Astronomers Built the Most Detailed Galaxy Cluster Ever Simulated

Pick any galaxy in a cluster, and it's probably had a rough life. Clusters strip gas from spirals, crush dwarfs, and kill off star formation on billion-year timescales. Watching that unfold has always required a simulation the field didn't quite have — until now. In January 2026, a team led by S. Han and S. K. Yi at Yonsei University in Seoul, with colleagues at the Institut d'Astrophysique de Paris, published the first results from NEWCLUSTER in Astronomy & Astrophysics. They've already opened the early data to the public.

Clusters Are Ruthless — and That's Exactly Why They're Hard to Simulate

Galaxy clusters sit at the top of the cosmic food chain — hundreds of galaxies bound by gravity, bathed in hot gas, scaffolded by dark matter. They're also the most transformative environment a galaxy can pass through. A spiral that wanders in loses its cold gas to ram pressure, slows its star formation, and over a few billion years quietly dies. Whether it survives with its disk intact depends on details that happen well below cluster scales. Dwarf galaxies are especially tricky — so small that most large-volume runs never resolve them properly. For decades, researchers have had to choose: full cluster, or fine detail. Not both.

Why Bigger Simulations Have Always Meant Blurrier Galaxies

The problem isn't ignorance — it's arithmetic. Simulations like IllustrisTNG and EAGLE are genuinely impressive, but they spread resolution across enormous volumes. A typical large-box run achieves roughly 1 kiloparsec per element. Fine for counting galaxies, useless for watching one's insides. At 1 kpc, a Milky Way-sized disk gets maybe ten resolution elements across its full width. You're not studying a galaxy; you're studying a smeared-out blob that vaguely resembles one. Add a few hundred of those blobs to a cluster environment and the problem only multiplies.

How Does NewCluster Resolve a Full Cluster at Galaxy-Scale Detail?

The short answer: it doesn't pretend to be a box simulation. NEWCLUSTER targets a single overdense region — a 4.1-sigma peak in the early universe destined to collapse into something Virgo-sized. The zoom-in volume stretches 3.5 virial radii out from the cluster center, far enough to catch infalling galaxies before they hit the dense core. At 68 parsecs in the most refined regions, you're inside molecular cloud territory — the actual scale where stars form. The adaptive mesh cranks resolution up automatically wherever gas density spikes. Snapshot cadence matters too: most comparable runs output every 100–200 Myr, while NEWCLUSTER does it every 15 Myr. Ram-pressure stripping events can begin and end inside that window. The team also tracked ten chemical elements separately and ran a full dust lifecycle — grain formation, size change, destruction. Monte Carlo tracer particles were added to follow individual gas parcels, something the underlying Eulerian code can't do on its own.

What Researchers Can Now Actually Study That They Couldn't Before

Take dwarf galaxies. At 20,000 solar masses per particle, NEWCLUSTER can form and track dwarfs as they fall into the cluster — something previous cluster simulations couldn't do because dwarfs simply vanished below the resolution floor. You can't study tidal stripping if you can't see the galaxy being stripped. The Virgo comparison is deliberate too. The Virgo Cluster is 65 million light-years away and the most-observed cluster we have. Matching its virial mass — around 5 × 10¹⁴ solar masses — means every NEWCLUSTER result can be tested against real catalogued data. And because supernova feedback, AGN feedback, and chemical enrichment all run together, the simulation can trace causal chains that need all three: how iron from one galaxy's supernovae ends up diffused through the intracluster gas and eventually inside a completely different galaxy.

What's Still Missing — and What Comes Next

The paper calls itself a first-half report — and means it. The run hasn't reached z = 0 yet, so the final assembled cluster is still coming. The Eulerian gas method struggles with diffuse stripped material on the outskirts; the Monte Carlo tracers help, but that output is still being validated. AGN feedback behavior across the full mass range — from dwarf to brightest cluster galaxy — is another open question. What's notable is that the team didn't wait. Data up to z = 0.8 is already public. Other groups can run their own analyses on the chemistry, the dust, the dwarf population right now, while the simulation keeps running toward present day.

• Dwarfs are finally visible — At 20,000 solar masses per particle, NEWCLUSTER resolves small galaxies that vanish in every large-box cluster run, making it possible to study their fate in a real cluster environment for the first time.
• Dust and chemistry aren't afterthoughts — Ten elements plus a full dust grain lifecycle means outputs can be compared directly against infrared and submillimeter observations, not just optical data.
• Early access, no waiting — Data released at z = 0.8 before the simulation finishes lets other research teams start publishing science now.

"NEWCLUSTER is a novel high-resolution cluster simulation designed to serve as the massive halo counterpart of the modern cosmological galaxy evolution framework." — Han, Yi et al., Astronomy & Astrophysics, 2026.