Hyperion Proto-Supercluster

• Hyperion Proto-Supercluster is a vast multi-component structure at z~2.45, composed of interconnected proto-clusters and groups.
• Its 3D mapping employs density-field reconstructions from deep spectroscopy and high-quality photometric redshifts to reveal detailed substructure.
• Simulations predict that Hyperion will evolve into a filamentary supercluster with multiple virialized clusters rather than a single collapsed halo.

The Hyperion Proto-Supercluster is a massive, extended, multi-component proto-supercluster in the COSMOS field at $z\sim2.45$, first presented as a homogeneous three-dimensional overdensity reconstruction from VUDS, zCOSMOS, and COSMOS2015 data. It is not a single virialized cluster, but a vast, still-forming complex of multiple proto-clusters and proto-groups embedded in a larger collapsing network, extending over roughly $60\times60\times150\ {\rm cMpc}^3$ and containing at least seven dense peaks connected by filaments. Observationally, Hyperion has become a benchmark system for studying structure assembly, environmental effects on galaxy growth, and the relation between galaxy-defined overdensities and the intergalactic medium during cosmic noon (Cucciati et al., 2018).

1. Discovery, definition, and nomenclature

Hyperion was identified in the central COSMOS field through a 3D density-field reconstruction based on deep spectroscopy from VUDS and zCOSMOS, supplemented by high-quality photometric redshifts from COSMOS2015. The discovery paper assigned the official designation PSC J1001+0218 and framed Hyperion as a panoramic synthesis of several previously reported overdensities that had been detected piecemeal with heterogeneous tracers, including Ly$\alpha$ emitters, dusty star-forming galaxies, CO emitters, proto-group searches, and Ly$\alpha$-forest tomography (Cucciati et al., 2018).

The defining interpretation is explicitly supercluster-scale rather than cluster-scale. Hyperion was described as a connected overdense volume with an estimated total mass of $\sim4.8\times10^{15}\ M_\odot$, containing at least seven density peaks within $2.4$\sim0.1\times10^{14}\ M_\odot$ to $\sim2.7\times10^{14}\ M_\odot$ (Cucciati et al., 2018). Later work retained this description and emphasized that Hyperion is a collection of seven groups or peaks connected by lower-density filamentary regions, making it an especially useful laboratory for studying galaxies across a broad environmental dynamic range at $z\sim2.45$ (Gururajan et al., 11 Apr 2025).

The terminology surrounding Hyperion is partly semantic and partly dynamical. Some studies refer to it as a “protocluster,” others as an “aggregation” or “supercluster,” but the consistent point is that it is not a single collapsed halo. A later constrained-simulation analysis explicitly adopted the view that the various overdensities at $2.423\le z^{\rm obs}\le2.507$, spanning approximately $60\times60\times150\ {\rm cMpc}^3$0, are likely a connected structure with seven distinct density peaks, and that this connectedness is the salient physical attribute (Ata et al., 2022).

2. Three-dimensional mapping and spectroscopic consolidation

The original reconstruction of Hyperion relied on 2D Voronoi tessellation in overlapping redshift slices converted into a 3D density field. In that framework, spectroscopic objects were retained probabilistically according to redshift-quality flags, photometric-only galaxies were repeatedly resampled from their redshift uncertainties, and slice-by-slice surface-density fields were combined into a cube. The galaxy overdensity was defined as

$60\times60\times150\ {\rm cMpc}^3$1

with structure identification performed through thresholds in $60\times60\times150\ {\rm cMpc}^3$2. Hyperion itself was defined as a connected $60\times60\times150\ {\rm cMpc}^3$3 overdense region, while the major substructures were isolated with a $60\times60\times150\ {\rm cMpc}^3$4 threshold (Cucciati et al., 2018).

Subsequent observations substantially improved the spectroscopic map. The HST-Hyperion survey added 50 orbits of WFC3/F160W imaging and WFC3/G141 slitless grism spectroscopy over 25 pointings targeted on the most overdense Hyperion regions, analyzed jointly with adjacent 3D-HST COSMOS data comprising 56 orbits and 28 pointings. Using GRIZLI and visually inspected grism fits, that program inspected 12,814 objects with $60\times60\times150\ {\rm cMpc}^3$5, of which 5,629, or 44%, received reliable grism redshifts. Comparison to high-quality ground-based spectroscopy yielded $60\times60\times150\ {\rm cMpc}^3$6. The resulting catalogs contain 125 confirmed members of the Hyperion structure within $60\times60\times150\ {\rm cMpc}^3$7, plus an additional 71 confirmed galaxies in projection within $60\times60\times150\ {\rm cMpc}^3$8 (Forrest et al., 6 Mar 2025).

These HST data materially altered the practical observability of Hyperion. Within the broader cube $60\times60\times150\ {\rm cMpc}^3$9, $\alpha$0, and $\alpha$1, the combined spectroscopy now exceeds 650 galaxies with high-quality redshifts, and the HST component alone contributes 196 galaxies with high-quality grism redshifts in that interval, 149 of them new. The survey therefore shifted Hyperion from a partially sampled overdensity map to a system with a much denser, more spatially uniform membership census (Forrest et al., 6 Mar 2025).

Later HST-Hyperion analyses extended the same logic to environment-tagged galaxy samples using COSMOS2020 photometry with ground-based and new HST grism spectroscopy to build a 3D overdensity map assigning galaxies to peaks, outskirts, or field. That workflow preserved the original emphasis on probabilistic treatment of redshift and mass uncertainties and reinforced the view that Hyperion should be analyzed as a structured environment rather than as a single object (Sikorski et al., 2 Sep 2025).

3. Internal architecture and overdense components

Hyperion has a strongly anisotropic internal morphology. The original 3D reconstruction showed two visually dominant components, a South-West component at lower redshift around $\alpha$2 and a North-East component at slightly higher redshift around $\alpha$3, linked by an intermediate overdense bridge or filamentary region. The connected overdense volume enclosed by the $\alpha$4 surface is about $\alpha$5, with average galaxy overdensity $\alpha$6 (Cucciati et al., 2018).

Within that volume, seven $\alpha$7 peaks were isolated. Four were given mythological names—Theia, Eos, Helios, and Selene—while three remained unnamed in the original nomenclature. Their measured peak redshifts span $\alpha$8 to $\alpha$9, and their overdensity-based masses span a factor of about 30 (Cucciati et al., 2018).

Peak	Name	Key value
[1]	Theia	$\alpha$0; $\alpha$1
[2]	Eos	$\alpha$2; $\alpha$3
[3]	Helios	$\alpha$4; $\alpha$5
[4]	Selene	$\alpha$6; $\alpha$7
[5]	—	$\alpha$8; $\alpha$9
[6]	—	$\sim4.8\times10^{15}\ M_\odot$0; $\sim4.8\times10^{15}\ M_\odot$1
[7]	—	$\sim4.8\times10^{15}\ M_\odot$2; $\sim4.8\times10^{15}\ M_\odot$3

The peak census was later re-tabulated in the HST-Hyperion survey, which confirmed member galaxies in most of the $\sim4.8\times10^{15}\ M_\odot$4 peaks and quantified the densest known region at

$\sim4.8\times10^{15}\ M_\odot$5

corresponding to the compact X-ray-detected Peak [5], with quoted velocity dispersion

$\sim4.8\times10^{15}\ M_\odot$6

That region had already been associated with the Wang et al. structure, but its physical interpretation remained unsettled (Forrest et al., 6 Mar 2025).

A recurring controversy concerns whether one of Hyperion’s cores was already a bona fide cluster traced by hot intracluster gas. A detailed gas study of a $\sim4.8\times10^{15}\ M_\odot$7 Hyperion-associated protocluster core re-analyzed stacked Chandra/XMM data and rejected earlier claims of confirmed intracluster-medium emission, arguing instead that at least $\sim4.8\times10^{15}\ M_\odot$8 of extended X-ray sources of similar luminosity and size at this redshift arise from inverse Compton scattering off recently extinguished radio galaxies rather than from intracluster gas (Champagne et al., 2021). The HST-Hyperion survey correspondingly treated the X-ray origin in Peak [5] as unresolved, rather than as definitive evidence of cluster virialization (Forrest et al., 6 Mar 2025).

4. Environmental effects on galaxies inside Hyperion

Hyperion has increasingly been used as an environmental laboratory rather than solely as a structure-finding result. One line of work concerns merger and interaction activity. Using COSMOS2020, archival spectroscopy, C3VO, and the HST-Hyperion grism data, the close kinematic companion fraction was measured with a Monte Carlo methodology designed to combine spectroscopy and photometric redshift PDFs. After correction for sampling incompleteness, the analysis found

$\sim4.8\times10^{15}\ M_\odot$9

for Hyperion and

$2.4

for the coeval field, a $2.4Giddings et al., 6 Mar 2025).

A second line of work addresses gas reservoirs and star formation. A multiwavelength study combining COSMOS2020, COSMOS Super-deblended, A$2.4

$2.4

adopting $2.4SFR relation had slope $\sim0.1\times10^{14}\ M_\odot$0. The mean depletion timescale was $\sim0.1\times10^{14}\ M_\odot$1 Myr and the mean gas fraction was $\sim0.1\times10^{14}\ M_\odot$2. The authors characterized these as tentative, mostly $\sim0.1\times10^{14}\ M_\odot$3-level indications that denser Hyperion environments host higher SFRs, shorter depletion times, and somewhat lower gas fractions (Gururajan et al., 11 Apr 2025).

A third result concerns the stellar-mass function. In the HST-Hyperion survey analysis of the SMF at $\sim0.1\times10^{14}\ M_\odot$4, 100 Monte Carlo realizations were used to propagate redshift and mass uncertainties after assigning galaxies to peaks, outskirts, or field. The overdense peaks show a clear excess of massive galaxies: number densities at $\sim0.1\times10^{14}\ M_\odot$5 are $\sim0.1\times10^{14}\ M_\odot$6 higher than the field, while those at $\sim0.1\times10^{14}\ M_\odot$7 are enhanced by only $\sim0.1\times10^{14}\ M_\odot$8. By contrast, the outskirts and Hyperion as a whole mirror the field (Sikorski et al., 2 Sep 2025). This supports a picture in which the densest peaks, not the proto-supercluster-averaged environment, are where accelerated stellar-mass assembly is most readily visible.

Taken together, these studies suggest that different observables respond to different environmental scales inside Hyperion. The global structure can resemble the field in some aggregate statistics, while the densest peaks already show elevated massive-galaxy abundance, enhanced close-companion incidence, and signs of faster gas consumption. A plausible implication is that Hyperion’s environmental processing is spatially localized and stage-dependent rather than uniform across the full $\sim0.1\times10^{14}\ M_\odot$9 complex.

5. Multi-tracer views: Ly$\sim2.7\times10^{14}\ M_\odot$0 emitters, IGM tomography, and AGN

Hyperion has also been reconstructed with tracers other than continuum-selected galaxies. Wide-field narrow-band imaging around $\sim2.7\times10^{14}\ M_\odot$1 produced a sample of 157 Ly$\sim2.7\times10^{14}\ M_\odot$2 emitters over 3,455 arcmin$\sim2.7\times10^{14}\ M_\odot$3, and the LAE distribution was found to closely mirror the known spectroscopic members, tracing both protocluster cores and extended filamentary arms. The inferred LAE overdensity was $\sim2.7\times10^{14}\ M_\odot$4 within an effective volume of $\sim2.7\times10^{14}\ M_\odot$5, a value the authors regarded as too high for a single protocluster and therefore consistent with Hyperion being composed of multiple protoclusters. The same work reported an AGN population about $\sim2.7\times10^{14}\ M_\odot$6 times more abundant than in the field, while also noting weak evidence that LAEs may be less abundant in the highest H I regions, perhaps because Ly$\sim2.7\times10^{14}\ M_\odot$7 is suppressed there (Huang et al., 2022).

A distinct perspective comes from Ly$\sim2.7\times10^{14}\ M_\odot$8-forest tomography. The LATIS survey reconstructed IGM maps from 3,012 background sight lines and examined Hyperion as the richest overdense region in the COSMOS/D2 volume. In the tomographic framework, the transmitted-flux contrast is

$\sim2.7\times10^{14}\ M_\odot$9

with maps Wiener filtered and then smoothed with $z\sim2.45$0. LATIS identified six IGM-selected overdensities within Hyperion and reported a new eastern galaxy-density component,

$z\sim2.45$1

with $z\sim2.45$2, labeled H8 and associated with the Ly$z\sim2.45$3 absorption peak LATIS2-D2-18. For that IGM peak, the reported quantities are $z\sim2.45$4, significance $z\sim2.45$5, $z\sim2.45$6, and $z\sim2.45$7 (Newman et al., 15 Jul 2025).

The LATIS analysis is notable because it showed that Hyperion is not only a galaxy-defined superstructure but also a broad, connected IGM absorption complex. At the same time, the paper emphasized that galaxy-density peaks and IGM minima do not correspond in a one-to-one fashion; Hyperion is better represented as a continuous absorption field with embedded peaks than as a set of isolated matched centroids (Newman et al., 15 Jul 2025). This resolves an apparent misconception that different tracers should recover identical substructure geometry.

LATIS also tested whether the gas in Hyperion requires non-standard ionization physics. Comparing the observed Ly$z\sim2.45$8 transmission at the galaxy density peaks H1–H8 to mock conditional distributions $z\sim2.45$9, the authors found mean observed absorption $2.423\le z^{\rm obs}\le2.507$0, consistent with the mock expectation $2.423\le z^{\rm obs}\le2.507$1. They interpreted this as evidence that, on $2.423\le z^{\rm obs}\le2.507$2 scales, Hyperion’s gas is consistent with a simple fluctuating Gunn-Peterson approximation and that active galactic nucleus feedback or other processes have not affected the large-scale gas ionization within the structure as a whole (Newman et al., 15 Jul 2025).

6. Dynamical state and predicted descendant

The discovery analysis already suggested that Hyperion’s peaks are at different dynamical stages. Using elongation-corrected overdensities and a spherical-collapse model, the seven main peaks were inferred to have already started or to be about to start collapsing, with predicted collapse or virialization redshifts spanning approximately $2.423\le z^{\rm obs}\le2.507$3 to $2.423\le z^{\rm obs}\le2.507$4. Peak [6] emerged as the most evolved and Peak [2] as the least evolved, reinforcing that Hyperion is a network of overdensities at different stages of collapse rather than a monolithic object (Cucciati et al., 2018).

Later constrained cosmological simulations sharpened the endpoint of that evolution. Using COSTCO constrained simulations based on spectroscopic surveys in the central square degree of COSMOS, Hyperion was matched in all 50 realizations. The key result is explicit: Hyperion does not collapse into one single cluster by $2.423\le z^{\rm obs}\le2.507$5. Instead, “on average, four virialized clusters will coalesce out of Hyperion to form a massive filamentary group of clusters,” and “a partial merging of Hyperion is likely by $2.423\le z^{\rm obs}\le2.507$6, but we never witnessed a complete merging of all the constituent peaks to form one cluster only” (Ata et al., 2022).

The simulated $2.423\le z^{\rm obs}\le2.507$7 descendant is therefore a filamentary supercluster-like complex rather than a single Coma-like terminal halo. The most massive node reaches

$2.423\le z^{\rm obs}\le2.507$8

the total mass within collapsed cluster halos is

$2.423\le z^{\rm obs}\le2.507$9

and the spatial span of the collapsed system is

$60\times60\times150\ {\rm cMpc}^3$00

The abstract rendered this more approximately as a giant filamentary supercluster spanning 100 Megaparsecs (Ata et al., 2022).

This descendant picture revises, but does not negate, the original discovery interpretation. The 2018 observational reconstruction argued for a proto-supercluster based on Hyperion’s mass, extent, multiplicity of peaks, and filamentary connectivity (Cucciati et al., 2018). The constrained simulations support that multi-component interpretation while narrowing the dynamical claim: Hyperion’s likely progeny is not one virialized object, but multiple massive cluster cores embedded within a giant filamentary supercluster (Ata et al., 2022). In that sense, the term proto-supercluster is not merely descriptive; it is the dynamical category that best matches Hyperion’s predicted fate.