arXiv subjects

Sebastian Trujillo-Gomez

Publications and source records attributed to Sebastian Trujillo-Gomez.

At least 19 recordsLinked to original sources

arXiv2025

EuroHPC SPACE CoE: Redesigning Scalable Parallel Astrophysical Codes for Exascale

High Performance Computing (HPC) based simulations are crucial in Astrophysics and Cosmology (A&C), helping scientists investigate and understand complex astrophysical phenomena. Taking advantage of exascale computing capabilities is essential for these efforts. However, the unprecedented architectural complexity of exascale systems impacts legacy codes. The SPACE Centre of Excellence (CoE) aims to re-engineer key astrophysical codes to tackle new computational challenges by adopting innovative programming paradigms and software (SW) solutions. SPACE brings together scientists, code developers, HPC experts, hardware (HW) manufacturers, and SW developers. This collaboration enhances exascale A&C applications, promoting the use of exascale and post-exascale computing capabilities. Additionally, SPACE addresses high-performance data analysis for the massive data outputs from exascale simulations and modern observations, using machine learning (ML) and visualisation tools. The project facilitates application deployment across platforms by focusing on code repositories and data sharing, integrating European astrophysical communities around exascale computing with standardised SW and data protocols.

astro-ph.IM↗

arXiv2025

Machine Learning Workflow for Morphological Classification of Galaxies

As part of the EU-funded Center of Excellence SPACE (Scalable Parallel Astrophysical Codes for Exascale), seven commonly used astrophysics simulation codes are being optimized to exploit exascale computing platforms. Exascale cosmological simulations will produce large amounts of data (i.e. several petabytes) that will soon be waiting to be analyzed, with enormous potential for scientific breakthroughs. Our tool Spherinator enables the reduction of these complex data sets to a low-dimensional space using Generative Deep Learning to understand the morphological structure of simulated galaxies. A spherical latent space allows the HiPSter module to provide explorative visualization using Hierarchical Progressive Surveys (HiPS) in the Aladin software. Here we present a machine-learning workflow covering all stages, from data collection to preprocessing, training, prediction, and final deployment. This workflow ensures full reproducibility by keeping track of the code, data, and environment. Additionally, the workflow allows for scalability in managing a large amount of data and complex pipelines. We use only open source software and standards that align with the FAIR (Findability, Accessibility, Interoperability and Reproducibility) principles. In this way, we are able to distribute the workflow reliably and enable collaboration by sharing code, data, and results efficiently.

astro-ph.IM↗

arXiv2025

JAvaScript Multimodal INformation Explorer

Astronomical data is rich in volume, information and facets. Although this offers multiple research perspectives, processing the data remains a challenge. Infrastructures for analyzing, inspecting, exploring and communicating with data are mandatory. To address this issue, we introduce Jasmine, the JAvaScript Multimodal INformation Explorer. Jasmine allows users to open different data viewer modals that show a specific data point from a set. The viewer currently supports image data, as well as point cloud objects. Users can decide on which information about the data point they like to have displayed. Point clouds are interactive and allow for zooming, tossing, and turning. Picking a data point is enabled by providing a structured view of the set, arranged by a key property. This arrangement is achieved by autoencoding.

astro-ph.IM↗

arXiv2024

The cosmic globular cluster formation history in the E-MOSAICS simulations

We present a comprehensive analysis of globular cluster (GC) formation and evolution across the $34^3$ Mpc$^3$ volume of the E-MOSAICS galaxy formation simulations. Defining GCs as surviving, high-mass ($>10^5$ M$_\odot$) clusters, we analyse their formation histories as a function of their metallicity and host galaxy mass, also distinguishing between central and satellite galaxies. The redshift of peak GC formation rate increases weakly with galaxy mass, decreases with metallicity, and does not differ between centrals and satellites. The epoch of peak GC formation precedes that of the stars by a factor of $1.1{-}1.6$, primarily due to `downsizing', i.e. low-mass galaxies form their stars later. Consequently, this offset decreases with galaxy mass, leading to nearly coeval stellar and GC populations in massive galaxies ($>10^{11}$ M$_\odot$). GCs themselves do not exhibit strong downsizing, because they predominantly formed at early cosmic epochs conducive to the formation (through high gas pressures) and survival (through high galaxy merger and GC migration rates) of massive, compact stellar systems. The total GC formation rate in the volume peaks at $z\approx 2.5$, shortly before star formation peaks at $z\approx 2$, but well after the general cluster formation rate at $z\approx 4$, reflecting a survivor bias where surviving GCs formed more recently. We find that GC formation commenced early, at $z>10$, such that the results of this work may provide a framework for interpreting direct observations of proto-GC formation with the JWST, especially as these observations accumulate to enable statistical studies.

astro-ph.GA↗

arXiv2024

Spherinator and HiPSter: Representation Learning for Unbiased Knowledge Discovery from Simulations

Simulations are the best approximation to experimental laboratories in astrophysics and cosmology. However, the complexity, richness, and large size of their outputs severely limit the interpretability of their predictions. We describe a new, unbiased, and machine learning based approach to obtaining useful scientific insights from a broad range of simulations. The method can be used on today's largest simulations and will be essential to solve the extreme data exploration and analysis challenges posed by the Exascale era. Furthermore, this concept is so flexible, that it will also enable explorative access to observed data. Our concept is based on applying nonlinear dimensionality reduction to learn compact representations of the data in a low-dimensional space. The simulation data is projected onto this space for interactive inspection, visual interpretation, sample selection, and local analysis. We present a prototype using a rotational invariant hyperspherical variational convolutional autoencoder, utilizing a power distribution in the latent space, and trained on galaxies from IllustrisTNG simulation. Thereby, we obtain a natural Hubble tuning fork like similarity space that can be visualized interactively on the surface of a sphere by exploiting the power of HiPS tilings in Aladin Lite.

astro-ph.IM↗

arXiv2024

HI discs of L$_{\ast}$ galaxies as probes of the baryonic physics of galaxy evolution

Understanding what shapes the cold gas component of galaxies, which both provides the fuel for star formation and is strongly affected by the subsequent stellar feedback, is a crucial step towards a better understanding of galaxy evolution. Here, we analyse the HI properties of a sample of 46 Milky Way halo-mass galaxies, drawn from cosmological simulations (EMP-Pathfinder and FIREbox). This set of simulations comprises galaxies evolved self-consistently across cosmic time with different baryonic sub-grid physics: three different star formation models [constant star formation efficiency (SFE) with different star formation eligibility criteria, and an environmentally-dependent, turbulence-based SFE] and two different feedback prescriptions, where only one sub-sample includes early stellar feedback. We use these simulations to assess the impact of different baryonic physics on the HI content of galaxies. We find that the galaxy-wide HI properties agree with each other and with observations. However, differences appear for small-scale properties. The thin HI discs observed in the local Universe are only reproduced with a turbulence-dependent SFE and/or early stellar feedback. Furthermore, we find that the morphology of HI discs is particularly sensitive to the different physics models: galaxies simulated with a turbulence-based SFE have discs that are smoother and more rotationally symmetric, compared to those simulated with a constant SFE; galaxies simulated with early stellar feedback have more regular discs than supernova-feedback-only galaxies. We find that the rotational asymmetry of the HI discs depends most strongly on the underlying physics model, making this a promising observable for understanding the physics responsible for shaping the interstellar medium of galaxies.

astro-ph.GA↗

arXiv2024

On the origin of globular clusters in a hierarchical Universe

We present an end-to-end description of the formation process of globular clusters (GCs) which combines a treatment for their formation and dynamical evolution within galaxy haloes with a state-of-the-art semi-analytic simulation of galaxy formation. Our approach allows us to obtain exquisite statistics to study the effect of the environment and assembly history of galaxies, while still allowing a very efficient exploration of the parameter space of star cluster physics. Our reference model, including both efficient cluster disruption during galaxy mergers and a model for the dynamical friction of GCs within the galactic potential, accurately reproduces the observed correlation between the total mass in GCs and the parent halo mass. A deviation from linearity is predicted at low halo masses, which is driven by a strong dependence on morphological type: bulge-dominated galaxies tend to host larger masses of GCs than their later-type counterparts of similar stellar mass. While the significance of the difference might be affected by resolution at the lowest halo masses considered, this is a robust prediction of our model and represents a natural consequence of the assumption that cluster migration from the disk to the halo is triggered by galaxy mergers. Our model requires an environmental dependence of GC radii to reproduce the observed mass distribution of GCs in our Galaxy at the low-mass end. At GC masses $>10^6\,{\rm M}_\odot$, our model predicts fewer GCs than observed due to an overly aggressive treatment of dynamical friction. The metallicity distribution measured for Galactic GCs is well reproduced by our model, even though it predicts systematically younger GCs than observed. We argue that this adds further evidence for an anomalously early formation of the stars in our Galaxy.

astro-ph.GA↗

arXiv2023

Globular cluster metallicity distributions in the E-MOSAICS simulations

The metallicity distributions of globular cluster (GC) systems in galaxies are a critical test of any GC formation scenario. In this work, we investigate the predicted GC metallicity distributions of galaxies in the MOdelling Star cluster population Assembly In Cosmological Simulations within EAGLE (E-MOSAICS) simulation of a representative cosmological volume ($L = 34.4$ comoving Mpc). We find that the predicted GC metallicity distributions and median metallicities from the fiducial E-MOSAICS GC formation model agree well the observed distributions, except for galaxies with masses $M_\ast \sim 2 \times 10^{10}$ M$_\odot$, which contain an overabundance of metal-rich GCs. The predicted fraction of galaxies with bimodal GC metallicity distributions ($37 \pm 2$ per cent in total; $45 \pm 7$ per cent for $M_\ast > 10^{10.5}$ M$_\odot$) is in good agreement with observed fractions ($44^{+10}_{-9}$ per cent), as are the mean metallicities of the metal-poor and metal-rich peaks. We show that, for massive galaxies ($M_\ast > 10^{10}$ M$_\odot$), bimodal GC distributions primarily occur as a result of cluster disruption from initially-unimodal distributions, rather than as a result of cluster formation processes. Based on the distribution of field stars with GC-like abundances in the Milky Way, we suggest that the bimodal GC metallicity distribution of Milky Way GCs also occurred as a result of cluster disruption, rather than formation processes. We conclude that separate formation processes are not required to explain metal-poor and metal-rich GCs, and that GCs can be considered as the surviving analogues of young massive star clusters that are readily observed to form in the local Universe today.

astro-ph.GA↗

arXiv2023

In-situ or accreted? Using deep learning to infer the origin of extragalactic globular clusters from observables

Globular clusters (GCs) are powerful tracers of the galaxy assembly process, and have already been used to obtain a detailed picture of the progenitors of the Milky Way. Using the E-MOSAICS cosmological simulation of a (34.4 Mpc)$^3$ volume that follows the formation and co-evolution of galaxies and their star cluster populations, we develop a method to link the origin of GCs to their observable properties. We capture this complex link using a supervised deep learning algorithm trained on the simulations, and predict the origin of individual GCs (whether they formed in the main progenitor or were accreted from satellites) based solely on extragalactic observables. An artificial neural network classifier trained on $\sim50,000$ GCs hosted by $\sim 700$ simulated galaxies successfully predicts the origin of GCs in the test set with a mean accuracy of $89$ per cent for the objects with [Fe/H]<-0.5 that have unambiguous classifications. The network relies mostly on the alpha-element abundances, metallicities, projected positions, and projected angular momenta of the clusters to predict their origin. A real-world test using the known progenitor associations of the Milky Way GCs achieves up to $90$ per cent accuracy, and successfully identifies as accreted most of the GCs in the inner Galaxy associated to the Kraken progenitor, as well as all the Gaia-Enceladus GCs. We demonstrate that the model is robust to observational uncertainties, and develop a method to predict the classification accuracy across observed galaxies. The classifier can be optimized for available observables (e.g. to improve the accuracy by including GC ages), making it a valuable tool to reconstruct the assembly histories of galaxies in upcoming wide-field surveys.

astro-ph.GA↗

arXiv2022

Monochromatic globular clusters as a critical test of formation models for the dark matter deficient galaxies NGC1052-DF2 and NGC1052-DF4

It was recently proposed that the dark matter-deficient ultra-diffuse galaxies DF2 and DF4 in the NGC1052 group could be the products of a "bullet dwarf" collision between two gas-rich progenitor galaxies. In this model DF2 and DF4 formed at the same time in the immediate aftermath of the collision, and a strong prediction is that their globular clusters should have nearly identical stellar populations. Here we test this prediction by measuring accurate F606W-F814W colors from deep HST/ACS imaging. We find that the clusters are extremely homogeneous. The mean color difference between the globular clusters in DF2 and DF4 is $-0.003\pm 0.005$ mag and the observed scatter for the combined sample of 18 clusters with $M_V<-8.6$ in both galaxies is $0.015 \pm 0.002$ mag. After accounting for observational uncertainties and stochastic cluster-to-cluster variation in the number of red giants, the remaining scatter is $0.008^{+0.005}_{-0.006}$ mag. Both the color difference and the scatter are an order of magnitude smaller than in other dwarf galaxies, and we infer that the bullet scenario passes an important test that could have falsified it. No other formation models have predicted this extreme uniformity of the globular clusters in the two galaxies. We find that the galaxies themselves are slightly redder than the clusters, consistent with a previously-measured metallicity difference. Numerical simulations have shown that such differences are expected in the bullet scenario, as the galaxies continued to self-enrich after the formation of the globular clusters.

astro-ph.GA↗

arXiv2022

A trail of dark matter-free galaxies from a bullet dwarf collision

The ultra-diffuse galaxies DF2 and DF4 in the NGC1052 group share several unusual properties: they both have large sizes, rich populations of overluminous and large globular clusters, and very low velocity dispersions indicating little or no dark matter. It has been suggested that these galaxies were formed in the aftermath of high velocity encounters of gas rich galaxies, events that resemble the collision that created the bullet cluster but on much smaller scales. The gas separates from the dark matter in the collision and subsequent star formation leads to the formation of one or more dark matter-free galaxies. Here we show that the present-day line-of-sight distances and radial velocities of DF2 and DF4 are consistent with their joint formation in the aftermath of a single bullet-dwarf collision, around eight billion years ago. Moreover, we find that DF2 and DF4 are part of an apparent linear substructure of 7-11 large, low-luminosity objects. We propose that these all originated in the same event, forming a trail of dark matter-free galaxies that is more than 2 Mpc long and angled 7 +- 2 degrees from the line of sight. We also tentatively identify the highly dark matter-dominated remnants of the two progenitor galaxies that are expected at the leading edges of the trail.

astro-ph.GA↗

arXiv2022

Constraining the shape of dark matter haloes with globular clusters

We explore how diffuse stellar light and globular clusters (GCs) can be used to trace the matter distribution of their host halo using an observational methodology. For this, we use 117 simulated dark matter (DM) haloes from the $(34.4~\rm cMpc)^3$ periodic volume of the E-MOSAICS project. For each halo, we compare the stellar surface brightness and GC projected number density maps to the surface densities of DM and total mass. We find that the dominant structures identified in the stellar light and in the GCs correspond closely with those from the DM and total mass. Our method is unaffected by the presence of satellites and its precision improves with fainter GC samples. We recover tight relations between the profiles of stellar surface brightness and GC number density to those of the DM, suggesting that the profile of DM can be accurately recovered from the stars and GCs ($σ\leq0.5~$dex). We quantify the projected morphology of DM, stars and GCs, and find that the stars and GCs are more flattened than the DM. Additionally, the semi-major axes of the distribution of stars and GCs are typically misaligned by $\sim 10~$degrees from that of DM. We demonstrate that deep imaging of diffuse stellar light and GCs can place constraints on the shape, profile and orientation of their host halo. These results extend down to haloes with central galaxies $M_{\star}\geq10^{10}~M_{\odot}$, and the analysis will be applicable to future data from the Euclid, Roman and the Rubin observatories.

astro-ph.GA↗

arXiv2022

Examining quasar absorption-line analysis methods: the tension between simulations and observational assumptions key to modelling clouds

A key assumption in quasar absorption line studies of the circumgalactic medium (CGM) is that each absorption component maps to a spatially isolated "cloud" structure that has single valued properties (e.g. density, temperature, metallicity). We aim to assess and quantify the degree of accuracy underlying this assumption. We used adaptive mesh refinement hydrodynamic cosmological simulations of two $z=1$ dwarf galaxies and generated synthetic quasar absorption-line spectra of their CGM. For the SiII $λ1260$ transition, and the CIV $λ\lambda1548, 1550$ and OVI $λ\lambda1031, 1037$ fine-structure doublets, we objectively determined which gas cells along a line-of-sight (LOS) contribute to detected absorption. We implemented a fast, efficient, and objective method to define individual absorption components in each absorption profile. For each absorption component, we quantified the spatial distribution of the absorbing gas. We studied a total of 1,302 absorption systems containing a total of 7,755 absorption components. 48% of SiII, 68% of CIV, and 72% of OVI absorption components arise from two or more spatially isolated "cloud" structures along the LOS. Spatially isolated "cloud" structures were most likely to have cloud-cloud LOS separations of 0.03$R_{vir}$, 0.11$R_{vir}$, and 0.13$R_{vir}$ for SiII, CIV, and OVI, respectively. There can be very little overlap between multi-phase gas structures giving rise to absorption components. If our results reflect the underlying reality of how absorption lines record CGM gas, they place tension on current observational analysis methods as they suggest that component-by-component absorption line formation is more complex than is assumed and applied for chemical-ionisation modelling.

astro-ph.GA↗

arXiv2022

Introducing EMP-Pathfinder: modelling the simultaneous formation and evolution of stellar clusters in their host galaxies

The formation and evolution of stellar clusters is intimately linked to that of their host galaxies. To study this connection, we present the EMP-Pathfinder suite of cosmological zoom-in Milky Way-mass simulations. These simulations contain a sub-grid description for stellar cluster formation and evolution, allowing us to study the simultaneous formation and evolution of stellar clusters alongside their host galaxies across cosmic time. As a key ingredient in these simulations, we include the physics of the multi-phase nature of the interstellar medium (ISM), which enables studies of how the presence of a cold, dense ISM affects cluster formation and evolution. We consider two different star formation prescriptions: a constant star formation efficiency per free-fall time, as well as an environmentally-dependent, turbulence-based prescription. We identify two key results drawn from these simulations. Firstly, we find that tidal shock-driven disruption caused by the graininess of the cold ISM produces old ($τ>10~$Gyr) stellar cluster populations with properties that are in excellent agreement with the observed populations in the Milky Way and M31. Importantly, the addition of the cold ISM addresses the areas of disagreement found in previous simulations that lacked the cold gas phase. Secondly, the formation of stellar clusters is extremely sensitive to the baryonic physics that govern the properties of the cold, dense gas reservoir in the galaxy. This implies that the demographics of stellar cluster populations represent an important diagnostic tool for constraining baryonic physics models in upcoming galaxy formation simulations that also include a description of the cold ISM.

astro-ph.GA↗

arXiv2022

Constraining the formation of NGC1052-DF2 from its unusual globular cluster population

The ultra-diffuse galaxy (UDG) NGC1052-DF2 has a low dark matter content and hosts a very unusual globular cluster (GC) population, with a median luminosity $\sim4$ times higher than in most galaxies and containing about 5~per~cent of the galaxy's stars. We apply a theoretical model that predicts the initial cluster mass function as a function of the galactic environment to investigate the origin of DF2's peculiar GC system. Using the GC mass function, the model constrains the star-forming conditions in the galaxy during the formation of its GCs, $\sim9~\rm{Gyr}$ ago. We predict that the GCs formed in an environment with very high gas surface density, $Σ_{\rm ISM} \gtrsim 10^3 M_{\odot} \rm{pc}^{-2}$, and strong centrifugal support, $Ω\gtrsim 0.7~\rm{Myr}^{-1}$, similar to nearby circum-nuclear starbursts and the central region of the Milky Way. These extreme conditions required to form the observed GC population imply a very high cluster formation efficiency of $\gtrsim 78$ per cent, and contrast strongly with the current diffuse nature of the galaxy. Since a nuclear starburst would lead to the rapid in-spiral of the GCs and is ruled out by the absence of a nuclear star cluster, we propose that the GCs plausibly formed during a major merger at $z\sim1.3$. The merger remnant must have undergone significant expansion of its stellar (and perhaps also its dark matter) component to reach its low present surface brightness, leading to the interesting possibility that it was the formation of DF2's extreme GC population that caused it to become a UDG. If true, this strong structural evolution would have important implications for understanding the origins of UDGs.

astro-ph.GA↗

arXiv2022

The emergence of dark matter-deficient ultra-diffuse galaxies driven by scatter in the stellar mass-halo mass relation and feedback from globular clusters

In addition to their low stellar densities, ultra-diffuse galaxies (UDGs) have a broad variety of dynamical mass-to-light ratios, ranging from dark matter (DM) dominated systems to objects nearly devoid of DM. To investigate the origin of this diversity, we develop a simple, semi-empirical model that predicts the structural evolution of galaxies, driven by feedback from massive star clusters, as a function of their departure from the mean SMHM relation. The model predicts that a galaxy located $\gtrsim 0.5$ dex above the mean relation at $M_{\rm halo}=10^{10}M_{\odot}$ will host a factor of $\sim 10-100$ larger globular cluster (GC) populations, and that feedback from these GCs drives a significant expansion of the stellar component and loss of DM compared to galaxies on the SMHM relation. This effect is stronger in haloes that collapse earlier and have enhanced star formation rates at $z\gtrsim2$, which leads to increased gas pressures, stellar clustering, and mean cluster masses, and significantly enhances the energy loading of galactic winds and its impact on the DM and stellar orbits. The impact on galaxy size and DM content can be large enough to explain observed galaxies that contain nearly the universal baryon fraction, as well as NGC1052-DF2 and DF4 and other isolated UDGs that contain almost no DM. The trend of increasing galaxy size with GC specific frequency observed in galaxy clusters also emerges naturally in the model. Our predictions can be tested with large and deep surveys of the stellar and GC populations in dwarfs and UDGs. Because stellar clustering drives the efficiency of galactic winds, it may be a dominant factor in the structural evolution of galaxies and should be included as an essential ingredient in galaxy formation models.

astro-ph.GA↗

arXiv2021

The physics governing the upper truncation mass of the globular cluster mass function

The mass function of globular cluster (GC) populations is a fundamental observable that encodes the physical conditions under which these massive stellar clusters formed and evolved. The high-mass end of star cluster mass functions are commonly described using a Schechter function, with an exponential truncation mass $M_{c,*}$. For the GC mass functions in the Virgo galaxy cluster, this truncation mass increases with galaxy mass ($M_{*}$). In this paper we fit Schechter mass functions to the GCs in the most massive galaxy group ($M_{\mathrm{200}} = 5.14 \times 10^{13} M_{\odot}$) in the E-MOSAICS simulations. The fiducial cluster formation model in E-MOSAICS reproduces the observed trend of $M_{c,*}$ with $M_{*}$ for the Virgo cluster. We therefore examine the origin of the relation by fitting $M_{c,*}$ as a function of galaxy mass, with and without accounting for mass loss by two-body relaxation, tidal shocks and/or dynamical friction. In the absence of these mass-loss mechanisms, the $M_{c,*}$-$M_{*}$ relation is flat above $M_* > 10^{10} M_{\odot}$. It is therefore the disruption of high-mass GCs in galaxies with $M_{*}\sim 10^{10} M_{\odot}$ that lowers the $M_{c,*}$ in these galaxies. High-mass GCs are able to survive in more massive galaxies, since there are more mergers to facilitate their redistribution to less-dense environments. The $M_{c,*}-M_*$ relation is therefore a consequence of both the formation conditions of massive star clusters and their environmentally-dependent disruption mechanisms.

astro-ph.GA↗

arXiv2021

NGC5846-UDG1: A galaxy formed mostly by star formation in massive, extremely dense clumps of gas

It has been shown that ultra-diffuse galaxies (UDGs) have higher specific frequencies of globular clusters on average than other dwarf galaxies with similar luminosities. The UDG NGC5846-UDG1 is among the most extreme examples of globular cluster-rich galaxies found so far. Here we present new Hubble Space Telescope (HST) observations and analysis of this galaxy and its globular cluster system. We find that NGC5846-UDG1 hosts $54 \pm 9$ globular clusters, three to four times more than any previously known galaxy with a similar luminosity, and higher than reported in previous studies. With a galaxy luminosity of $L_{V,\mathrm{gal}} \approx 6 \times 10^7\,{\rm L}_{\odot}$ ($M_\star \approx 1.2 \times 10^8\,{\rm M}_\odot$) and a total globular cluster luminosity of $L_{V,\mathrm{GCs}} \approx 7.6 \times 10^6\,{\rm L}_{\odot}$, we find that the clusters currently comprise $\sim 13 \%$ of the total light. Taking into account the effects of mass loss from clusters during their formation and throughout their lifetime, we infer that most of the stars in the galaxy likely formed in globular clusters, and very little to no "normal" low-density star formation occurred. This result implies that the most extreme conditions during early galaxy formation promoted star formation in massive and dense clumps, in contrast to the dispersed star formation observed in galaxies today.

astro-ph.GA↗