{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/cosmoflow-using-deep-learning-to-learn-the","title":"CosmoFlow: Using Deep Learning to Learn the Universe at Scale","arxiv_id":"1808.04728","date":"2018-08-14","proceeding":null,"authors":["Amrita Mathuriya","Deborah Bard","Peter Mendygral","Lawrence Meadows","James Arnemann","Lei Shao","Siyu He","Tuomas Karna","Daina Moise","Simon J. Pennycook","Kristyn Maschoff","Jason Sewall","Nalini Kumar","Shirley Ho","Mike Ringenburg","Prabhat","Victor Lee"],"abstract":"Deep learning is a promising tool to determine the physical model that\ndescribes our universe. To handle the considerable computational cost of this\nproblem, we present CosmoFlow: a highly scalable deep learning application\nbuilt on top of the TensorFlow framework. CosmoFlow uses efficient\nimplementations of 3D convolution and pooling primitives, together with\nimprovements in threading for many element-wise operations, to improve training\nperformance on Intel(C) Xeon Phi(TM) processors. We also utilize the Cray PE\nMachine Learning Plugin for efficient scaling to multiple nodes. We demonstrate\nfully synchronous data-parallel training on 8192 nodes of Cori with 77%\nparallel efficiency, achieving 3.5 Pflop/s sustained performance. To our\nknowledge, this is the first large-scale science application of the TensorFlow\nframework at supercomputer scale with fully-synchronous training. These\nenhancements enable us to process large 3D dark matter distribution and predict\nthe cosmological parameters $\\Omega_M$, $\\sigma_8$ and n$_s$ with unprecedented\naccuracy.","url_abs":"http://arxiv.org/abs/1808.04728v2","url_pdf":"http://arxiv.org/pdf/1808.04728v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"}],"methods":[{"method_slug":"3d-convolution","method_name":"3D Convolution"},{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[{"slug":"cosmoflow","name":"CosmoFlow","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}