NewCluster: Virgo-Mass Zoom Simulation

• NewCluster simulation is a high-resolution zoom-in realization of a Virgo-mass galaxy cluster, bridging the gap between sub-kpc galaxy scales and large-volume cluster simulations.
• It employs stellar mass resolution of 2×10⁴ M⊙, 68 pc spatial resolution, and 15 Myr snapshot cadence to accurately capture processes like galaxy assembly, ram-pressure stripping, and intracluster light formation.
• The simulation integrates advanced hydrodynamics, feedback, chemical evolution, dust modeling, and Monte Carlo tracer particles, enabling detailed studies of galaxy kinematics and environmental dynamics.

NewCluster is a high-resolution zoom-in realization of a massive galaxy cluster in a cosmological context, introduced as the massive-halo counterpart of modern cosmological galaxy-evolution zoom simulations. It targets a $4.1\sigma$ overdensity expected to evolve into a Virgo-like system with $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$, follows the region from $z=50$ to $z\simeq0.8$, and is defined by a stellar mass resolution of $2\times10^4\,M_\odot$, a best physical spatial resolution of $\simeq68$ pc in AMR cells, and a snapshot cadence of 15 Myr. The simulation is designed to study early massive-galaxy build-up, dwarf-galaxy formation, galaxy kinematics, ram-pressure stripping, intracluster light, and gas-flow histories inside a single Virgo-mass cluster environment (Han et al., 8 Jul 2025).

1. Position within cluster-simulation research

NewCluster was motivated by a specific scale gap in numerical cosmology: sub-kpc “galaxy-scale” zooms such as NewHorizon and NewHorizon2 on one side, and large-volume cluster runs such as Cluster-EAGLE and TNG-Cluster on the other. Its stated purpose is to bridge that gap by following the formation and evolution of a single Virgo-mass object with unusually fine mass, spatial, and temporal resolution (Han et al., 8 Jul 2025).

This design differs from large statistical suites. TNG-Cluster, for example, re-simulates 352 cluster regions drawn from a $1\,{\rm Gpc}\,h^{-1}$ volume, keeps the IllustrisTNG physical model fixed, and operates at an effective TNG300-1 resolution with gas-cell mass $m_{\rm gas}\approx1.2\times10^7\,M_\odot$ and minimum adaptive gas softening $\epsilon_{\rm gas,min}=370$ pc (Nelson et al., 2023). NewCluster instead concentrates computational effort on a single high-density region and pushes stellar-mass resolution to $2\times10^4\,M_\odot$ and AMR resolution to $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$0 pc (Han et al., 8 Jul 2025). This suggests a complementary role: TNG-Cluster emphasizes statistics at the high-mass end, whereas NewCluster emphasizes local physical fidelity inside one Virgo-like assembly history.

2. Initial conditions, target halo, and resolution architecture

The target halo was selected from a dark-matter-only run in a $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$1 box with WMAP7 cosmology, using $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$2, $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$3, and $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$4. At $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$5, the selected halo has $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$6. The high-resolution Lagrangian region was defined by a convex hull around all particles within $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$7 of that target at $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$8; this region has an initial overdensity of $M_{\rm vir}\simeq5\times10^{14}\,M_\odot$9 above the cosmic mean, encloses $z=50$0 of matter, and spans a comoving volume of $z=50$1. With $z=50$2 Mpc at $z=50$3, the zoom-in radius corresponds to $z=50$4 Mpc in physical units (Han et al., 8 Jul 2025).

A compact summary of the principal resolution scales is useful because they largely determine the accessible science.

Quantity	Value	Notes
Stellar particle mass	$z=50$5	Minimum star-particle mass
DM particle mass	$z=50$6	Finest level, $z=50$7
Spatial resolution	$z=50$8 pc	Physical at $z=50$9, level 21
Snapshot interval	15 Myr	459 outputs to $z\simeq0.8$0

Refinement proceeds on mass and Jeans criteria, with a smooth transition at scale-factor thresholds $z\simeq0.8$1:

$z\simeq0.8$2

where $z\simeq0.8$3 is either the mass-refinement threshold $z\simeq0.8$4 or the Jeans parameter $z\simeq0.8$5 (Han et al., 8 Jul 2025).

3. Hydrodynamics, feedback, chemistry, dust, and tracer infrastructure

NewCluster is run with RAMSES-yOMP, a hybrid MPI+OpenMP branch of RAMSES that includes load-balancing and Poisson-solver optimizations. Gas dynamics are evolved on an AMR mesh, while star formation follows a gravo-thermo-turbulent prescription above $z\simeq0.8$6; star particles are formed with mass $z\simeq0.8$7 via a Poisson draw (Han et al., 8 Jul 2025).

The feedback model combines stellar and black-hole channels. Mechanical supernova feedback is injected once star-particle ages satisfy $z\simeq0.8$8 Myr, with

$z\simeq0.8$9

and Type Ia supernovae follow a delay-time distribution

$2\times10^4\,M_\odot$0

AGN feedback is dual-mode, with quasar and radio channels following Dubois et al. (2012). Black-hole seeds form in cells satisfying $2\times10^4\,M_\odot$1 and $2\times10^4\,M_\odot$2, and the dynamical-friction boost is

$2\times10^4\,M_\odot$3

These choices were made to support simultaneous studies of galaxy assembly, AGN-driven outflows, and environmental processing in the intracluster medium (Han et al., 8 Jul 2025).

Chemical evolution is computed on-the-fly for ten elements—H, D, He, C, N, O, Mg, Si, S, and Fe—and dust is treated as two compositions, carbonaceous and silicate, each split into two size bins, “small” at 5 nm and “large” at $2\times10^4\,M_\odot$4m. The dust model includes stellar condensation from SNII, SNIa, and AGB sources, accretion in the ISM through a subgrid log-normal density PDF, thermal sputtering, supernova destruction, shattering, and coagulation (Han et al., 8 Jul 2025). A later morphology study made this infrastructure explicit by evolving dust as a passive scalar with on-the-fly accretion, supernova destruction, thermal sputtering, astration, coagulation, and shattering; that analysis used the same two-size approximation and showed that the model reproduces observed dust-to-gas versus metallicity scalings (Byun et al., 25 Aug 2025).

A distinctive numerical element is the use of $2\times10^4\,M_\odot$5 Monte Carlo tracer particles, exchanged probabilistically between gas cells, stars, and black holes in proportion to mass fluxes. This Lagrangian tagging is intended to compensate for the Eulerian difficulty of following individual fluid elements and enables direct analyses of inflow, outflow, recycling, stripping, and particle-origin histories (Han et al., 8 Jul 2025).

4. Resolved phenomena and early physical results

At its nominal resolution, NewCluster is intended to resolve galaxy assembly across a wide dynamic range. The run resolves ultra-faint dwarfs and low-surface-brightness features down to $2\times10^4\,M_\odot$6 at $2\times10^4\,M_\odot$7, while massive galaxies with $2\times10^4\,M_\odot$8 exhibit sub-100 pc disc structure and early compaction events. Stellar half-mass radii and kinematic ratios $2\times10^4\,M_\odot$9 evolve in line with observations, and the 68 pc resolution in the densest regions allows rotation curves, velocity-dispersion maps, and disc settling to be followed out to $\simeq68$0 (Han et al., 8 Jul 2025).

The 15 Myr output cadence is central to the simulation’s environmental science. It captures rapid changes during pericentric passages, enabling phase-space tracking of quenching and tidal mass loss. Tracer-particle analysis around a central galaxy at $\simeq68$1 resolves distinct inflow, outflow, and recycled components; the outflows trace bipolar AGN jets, while recycled gas returns to fuel star formation, illustrating a galactic fountain (Han et al., 8 Jul 2025).

NewCluster also emphasizes cluster-specific gas dynamics. Even at $\simeq68$2, satellite galaxies near the cluster core experience ram pressure of

$\simeq68$3

and representative satellites display asymmetric cold-gas morphologies characteristic of jellyfish galaxies. Low-mass systems with $\simeq68$4 can be entirely stripped. At $\simeq68$5, the main halo approaches a $\simeq68$6 merger, and compression at the interface boosts ICM temperature and pressure, increasing ram pressure on galaxies caught between the two systems. A plausible implication is that NewCluster is optimized not merely for equilibrium cluster structure, but for time-dependent preprocessing, satellite transformation, and pre-merger environmental amplification (Han et al., 8 Jul 2025).

5. Follow-up analyses: morphology, dust evolution, and intracluster light

The first major follow-up study of NewCluster’s internal baryonic modeling examined how dust prescriptions alter mock JWST morphologies. In that work, mock images were generated with SKIRT using $\simeq68$7 photon packets, 16 viewing directions, a spatial resolution of $\simeq68$8, and JWST bands including F277W and F356W. The comparison was between the on-the-fly dust model inherited from NewCluster and a fixed dust-to-metal model with DTM $\simeq68$9 in cells with $1\,{\rm Gpc}\,h^{-1}$0 K. Morphology was quantified using the Gini coefficient, $1\,{\rm Gpc}\,h^{-1}$1, Concentration, and Asymmetry. The on-the-fly models show systematically higher concentration and Gini, especially at $1\,{\rm Gpc}\,h^{-1}$2 and $1\,{\rm Gpc}\,h^{-1}$3, while fixed-DTM images under-predict central surface brightness by $1\,{\rm Gpc}\,h^{-1}$4. The same study identified a “DTM cavity,” defined by the ratio $1\,{\rm Gpc}\,h^{-1}$5, in which central bulge starbursts drive $1\,{\rm Gpc}\,h^{-1}$6 for $1\,{\rm Gpc}\,h^{-1}$7 Myr and depress inner-kpc DTM profiles by $1\,{\rm Gpc}\,h^{-1}$8 (Byun et al., 25 Aug 2025).

A separate follow-up focused on the origin of intracluster light. That analysis used AdaptaHOP in HaloMaker and GalaxyMaker modes together with the YoungTree merger-tree algorithm, which scores halo matches with a shared-particle term, a kinematic term, and a stellar-mass term, and evaluates candidate links across up to four snapshots, or $1\,{\rm Gpc}\,h^{-1}$9 Myr. Stellar particles were then classified as “tight” or “loose” members according to whether they remained associated with a galaxy for $m_{\rm gas}\approx1.2\times10^7\,M_\odot$0 or $m_{\rm gas}\approx1.2\times10^7\,M_\odot$1–$m_{\rm gas}\approx1.2\times10^7\,M_\odot$2 of the relevant snapshots, and were assigned to four ICL channels: stripped, disrupted, in situ, and preprocessed. Within $m_{\rm gas}\approx1.2\times10^7\,M_\odot$3 at $m_{\rm gas}\approx1.2\times10^7\,M_\odot$4, the resulting BCG+ICL sample has $m_{\rm gas}\approx1.2\times10^7\,M_\odot$5, about 41% of the cluster’s total stellar mass; its fractional contributions are 33.4% stripped from surviving satellites, 22.0% stripped from disrupted satellites, 32.3% formed in situ in the BCG, and 12.3% preprocessed. Among these components, the preprocessed channel is the shallowest in density and follows the dark-matter profile to better than 20% over $m_{\rm gas}\approx1.2\times10^7\,M_\odot$6–$m_{\rm gas}\approx1.2\times10^7\,M_\odot$7; it is also old, metal-poor, and $m_{\rm gas}\approx1.2\times10^7\,M_\odot$8-enhanced, with $m_{\rm gas}\approx1.2\times10^7\,M_\odot$9 dex. The stripped fraction of satellites increases most clearly with longer time since infall, smaller pericentric distance, and larger number of completed orbits (Jeon et al., 5 Dec 2025).

Taken together, these analyses show that NewCluster is not only a hydrodynamical run but also a platform for physically constrained radiative-transfer post-processing and particle-history decomposition. The morphology and ICL studies both exploit the simulation’s unusually fine spatial resolution and its high-cadence outputs, but they interrogate different observables: one emphasizes light redistribution by dust physics, the other emphasizes dynamical and chemical decomposition of diffuse stellar components.

6. Nomenclature, scope, and other uses of the label

The label “NewCluster” is not unique to the Virgo-mass cosmological zoom. In another 2024 study, “NewCluster” denotes a simulation of the collision and merger of two identical, gas-rich dwarf galaxies. There, each dwarf has virial mass $\epsilon_{\rm gas,min}=370$0, a stellar disk of $\epsilon_{\rm gas,min}=370$1, and a gaseous disk of $\epsilon_{\rm gas,min}=370$2; gas is represented by SPH particles of $\epsilon_{\rm gas,min}=370$3 with a gravitational softening length of 0.1 pc. That simulation was used to study bound star clusters with a mass function $\epsilon_{\rm gas,min}=370$4 and $\epsilon_{\rm gas,min}=370$5, an empirical scaling $\epsilon_{\rm gas,min}=370$6 with $\epsilon_{\rm gas,min}=370$7, a spatial-segregation measure

$\epsilon_{\rm gas,min}=370$8

with $\epsilon_{\rm gas,min}=370$9, and cluster coalescence constrained by $2\times10^4\,M_\odot$0; bound nearest-neighbor pairs constitute $2\times10^4\,M_\odot$1 percent by number over $2\times10^4\,M_\odot$2 Myr and up to $2\times10^4\,M_\odot$3 percent by mass for the youngest clusters during peak-SFR phases (Elmegreen et al., 2024).

In a different methodological usage, “NewCluster” refers not to an astrophysical simulation at all, but to an agglomerative clustering framework for empirical output distributions from stochastic simulations. That framework measures pairwise dissimilarity by the entropy-regularized Wasserstein distance

$2\times10^4\,M_\odot$4

uses complete-linkage agglomeration, and selects the stopping level by maximizing the silhouette index. Its reported use cases include anomaly detection, pre-optimization, and online monitoring, illustrated with a call-center case study (Ghasemloo et al., 2024).

The coexistence of these usages means that “NewCluster simulation” is context-dependent. In cosmological galaxy-cluster research, however, the term most commonly denotes the high-resolution Virgo-mass zoom-in simulation introduced in 2025 and subsequently extended to studies of dust-regulated morphology and intracluster-light origin (Han et al., 8 Jul 2025).