用機器學習區分部分可視噴流

Classification of Semi-visible Jets with Machine Learning

鄭博仁(Po-Jen Cheng)

指導教授 ： 張敬民 ； 徐士傑

國立清華大學/理學院/物理學系/碩士(2020年)

摘要

摘要部分可視噴流是一種我們尚未觀察到的暗物質模型預測,它在模型中是由可視的強子與不可視的穩定暗介子組成,因為它有一部分是可以被偵測到的,所以我們就能在大型強子對撞機裡面搜尋這個訊號.但我們很難去區分部分可視噴流與量子色動力學的噴流,所以我們應用機器學習去做這些種類的區分,我們在研究中應用深度神經網路、卷積神經網路與ParticleNet來區分它們,而我們所輸入的參數是噴流的結構變數、噴流的影像還有噴流組成粒子的參數.最後我們比較這些模型的區分能力.

關鍵字

機器學習 ； 部分可視噴流 ； 對撞機 ； 噴流辨識

並列摘要

Semi-visible jets (SVJ) are predictions of dark sectors which are not observed yet, and they are composed of visible hadrons and stable dark hadrons. Since they have a visible part, we can search SVJ in the Large Hadron Collider. We apply Machine Learning (ML) techniques to classify SVJ from Quantum Chromodynamics jets because it’s extremely challenge to distinguish them. In this thesis, we deployed three different deep learning models using combinations of low-level features and high-level features and compared their performances. Models include Deep Neural Network, Convolution Neural Network, and ParticleNet. High-level features are jet-substructure variables, and low-level features are jet images or variables of jet’s constituents. We find that ParticleNet with constituent variables as inputsprovides the best classification power.

並列關鍵字

semi-visible ； jets ； Machine ； Learning ； Classificatio ； Hidden Valley

參考文獻

[1] Lisa Carloni and Torbj ̈orn Sj ̈ostrand. “Visible effects of invisible HiddenValley radiation”. In:Journal of High Energy Physics2010.9 (Sept. 2010).issn: 1029-8479.doi:10.1007/jhep09(2010)105.url:http://dx.doi.org/10.1007/JHEP09(2010)105.[2] Timothy Cohen, Mariangela Lisanti, and Hou Keong Lou. “Semivisible Jets:Dark Matter Undercover at the LHC”. In:Physical Review Letters115.17(Oct. 2015).issn: 1079-7114.doi:10.1103/physrevlett.115.171804.url:http://dx.doi.org/10.1103/PhysRevLett.115.171804.[3] Timothy Cohen et al. “LHC searches for dark sector showers”. In:Journal ofHigh Energy Physics2017.11 (Nov. 2017).issn: 1029-8479.doi:10.1007/jhep11(2017)196.url:http://dx.doi.org/10.1007/JHEP11(2017)196.[4] S. Chatrchyan et al. “Observation of a new boson at a mass of 125 GeV withthe CMS experiment at the LHC”. In:Physics Letters B716.1 (Sept. 2012),pp. 30–61.issn: 0370-2693.doi:10.1016/j.physletb.2012.08.021.url:http://dx.doi.org/10.1016/j.physletb.2012.08.021.[5] G. Aad et al. “Observation of a new particle in the search for the StandardModel Higgs boson with the ATLAS detector at the LHC”. In:PhysicsLetters B716.1 (Sept. 2012), pp. 1–29.issn: 0370-2693.doi:10.1016/j.physletb.2012.08.020.url:http://dx.doi.org/10.1016/j.physletb.2012.08.020.[6] Dimitri Bourilkov. “Machine and deep learning applications in particle physics”.In:International Journal of Modern Physics A34.35 (Dec. 2019), p. 1930019.

issn: 1793-656X.doi:10.1142/s0217751x19300199.url:http://dx.doi.org/10.1142/S0217751X19300199.[7] Fabio Maltoni and Tim Stelzer. “MadEvent: automatic event generationwith MadGraph”. In:Journal of High Energy Physics2003.02 (Feb. 2003),pp. 027–027.issn: 1029-8479.doi:10.1088/1126-6708/2003/02/027.url:http://dx.doi.org/10.1088/1126-6708/2003/02/027.[8] Torbj ̈orn Sj ̈ostrand et al. “An introduction to PYTHIA 8.2”. In:ComputerPhysics Communications191 (June 2015), pp. 159–177.issn: 0010-4655.doi:10.1016/j.cpc.2015.01.024.url:http://dx.doi.org/10.1016/j.cpc.2015.01.024.[9] J. de Favereau et al. “DELPHES 3: a modular framework for fast simulationof a generic collider experiment”. In:Journal of High Energy Physics2014.2(Feb. 2014).issn: 1029-8479.doi:10.1007/jhep02(2014)057.url:http://dx.doi.org/10.1007/JHEP02(2014)057.[10] Matteo Cacciari, Gavin P Salam, and Gregory Soyez. “The anti-ktjet clus-tering algorithm”. In:Journal of High Energy Physics2008.04 (Apr. 2008),pp. 063–063.issn: 1029-8479.doi:10.1088/1126-6708/2008/04/063.url:http://dx.doi.org/10.1088/1126-6708/2008/04/063.[11] Matteo Cacciari, Gavin P. Salam, and Gregory Soyez. “FastJet user manual”.In:The European Physical Journal C72.3 (Mar. 2012).issn: 1434-6052.doi:10.1140/epjc/s10052-012-1896-2.url:http://dx.doi.org/10.1140/epjc/s10052-012-1896-2.[12] Andrew J. Larkoski, Gavin P. Salam, and Jesse Thaler. “Energy correlationfunctions for jet substructure”. In:Journal of High Energy Physics2013.6(June 2013).issn: 1029-8479.doi:10.1007/jhep06(2013)108.url:http://dx.doi.org/10.1007/JHEP06(2013)108.[13] Davide Napoletano and Gregory Soyez. “Computing N -subjettiness forboosted jets”. In:Journal of High Energy Physics2018.12 (Dec. 2018).issn:1029-8479.doi:10.1007/jhep12(2018)031.url:http://dx.doi.org/10.1007/JHEP12(2018)031.

Yue Wang et al. “Dynamic Graph CNN for Learning on Point Clouds”. In:ACM Transactions on Graphics (TOG)(2019).