Cosmological neutrino simulations at extreme scale

• Published in: Research in astronomy and astrophysics Vol. 17; no. 8; pp. 89 - 100
• Main Authors: Emberson, J. D., Yu, Hao-Ran, Inman, Derek, Zhang, Tong-Jie, Pen, Ue-Li, Harnois-Déraps, Joachim, Yuan, Shuo, Teng, Huan-Yu, Zhu, Hong-Ming, Chen, Xuelei, Xing, Zhi-Zhong
• Format: Journal Article
• Language: English
• Published: Beijing National Astronomical Observatories, CAS and IOP Publishing Ltd 01.08.2017; IOP Publishing
• Subjects: ASTRONOMY AND ASTROPHYSICS; Clustering; Cold dark matter; Computer simulation; cosmology; Cosmology: theory; Dark matter; Data compression; Large scale structure of the universe; large-scale; large-scale structure of universe; methods; methods: numerical; Neutrinos; numerical; Numerical simulations; Simulation; structure; theory; universe
• ISSN: 1674-4527; 2397-6209
• DOI: 10.1088/1674-4527/17/8/85

Constraining neutrino mass remains an elusive challenge in modern physics. Precision mea-surements are expected from several upcoming cosmological probes of large-scale structure. Achievingthis goal relies on an equal level of precision from theoretical predictions of neutrino clustering.Numerical simulations of the non-linear evolution of cold dark matter and neutrinos play a pivotal rolein this process. We incorporate neutrinos into the cosmological N-body code CUBEP^3M and discuss thechallenges associated with pushing to the extreme scales demanded by the neutrino problem. We high-light code optimizations made to exploit modern high performance computing architectures and presenta novel method of data compression that reduces the phase-space particle footprint from 24 bytes insingle precision to roughly 9 bytes. We scale the neutrino problem to the Tianhe-2 supercomputer andprovide details of our production run, named TianNu, which uses 86% of the machine （13 824 computenodes）. With a total of 2.97 trillion particles, TianNu is currently the world＇s largest cosmological N-body simulation and improves upon previous neutrino simulations by two orders of magnitude in scale.We finish with a discussion of the unanticipated computational challenges that were encountered duringthe TianNu runtime.