Scott Lucchini Computational Astrophysicist
Public data release

ENGAWA

ENhanced Galactic Atmospheres With Arepo

A suite of cosmological zoom-in simulations that resolves the circumgalactic medium of Milky Way-like galaxies down to 200 pc — using fixed-volume mesh refinement in Arepo together with the IllustrisTNG feedback model. The full snapshot and catalog data are now public via Globus.

Lucchini, Abramson, Hummels, Conroy, Hernquist & Smith — submitted to ApJ (2026)

01 Overview

Resolving the atmospheres of galaxies

The circumgalactic medium (CGM) — the diffuse gas reservoir surrounding a galaxy — regulates how galaxies accrete fuel, recycle their winds, and ultimately form stars. But it is notoriously difficult to simulate: it fills an enormous volume at very low density, is intrinsically multiphase, and is stirred by chaotic environmental effects.

Standard cosmological simulations use mass-based refinement, which pours computational power into dense, star-forming regions and leaves the low-density CGM poorly resolved. ENGAWA takes the complementary path: a fixed-volume refinement criterion that caps the physical size of gas cells throughout the CGM, layered on top of the usual mass refinement. This is only practical with Arepo's hybrid, moving Voronoi mesh, where individual gas cells can be refined and de-refined on the fly.

The result is a consistent 200 pc spatial resolution maintained from the disk through the inner CGM out to 100 kpc — higher resolution, and extended over a larger region, than any previous work — combined with the well-studied IllustrisTNG stellar and AGN feedback model. Stellar radiation is then added in post-processing with the COLT radiative-transfer code for realistic ion populations.

02 The simulations

Four galaxies, three refinement levels

The suite comprises four Milky Way-mass halos — two drawn from the Auriga project and two TNG50 subhalos — each re-simulated with the fixed-volume refinement scheme. Because the resolution levels are chosen so the interstellar medium stays essentially unchanged, the galaxies themselves are consistent between runs (stellar masses agree to within 4% at z = 0).

The galaxies z = 0 properties

GalaxyOriginM★ (1010 M☉)M200 (1012 M☉)R200 (kpc)Max resolution
Au6Auriga halo 64.81.04214200 pc
Au8Auriga halo 83.01.08216500 pc
TNG-A (537941)TNG50 subhalo5.21.02212200 pc
TNG-B (519311)TNG50 subhalo9.21.39236500 pc
Auriga masses from Grand et al. (2017); TNG values from the public halo catalog. Initial conditions from the Auriga public data release and TNG50 zoom ICs.

Resolution levels

Each galaxy is run with the default mass-based scheme plus up to three fixed-volume levels. The target cell volume Vtarget is applied within a comoving radius Rri = 100 ckpc and tapers to the default scheme by Rro = 200 ckpc; inside 0.15 R200 the disk/ISM stays at the fiducial TNG50-1 resolution.

RunRefinementVtarget (kpc3)Spatial resolution
DefaultMass-based only—TNG50-1 fiducial
1 kpc+ fixed volume1.01 kpc
500 pc+ fixed volume0.125500 pc
200 pc+ fixed volume0.015200 pc (Au6, TNG-A)
Fixed maximum target volume Vtarget,max = 100 kpc3. The finest level is set so that mean cell sizes within the ISM match across all runs.
03 Key results

What higher CGM resolution changes

Cool gas emerges

H I and Mg II column densities rise by ~4 orders of magnitude as resolution increases, bringing them into agreement with COS-Halos observations. Higher resolution produces many more small, cool clouds.

O VI stays put

Total gas and O VI columns barely change with resolution, but the O VI-bearing gas becomes far more filamentary and structured. As in prior work, O VI still undercuts observations by a factor of a few.

Radiation matters

Adding stellar ionizing radiation with COLT reduces H I toward observed values, while Mg II and O VI are largely unaffected — tempering the pure-resolution effect while staying consistent with data.

The galaxy is stable

Stellar masses, gas masses, and star-formation rates are essentially unchanged by the added refinement — the extra resolution reshapes the CGM without perturbing the galaxy it surrounds.

Cloud populations

The number of small cold clouds keeps growing with resolution while the count of large clouds converges. Their mass–size relation follows the local Larson relation with a slope of ~2.2.

Sharper interfaces

The warm intermediate-temperature boundary layers around small clouds shrink at higher resolution, giving sharper, more rapid transitions to the ambient CGM temperature.

04 Data access

Getting the data via Globus

The ENGAWA data are hosted on a Globus collection. Globus is the standard tool for moving large scientific datasets: it manages reliable, resumable, high-throughput transfers between endpoints so you don't have to babysit multi-terabyte downloads. Access is free with any institutional or Globus ID login.

Collection / endpoint ID a4a95e3b-cd86-4a58-acf2-d68291249997 Open in Globus web app →

How to access

Open the collection

Follow the file-manager link and log in with your institutional identity or a free Globus ID. You'll land at the root of the ENGAWA collection.

Choose a destination

In the second Globus pane, select where the files should go: your institution's managed endpoint, or your own laptop/desktop after installing Globus Connect Personal.

Transfer

Select the galaxies/snapshots you need and hit Start. Globus handles retries and integrity checks; you'll be emailed when it completes. Individual small files can also be pulled straight over HTTPS from the web view.

From the command line

With the Globus CLI (pip install globus-cli):

# authenticate once
globus login

# set the ENGAWA collection as a shell variable
EP=a4a95e3b-cd86-4a58-acf2-d68291249997

# browse the tree
globus ls $EP:/
globus ls $EP:/Au6/200pc/

# recursive transfer to your endpoint (DEST_EP)
globus transfer $EP:/Au6/ DEST_EP:/data/engawa/Au6/ \
    --recursive --label "ENGAWA Au6"

Replace DEST_EP with your own endpoint UUID (globus endpoint search to find it). Adjust the source paths to match the layout shown at right.

05 Citation

Using ENGAWA in your work

If you use these simulations, please cite the ENGAWA paper. Please also acknowledge the Auriga and IllustrisTNG projects, whose initial conditions and feedback model this suite builds on.

@article{Lucchini2026_ENGAWA,
  author  = {Lucchini, S. and Abramson, C. and Hummels, C. and
             Conroy, C. and Hernquist, L. and Smith, A.},
  title   = {{ENhanced Galactic Atmospheres With Arepo:
             Resolving the CGM at 200 pc with the ENGAWA Simulations}},
  journal = {submitted to ApJ},
  year    = {2026},
  eprint  = {2603.05584},
  archivePrefix = {arXiv}
}