abs/ 1207.7214.
[2] CMS Collaboration, “Observation of a New Boson at a Mass of 125 GeV with the CMS Experiment at the LHC,” Phys. Lett. B 716, 30-61 (2012) [arXiv:1207.7235 [hep-ex]]. https://arxiv.org/abs/1207.7235.
[3] A. Vaswani et al., “Attention is all you need,” Advances in Neural Infor- mation Processing Systems, NIPS (2017) [arXiv:1706.03762[sc.CL]]. https:// arxiv.org/abs/1706.03762.
[4] P. A. Zyla et al. [Particle Data Group], “Review of Particle Physics,” PTEP 2020, no.8, 083C01 (2020) doi:10.1093/ptep/ptaa104
[5] A. Quade, “Top quark physics at hadron colliders,” doi:10.1007/978-3-540- 71060-8. https://www.springer.com/gp/book/9783540710592.
[6] CMS Collaboration, “Measurement of the top quark mass in the all-jets fi- nal state at √s = 13 TeV and combination with the lepton+jets channel,” Eur. Phys. J. C 79, no.4, 313 (2019) [arXiv:1812.10534 [hep-ex]]. https:// arxiv.org/abs/1812.10534.
[7] ATLAS Collaboration, “Top-quark mass measurement in the all-hadronic tt decay channel at √s = 8 TeV with the ATLAS detector,” JHEP 09, 118 (2017) [arXiv:1702.07546 [hep-ex]]. https://arxiv.org/abs/1702.07546.
26
[8] T. Mccarthy, “Measurement of the Top Quark Mass in the All-Hadronic Top- Antitop Decay Channel Using Proton-Proton Collision Data from the ATLAS Experiment at a Centre-of-Mass Energy of 8 TeV,” CERN-THESIS-2015-275. https://cds.cern.ch/record/2125782.
[9] ATLAS Collaboration, “Machine Learning Algorithms for b-Jet Tagging at the ATLAS Experiment,” J. Phys. Conf. Ser. 1085, no.4, 042031 (2018) [arXiv:1711.08811 [hep-ex]]. https://arxiv.org/abs/1711.08811.
[10] Hung-Yi, Lee, “NTU course lecture note-Transformer” https:// speech.ee.ntu.edu.tw/ hylee/ml/ml2021-course-data/self_v7.pdf
[11] J. de Favereau et al. [DELPHES 3], “DELPHES 3, A modular framework for fast simulation of a generic collider experiment,” JHEP 02, 057 (2014) [arXiv:1307.6346 [hep-ex]]. https://arxiv.org/abs/1307.6346.
[12] ATLAS Collaboration, “Optimisation of the ATLAS b-tagging performance for the 2016 LHC Run,” ATL-PHYS-PUB-2016-012. https://cds.cern.ch/ record/2160731
[13] CMS Collaboration, “Identification of heavy-flavour jets with the CMS detector in pp collisions at 13 TeV,” JINST 13, no.05, P05011 (2018) [arXiv:1712.07158 [physics.ins-det]]. https://arxiv.org/abs/1712.07158
[14] M. J. Fenton, A. Shmakov, T. W. Ho, S. C. Hsu, D. Whiteson and P. Baldi, “Permutationless Many-Jet Event Reconstruction with Symmetry Preserv- ing Attention Networks,” [arXiv:2010.09206 [hep-ex]]. https://arxiv.org/abs/ 2010.09206.
[15] A. Shmakov, M. J. Fenton, T. W. Ho, S. C. Hsu, D. Whiteson and P. Baldi, “SPANet: Generalized Permutationless Set Assignment for Particle Physics using Symmetry Preserving Attention,” [arXiv:2106.03898 [hep-ex]]. https:// arxiv.org/abs/2106.03898.
[16] I. Low, N. R. Shah and X. P. Wang, “Higgs Alignment and Novel CP- Violating Observables in 2HDM,” [arXiv:2012.00773 [hep-ph]]. https:// arxiv.org/pdf/2012.00773.pdf.
[17] CMS Collaboration, “Fully hadronic ttH(bb) search,” CERN-OPEN-2019- 005. https://cds.cern.ch/record/2689088
[18] CMS Collaboration, “Search for standard model production of four top quarks with same-sign and multilepton final states in proton–proton collisions at √s = 13 TeV,” Eur. Phys. J. C 78, no.2, 140 (2018) [arXiv:1710.10614 [hep- ex]]. https://arxiv.org/pdf/1710.10614.pdf.