{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T05:11:09Z","timestamp":1773724269758,"version":"3.50.1"},"reference-count":29,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2025,9,15]],"date-time":"2025-09-15T00:00:00Z","timestamp":1757894400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2025,9,15]],"date-time":"2025-09-15T00:00:00Z","timestamp":1757894400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/501100002347","name":"Federal Ministry of Education and Research","doi-asserted-by":"crossref","award":["05A23PAA"],"award-info":[{"award-number":["05A23PAA"]}],"id":[{"id":"10.13039\/501100002347","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/501100007210","name":"RWTH Aachen University","doi-asserted-by":"crossref","id":[{"id":"10.13039\/501100007210","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Comput Softw Big Sci"],"published-print":{"date-parts":[[2025,12]]},"abstract":"<jats:title>Abstract<\/jats:title>\n          <jats:p>Transformer networks excel in scientific applications. We explore two scenarios in ultra-high-energy cosmic ray simulations to examine what these network architectures learn. First, we investigate the trained positional encodings in air showers which are azimuthally symmetric. Second, we visualize the attention values assigned to cosmic particles originating from a galaxy catalog. In both cases, the Transformers learn plausible, physically meaningful features.<\/jats:p>","DOI":"10.1007\/s41781-025-00145-4","type":"journal-article","created":{"date-parts":[[2025,9,15]],"date-time":"2025-09-15T11:32:30Z","timestamp":1757935950000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["What Exactly Did the Transformer Learn from Our Physics Data?"],"prefix":"10.1007","volume":"9","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1653-1303","authenticated-orcid":false,"given":"Martin","family":"Erdmann","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7443-3042","authenticated-orcid":false,"given":"Niklas","family":"Langner","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5822-2445","authenticated-orcid":false,"given":"Josina","family":"Schulte","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Dominik","family":"Wirtz","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"297","published-online":{"date-parts":[[2025,9,15]]},"reference":[{"key":"145_CR1","unstructured":"Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser L, Polosukhin I (2017) Attention Is All You Need arXiv:1706.03762 [cs.CL]"},{"key":"145_CR2","unstructured":"Dosovitskiy A et al (2020) An image is worth 16x16 words: transformers for image recognition at scale arXiv:2010.11929 [cs.CV]"},{"key":"145_CR3","unstructured":"OpenAI (2022) ChatGPT: optimizing language models for dialogue. https:\/\/openai.com\/blog\/chatgpt\/. Online; Accessed 30. November 2022"},{"key":"145_CR4","unstructured":"Qu H, Li C, Qian S (2022) Particle transformer for jet tagging arXiv:2202.03772 [hep-ph]"},{"key":"145_CR5","doi-asserted-by":"publisher","first-page":"184","DOI":"10.1007\/JHEP06(2023)184","volume":"06","author":"T Finke","year":"2023","unstructured":"Finke T, Kr\u00e4mer M, M\u00fcck A, T\u00f6nshoff J (2023) Learning the language of QCD jets with transformers. JHEP 06:184. https:\/\/doi.org\/10.1007\/JHEP06(2023)184. arXiv:2303.07364 [hep-ph]","journal-title":"JHEP"},{"key":"145_CR6","doi-asserted-by":"publisher","DOI":"10.22323\/1.444.0371","author":"A Abdul Halim","year":"2023","unstructured":"Abdul Halim A et al (2023) Deep-learning-based cosmic-ray mass reconstruction using the Water-Cherenkov and scintillation detectors of AugerPrime. PoS ICRC2023. https:\/\/doi.org\/10.22323\/1.444.0371","journal-title":"PoS ICRC2023"},{"issue":"1","key":"145_CR7","doi-asserted-by":"publisher","DOI":"10.1088\/1674-1137\/ad7f3d","volume":"49","author":"Y Wu","year":"2025","unstructured":"Wu Y, Wang K, Li C, Qu H, Zhu J (2025) Jet tagging with more-interaction particle transformer. Chin Phys C 49(1):013110. https:\/\/doi.org\/10.1088\/1674-1137\/ad7f3d. arXiv:2407.08682 [hep-ph]","journal-title":"Chin Phys C"},{"key":"145_CR8","unstructured":"Brehmer J, Bres\u00f3 V, Haan P, Plehn T, Qu H, Spinner J, Thaler J (2024) A Lorentz-Equivariant Transformer for All of the LHC arXiv:2411.00446 [hep-ph]"},{"key":"145_CR9","unstructured":"Feickert M, Nachman B (2021) A living review of machine learning for particle physics arXiv:2102.02770 [hep-ph], https:\/\/github.com\/iml-wg\/HEPML-LivingReview"},{"key":"145_CR10","unstructured":"Wang A, Gandrakota A, Ngadiuba J, Sahu V, Bhatnagar P, Khoda EE, Duarte J (2024) Interpreting transformers for jet tagging arXiv:2412.03673 [hep-ph]"},{"key":"145_CR11","doi-asserted-by":"publisher","first-page":"172","DOI":"10.1016\/j.nima.2015.06.058","volume":"798","author":"A Aab","year":"2015","unstructured":"Aab A et al (2015) The Pierre auger cosmic ray observatory. Nucl Instrum Meth A 798:172\u2013213. https:\/\/doi.org\/10.1016\/j.nima.2015.06.058. arXiv:1502.01323 [astro-ph.IM]","journal-title":"Nucl Instrum Meth A"},{"issue":"07","key":"145_CR12","doi-asserted-by":"publisher","first-page":"07019","DOI":"10.1088\/1748-0221\/16\/07\/P07019","volume":"16","author":"A Aab","year":"2021","unstructured":"Aab A et al (2021) Deep-learning based reconstruction of the shower maximum $$X_{max}$$ using the water-Cherenkov detectors of the Pierre Auger Observatory. JINST 16(07):07019. https:\/\/doi.org\/10.1088\/1748-0221\/16\/07\/P07019. arXiv:2101.02946 [astro-ph.IM]","journal-title":"JINST"},{"issue":"07","key":"145_CR13","doi-asserted-by":"publisher","first-page":"07016","DOI":"10.1088\/1748-0221\/16\/07\/P07016","volume":"16","author":"A Aab","year":"2021","unstructured":"Aab A et al (2021) Extraction of the muon signals recorded with the surface detector of the Pierre Auger Observatory using recurrent neural networks. JINST 16(07):07016. https:\/\/doi.org\/10.1088\/1748-0221\/16\/07\/P07016. arXiv:2103.11983 [hep-ex]","journal-title":"JINST"},{"issue":"2","key":"145_CR14","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevD.111.022003","volume":"111","author":"A Abdul Halim","year":"2025","unstructured":"Abdul Halim A et al (2025) Measurement of the depth of maximum of air-shower profiles with energies between $$10^{18.5}$$ and $$10^{20}$$ eV using the surface detector of the Pierre Auger Observatory and deep learning. Phys Rev D 111(2):022003. https:\/\/doi.org\/10.1103\/PhysRevD.111.022003. arXiv:2406.06319 [astro-ph.HE]","journal-title":"Phys Rev D"},{"key":"145_CR15","unstructured":"Hoogeboom E, Peters JWT, Cohen TS, Welling M (2018) HexaConv arXiv:1803.02108 [cs.LG]"},{"key":"145_CR16","doi-asserted-by":"publisher","first-page":"192","DOI":"10.1088\/0004-637X\/738\/2\/192","volume":"738","author":"MS Pshirkov","year":"2011","unstructured":"Pshirkov MS, Tinyakov PG, Kronberg PP, Newton-McGee KJ (2011) Deriving global structure of the Galactic Magnetic Field from Faraday Rotation Measures of extragalactic sources. Astrophys J 738:192. https:\/\/doi.org\/10.1088\/0004-637X\/738\/2\/192. arXiv:1103.0814 [astro-ph.GA]","journal-title":"Astrophys J"},{"key":"145_CR17","doi-asserted-by":"publisher","first-page":"11","DOI":"10.1088\/2041-8205\/761\/1\/L11","volume":"761","author":"R Jansson","year":"2012","unstructured":"Jansson R, Farrar GR (2012) The galactic magnetic field. Astrophys J Lett 761:11. https:\/\/doi.org\/10.1088\/2041-8205\/761\/1\/L11. arXiv:1210.7820 [astro-ph.GA]","journal-title":"Astrophys J Lett"},{"issue":"1","key":"145_CR18","doi-asserted-by":"publisher","first-page":"95","DOI":"10.3847\/1538-4357\/ad4a54","volume":"970","author":"M Unger","year":"2024","unstructured":"Unger M, Farrar GR (2024) The coherent magnetic field of the milky way. Astrophys J 970(1):95. https:\/\/doi.org\/10.3847\/1538-4357\/ad4a54. arXiv:2311.12120 [astro-ph.GA]","journal-title":"Astrophys J"},{"key":"145_CR19","doi-asserted-by":"crossref","unstructured":"Unger M, Farrar GR (2017) Uncertainties in the magnetic field of the milky way arXiv:1707.02339 [astro-ph.GA]","DOI":"10.22323\/1.301.0558"},{"key":"145_CR20","doi-asserted-by":"publisher","unstructured":"Schulte J, Bister T, Erdmann M (2023) An all-sky search method for coherent magnetic field deflections of ultra-high-energy cosmic rays based on Deep Learning. PoS ICRC2023, 198 https:\/\/doi.org\/10.22323\/1.444.0198","DOI":"10.22323\/1.444.0198"},{"key":"145_CR21","unstructured":"Ardizzone L, L\u00fcth C., Kruse J, Rother C, K\u00f6the U (2019) Guided image generation with conditional invertible neural networks arXiv:1907.02392 [cs.CV]"},{"key":"145_CR22","unstructured":"Kingma DP, Dhariwal P (2018) Glow: generative flow with invertible 1x1 convolutions arXiv:1807.03039 [stat.ML]"},{"issue":"2","key":"145_CR23","doi-asserted-by":"publisher","first-page":"171","DOI":"10.1140\/epjc\/s10052-022-10138-x","volume":"82","author":"T Bister","year":"2022","unstructured":"Bister T, Erdmann M, K\u00f6the U, Schulte J (2022) Inference of cosmic-ray source properties by conditional invertible neural networks. Eur Phys J C 82(2):171. https:\/\/doi.org\/10.1140\/epjc\/s10052-022-10138-x. arXiv:2110.09493 [astro-ph.IM]","journal-title":"Eur Phys J C"},{"key":"145_CR24","doi-asserted-by":"publisher","first-page":"035","DOI":"10.1088\/1475-7516\/2022\/09\/035","volume":"09","author":"R Alves Batista","year":"2022","unstructured":"Alves Batista R et al (2022) CRPropa 3.2\u2014an advanced framework for high-energy particle propagation in extragalactic and galactic spaces. JCAP 09:035. https:\/\/doi.org\/10.1088\/1475-7516\/2022\/09\/035. arXiv:2208.00107 [astro-ph.HE]","journal-title":"JCAP"},{"key":"145_CR25","doi-asserted-by":"publisher","first-page":"022","DOI":"10.1088\/1475-7516\/2024\/01\/022","volume":"01","author":"AA Halim","year":"2024","unstructured":"Halim AA et al (2024) Constraining models for the origin of ultra-high-energy cosmic rays with a novel combined analysis of arrival directions, spectrum, and composition data measured at the Pierre Auger Observatory. JCAP 01:022. https:\/\/doi.org\/10.1088\/1475-7516\/2024\/01\/022. arXiv:2305.16693 [astro-ph.HE]","journal-title":"JCAP"},{"key":"145_CR26","doi-asserted-by":"crossref","unstructured":"Xiong Y, Zeng Z, Chakraborty R, Tan M, Fung G, Li Y, Singh V (2021) Nystr\u00f6mformer: A nystr\u00f6m-based algorithm for approximating self-attention arXiv:2102.03902 [cs.CL]","DOI":"10.1609\/aaai.v35i16.17664"},{"key":"145_CR27","doi-asserted-by":"publisher","first-page":"759","DOI":"10.1086\/427976","volume":"622","author":"KM G\u00f3rski","year":"2005","unstructured":"G\u00f3rski KM, Hivon E, Banday AJ, Wandelt BD, Hansen FK, Reinecke M, Bartelman M (2005) HEALPix\u2014a framework for high resolution discretization, and fast analysis of data distributed on the sphere. Astrophys J 622:759\u2013771. https:\/\/doi.org\/10.1086\/427976. arXiv:astro-ph\/0409513","journal-title":"Astrophys J"},{"issue":"35","key":"145_CR28","doi-asserted-by":"publisher","first-page":"1298","DOI":"10.21105\/joss.01298","volume":"4","author":"A Zonca","year":"2019","unstructured":"Zonca A, Singer L, Lenz D, Reinecke M, Rosset C, Hivon E, Gorski K (2019) healpy: equal area pixelization and spherical harmonics transforms for data on the sphere in Python. J Open Source Softw 4(35):1298. https:\/\/doi.org\/10.21105\/joss.01298","journal-title":"J Open Source Softw"},{"key":"145_CR29","unstructured":"Sundararajan M, Taly A, Yan Q (2017) Axiomatic attribution for deep networks arXiv:1703.01365 [cs.LG]"}],"container-title":["Computing and Software for Big Science"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s41781-025-00145-4.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s41781-025-00145-4\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s41781-025-00145-4.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,16]],"date-time":"2025-09-16T00:02:45Z","timestamp":1757980965000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s41781-025-00145-4"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,15]]},"references-count":29,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2025,12]]}},"alternative-id":["145"],"URL":"https:\/\/doi.org\/10.1007\/s41781-025-00145-4","relation":{},"ISSN":["2510-2036","2510-2044"],"issn-type":[{"value":"2510-2036","type":"print"},{"value":"2510-2044","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,9,15]]},"assertion":[{"value":"27 May 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"29 August 2025","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"15 September 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare no competing interests.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"16"}}