Publication: The AGORA High-resolution Galaxy Simulations Comparison Project. III: Cosmological Zoom-in Simulation of a Milky Way-mass halo
Full text at PDC
Advisors (or tutors)
We present a suite of high-resolution cosmological zoom-in simulations to z = 4 of a 10^(12) Mꙩ halo at z = 0, obtained using seven contemporary astrophysical simulation codes (ART-I, ENZO, RAMSES, CHANGA, GADGET3, GEAR, and GIZMO) widely used in the numerical galaxy formation community. Physics prescriptions for gas cooling, heating and star formation are similar to the ones used in our previous AGORA disk comparison (Kim et al. 2016) but now account for the effects of cosmological processes such as the expansion of the Universe, intergalactic gas inflow, and the cosmic ultraviolet background radiation emitted by massive stars and quasars. In this work, we introduce the most careful comparison yet of galaxy formation simulations run by different code groups, together with a series of four calibration steps each of which is designed to reduce the number of tunable simulation parameters adopted in the final run. In the first two steps, we methodically calibrate the gas physics such as cooling and heating, in simulations without star formation. In the third, we seek an agreement on the total stellar mass produced with the common star formation prescription used in the AGORA disk comparison, in stellar feedback-free simulations. In the last calibration step, we activate stellar feedback, where each code group is asked to set the feedback prescriptions to be as close to the most used one in each code community as possible, while aiming for convergence in the stellar mass at z = 4 to the values predicted by semi-empirical models. After all the participating code groups successfully completed the calibration steps, we reach a suite of cosmological simulations with similar mass assembly histories down to z = 4. With numerical accuracy that resolves the internal structure of a target halo (˂̰ 100 physical pc at z = 4), we find that the codes overall agree well with one another in e.g., gas and stellar properties, but also show differences in e.g., circumgalactic medium (CGM) properties. We argue that, if adequately tested in accordance with our proposed calibration steps and common parameters, the results of high-resolution cosmological zoom-in simulations can be robust and reproducible. New code groups are invited to join and enrich this comparison by generating equivalent models or to test the code’s compatibility on their own, by adopting the common initial conditions, the common easy-to-implement physics package, and the proposed calibration steps. Further analyses of the zoom-in simulations presented here will be in forthcoming reports from the AGORA Collaboration, including studies of the CGM, simulations by additional codes, and results at lower redshift.
© 2021. The American Astronomical Society. Artículo firmado por 24 autores. We thank all of our colleagues participating in the AGORA Project for their collaborative spirit which has allowed the AGORA Collaboration to remain strong as a platform to foster and launch multiple science-oriented comparison efforts. We thank Aldo Rodr´ıguez-Puebla for sharing results from the abundance matching semi-empirical models, and Volker Springel for providing the original versions of GADGET3 to be used in the AGORA Project. We also thank the anonymous referee for his/her insightful comments and suggestions. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Santi Roca-Fabrega acknowledges ` support from a Spanish postdoctoral fellowship, under grant number 2017-T2/TIC-5592. His work has been supported by the Madrid Government (Comunidad de Madrid-Spain) under the Multiannual Agreement with Complutense University in the line Program to Stimulate Research for Young Doctors in the context of the V PRICIT (Regional Programme of Research and Technological Innovation). He also acknowledges financial support from the Spanish Ministry of Economy and Competitiveness (MINECO) under grant number AYA2016-75808-R, AYA2017-90589-REDT and S2018/NMT-429, and from the CAM-UCM under grant number PR65/19-22462. Ji-hoon Kim acknowledges support by Samsung Science and Technology Foundation under Project Number SSTF-BA1802-04. His work was also supported by the National Institute of Supercomputing and Network/Korea Institute of Science and Technology Information with supercomputing resources including technical support, grants KSC-2018-CRE-0052 and KSC-2019-CRE0163. Kentaro Nagamine acknowledges the support by the MEXT/JSPS KAKENHI Grant Number JP17H01111, 19H05810 & 20H00180, as well as the travel support from the Kavli IPMU, World Premier Research Center Initiative (WPI), where part of this work was conducted. Alessandro Lupi acknowledges funding by the MIUR under the grant PRIN 2017-MB8AEZ. Daniel Ceverino is a Ramon-Cajal Researcher and is supported by the Ministerio de Ciencia, Innovacion y Universidades (MICIU/FEDER) under research ´ grant PGC2018-094975-C21. Hector Vel ´ azquez acknowl- ´ edges support from PAPIIT-UNAM under grant number IN101918 and also by the Centro Nacional de Supercomputo (CNS-IPICYT-CONACYT). ART-I simulations were performed on the BRIGIT/EOLO cluster at the Centro de Proceso de Datos, Universidad Complutense de Madrid, and on the ATOCATL ´ supercomputer at the LAMOD/IAUNAM. LAMOD is a collaborative effort between the IA, ICN and IQ institutes at UNAM. RAMSES simulations were performed on the MIZTLI supercomputer at the LANACAD, Universidad Nacional Autonoma de M ´ exico, within the re- ´ search project LANCAD-UNAM-DGTIC-151 and on the Laboratorio Nacional de Supercmputo del Sureste-Conacyt. ´ CHANGA simulations were performed on the ATOCATL ´ supercomputer at the Instituto de Astronom´ıa de la UNAM, and on the Extreme Science and Engineering Discovery Environment (XSEDE) allocations TG-AST20020 and TGMCA94P018. XSEDE is supported by the National Science Foundation (NSF) grant ACI-1053575. GADGET3-OSAKA simulations and analyses were performed on the XC50 systems at the Center for Computational Astrophysics (CfCA) of the National Astronomical Observatory of Japan (NAOJ), OCTOPUS at the Cybermedia Center, Osaka University, and Oakforest-PACS at the University of Tokyo as part of the HPCI system Research Project (hp190050, hp200041). The publicly available ENZO and yt codes used in this work are the products of collaborative efforts by many independent scientists from numerous institutions around the world. Their commitment to open science has helped make this work possible.