APOSTLE vs. AURIGA Simulations: How Subgrid Models Shape Milky Way Analogs

Jianhong Hu, Hang Yang, Liang Gao · 2026

Read APOSTLE vs. AURIGA Simulations: How Subgrid Models Shape Milky Way Analogs by Jianhong Hu, Hang Yang, Liang Gao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-21837.txt.

Introduction

Hu and co-authors compare two galaxy formation simulations that start from the same initial conditions but use different subgrid physics. The APOSTLE and AURIGA runs thus isolate how baryonic recipes shape Milky Way-like galaxies. AURIGA galaxies end up with higher stellar masses and surface densities, driven by faster gas cooling from the halo. Both runs form clear disks, but AURIGA disks form earlier and show stronger bars and spiral arms. Satellites differ too: AURIGA makes heavier ones, while counts above a million solar masses stay similar.

Catalog record

CREATORJianhong Hu, Hang Yang, Liang Gao
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41693
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.21837

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

APOSTLE vs. AURIGA Simulations: How Subgrid Models Shape Milky Way Analogs Jianhong Hu, Hang Yang, Liang Gao Abstract Despite significant progress in cosmological simulations of galaxy formation, the role of subgrid physics in shaping the detailed properties of galaxies remains incompletely understood. In this work, we analyze two sets of zoom-in simulations that share identical initial conditions but adopt distinct implementations of baryonic physics, enabling a controlled comparison of their predictions. We examine the stellar properties, morphological structures, and satellite populations of the simulated galaxies at z = 0 z=0 . We find that Auriga galaxies systematically exhibit higher stellar masses and surface densities than their Apostle counterparts. These differences are primarily driven by variations in the efficiency of gas cooling from the circumgalactic medium (CGM) into the star-forming gas. Both simulations form well-defined disk galaxies; however, Auriga systems generally display higher disk-to-total mass ratios, earlier disk formation, and more prominent dynamical structures such as bars and spiral arms. Nevertheless, strongly disk-dominated systems are present in both simulations, although they do not arise in the same host haloes. The vertical disk structure in both simulations is well described by a sech 2 \text{sech}^{2} density profile, with scale heights below ∼ 1 \sim 1 kpc in the inner regions. The satellite populations also differ, with Auriga producing systematically more massive satellites, including a ∼ 0.3 \sim 0.3 dex increase in the most massive system, while the number of satellites above 10 6 ​ M ⊙ 10^{6}M_{\odot} remains comparable in most halo pairs. Both simulations reproduce similar satellite stellar mass–metallicity relations, albeit ∼ 0.25 \sim 0.25 dex higher than observation. This comparative study therefore provides useful benchmarks for future efforts to better constrain galaxy formation models. I Introduction The standard \Lambda cold dark matter ( \Lambda CDM) paradigm has become the cornerstone of modern cosmology, providing a robust framework for understanding cosmic structure formation ( 31 ; 2 ; 8 , e.g.) . This model has demonstrated remarkable success in explaining diverse observations, particularly the statistical properties of large-scale structure ( 24 ; 18 ; 43 , e.g.) . Within this framework, galaxies form through the cooling and condensation of gas within hierarchically assembled dark matter haloes, while their luminous components are shaped by complex baryonic processes operating across a wide range of scales ( 50 ; 51 , e.g.) . The nonlinear coupling of processes like gas cooling, star formation, and feedback renders galaxy formation one of the most challenging problems in modern astrophysics, motivating the use of numerical simulations to study this complex system ( 41 ; 25 ; 6 , e.g.) . Although cosmological simulations are now essential tools for studying galaxy formation, the accurate modeling of baryonic physics remains one of the most formidable challenges, primarily due to its inherently multiscale nature spanning from galactic to interstellar scales. The importance of feedback processes has been established over decades of numerical work. Early simulations showed that, without valid feedback, gas undergoes catastrophic cooling, producing galaxies that are overly compact, with excessive central star formation and significant angular momentum loss ( 26 ; 27 ; 28 , e.g.) . Later studies demonstrated that suitably calibrated feedback—such as supernova-driven outflows can suppress runaway cooling, regulate star formation, and preserve extended gas reservoirs required for disk formation ( 49 ; 42 ; 46 ; 20 ; 22 , e.g.) . Comparative projects using identical initial conditions across different codes, such as Aquila ( 37 ) and AGORA ( 15 ; 34 ; 13 ) , have further shown that differences in subgrid physics—especially feedback prescriptions—are the dominant source of variation in hydrodynamical simulation results. Building upon these foundations, we conduct a systematic comparison of galaxy formation outcomes between the Auriga ( 11 ) and Apostle ( 36 ; 10 ) simulation suites to establish how variations in baryonic physics models translate into systematic differences in galaxy evolution and observable properties. The original versions of these simulations have been widely used to study galaxy formation across a range of environments, including detailed investigations of Milky Way-like systems ( 19 ; 9 ; 53 ; 33 ; 29 , e.g.) . However, a systematic comparison of these simulation suites under matched initial conditions has not yet been performed, limiting our ability to isolate the impact of different baryonic physics implementations. In this work, we analyze a set of new simulations that adopt identical initial conditions for Local Group-like volumes, while otherwise retaining the full subgrid physics implementations of the original Auriga and Apostle models, particularly for stellar feedback. Specifically, Apostle adopts a thermal energy injection scheme that heats gas particles to a fixed temperature ( 7 ) , whereas Auriga employs a kinetic wind model in which particles are temporarily decoupled and launched isotropically, depositing both thermal and kinetic energy into the interstellar medium upon recoupling ( 48 ) . Previous studies utilizing these controlled simulation datasets have already revealed significant differences in baryonic properties arising from these distinct implementations. 14 showed that the efficient stellar feedback in Apostle leads to substantial baryon deficiency due to strong gas ejection and suppressed accretion at high redshift. In addition, 54 examined the evolution of angular momentum, finding that the feedback in Apostle inhibits the recycled galactic scale fountain process, thereby limiting angular momentum acquisition from the circumgalactic medium. Building on these results, we present a systematic and unified analysis of galaxy properties across both simulations. In particular, we examine stellar population distributions, galaxy morphology, and satellite system properties to establish a coherent physical picture of how baryonic physics implementations shape the Milky Way analogs. This paper is organized as follows. Section II describes the simulation suites and the galaxy sample selection, following the methodology of 54 . Section III presents our comparative analysis, including stellar population properties (Section III.2 ), morphological characteristics (Section III.3 ), and satellite system properties (Section III.4 ). We summarize our findings and discuss their implications in Section IV . II Simulations and Sample Selection Our samples are selected from two suites of ‘zoom-in’ simulations with identical initial conditions but distinct hydrodynamical schemes. The first one is Apostle project ( 36 ; 10 ) , which was run using a modified version of Gadget-3 code ( 44 ) , with fluid properties calculated by the pressure–entropy formulation of smoothed particle hydrodynamics (SPH) technique ( 12 ) . The second one is Auriga -like simulations (with the same galaxy formation model as Auriga project ( 11 ) ), using moving-mesh magneto-hydrodynamics code Arepo ( 45 ) . All simulations are performed with the WMAP-7 cosmological parameters ( 17 ) : \Omega_{\Lambda}=0.728 , \Omega_{m}=0.272 , \Omega_{b}=0.0455 , h=0.704 , \sigma_{8}=0.81 , and n_{s}=0.967 . The simulations share identical mass resolutions, with initial gas (dark matter) particle/cell masses of about 1.2~(5.9)\times 10^{5}M_{\odot} and a common maximum gravitational softening length of 307 pc. Critically, two simulations incorporate markedly different subgrid models for key baryonic processes. Most notably, Apostle implements a thermal feedback scheme ( 7 ) where supernova energy is stochastically distributed among neighboring gas particles as pure thermal energy, heating them by a prescribed temperature increase. In contrast, Auriga utilizes a kinetic feedback model ( 48 ) that launches isotropic ‘wind particles’ carrying mass, metals,

Read the complete text