Validating the CROCODILE model within the AGORA galaxy simulation framework

TL;DR

This study systematically validates gadget4-osaka, a GADGET-4–based galaxy formation code, within the AGORA framework by direct comparison to gadget3-osaka across isolated and cosmological contexts. Through a staged calibration (Cal-1 to Cal-4) and a controlled set of feedback variations, the authors demonstrate that gravity and hydrodynamics differences introduce modest offsets, but the dominant driver of galaxy evolution is the stellar feedback model. Specifically, momentum-driven mechanical feedback combined with stochastic thermal heating yields a realistic, multiphase ISM, suppresses clump formation, launches CGM-enriching outflows, and aligns the Kennicutt–Schmidt relation with observations. The work provides a benchmark data point in the CosmoRun suite and highlights how numerical methods and sub-grid physics jointly shape galaxy evolution, with implications for code development and interpretation of simulation-based predictions.

Abstract

Numerical galaxy formation simulations are sensitive to numerical methods and sub-grid physics models, making code comparison projects essential for quantifying uncertainties. Here, we evaluate GADGET4-OSAKA within the AGORA project framework by conducting a systematic comparison with its predecessor. We perform an isolated disk galaxy and a cosmological zoom-in run of a Milky Way-mass halo, following the multi-step AGORA calibration procedure. By systematically deconstructing the updated stellar feedback model, we demonstrate that mechanical momentum injection is necessary to suppress unphysical gas fragmentation and regulate star formation, yielding agreement with the Kennicutt-Schmidt relation. Meanwhile, stochastic thermal heating is essential for driving a hot, metal-enriched gaseous halo, thereby creating a multiphase circumgalactic medium that is absent in the predecessor code. In the cosmological context, we calibrate the simulation to match the stellar mass growth history targeted by the AGORA collaboration. The validated GADGET4-OSAKA simulation has been contributed to the AGORA CosmoRun suite, providing a new data point for understanding the impact of numerical and physical modeling choices on galaxy evolution.

Figures (18)

• Figure S1: AGORA cosmological calibration procedure. Physical complexity increases from top to bottom: Cal-1 (gravity and adiabatic hydrodynamics), Cal-2 (radiative cooling), Cal-3 (star formation) and Cal-4 (stellar feedback). Each step has a specified target redshift ($z=7$ for Cal-1/2/3, $z=4$ for Cal-4) for inter-code comparison and a diagnostic focus (e.g., gas and temperature projections in Cal-1/2, stellar mass in Cal-3/4). This staged approach ensures differences in final calibrated runs can be confidently attributed to feedback implementations rather than accumulated numerical artifacts. Adapted and modified from Figure 1 of Roca_F_brega_2021.
• Figure S2: Dark matter density projections at $z=7$ for (left) gadget3-osaka and (right) gadget4-osaka. Particles are deposited via a cloud-in-cell (CIC) scheme on a 300 pc grid, and surface densities are then calculated for a slice of $150$ kpc thickness. Both codes produce visually indistinguishable dark matter distributions, demonstrating convergence of gravity solvers (TreePM vs. FMM-PM).
• Figure S3: Gas density projection (top) and density-weighted temperature projection (bottom), each projected through a slab of thickness $150$ kpc at $z=7$, for Cal-1 (adiabatic, cols 1-2) and Cal-2 (with cooling, cols 3-4). While both gadget3-osaka and gadget4-osaka converge in Cal-1, minor variations in the temperature distribution are present in Cal-2.
• Figure S4: (a) Gas density projection (top) and density-weighted temperature projection (bottom) (both face-on), each projected through a slab of $35$ kpc thickness at $500$ Myr, for NSFF runs. The left panel shows gadget3-osaka while the right panel shows gadget4-osaka. There is overall agreement between the two, with minor differences in the gas clumpiness and the distribution of cold gas. (b) Mass-weighted phase diagrams of gas density vs. temperature for the gas within $100$ kpc from the center of the galaxy in the NSFF (left), Cal-1 (middle), and Cal-2 (right) runs, at $500$ Myr for the first and $z = 7$ for the latter runs. To guide the eye, we use a thick dashed line in the NSFF panel to plot the mean temperature in each density bin for gadget3-osaka. Colors represent the total gas mass in each 2-dimensional bin.
• Figure S5: Spherically averaged density profiles for dark matter (solid line) and gas (dashed line, when applicable) at $z=7$, shown for DM-Only (col 1), Cal-1 (col 2), Cal-2 (col 3), and NSFF isolated runs at $t=500$ Myr (col 4). The upper subplot in each panel displays the logarithmic density ratio $\log_{10}(\rho_{\rm G3}/\rho_{\rm G4})$ to highlight deviations.