Galaxy Formation Models Research Articles

First Detection of Molecular Gas in the Giant Low Surface Brightness Galaxy Malin 1

After over three decades of unsuccessful attempts, we report the first detection of molecular gas emission in Malin 1, the largest spiral galaxy observed to date, and one of the most iconic giant low surface brightness galaxies. Using Atacama Large Millimeter/submillimeter Array, we detect significant 12CO (J = 1–0) emission in the galaxy’s central region and tentatively identify CO emission across three regions on the disk. These observations allow for a better estimate of the H2 mass and molecular gas mass surface density, both of which are remarkably low given the galaxy’s scale. By integrating data on its H i mass, we derive a very low molecular-to-atomic gas mass ratio. Overall, our results highlight the minimal presence of molecular gas in Malin 1, contrasting sharply with its extensive, homogeneous atomic gas reservoir. For the first time, we position Malin 1 on the Kennicutt–Schmidt diagram, where it falls below the main sequence for normal spirals, consistent with previous upper limits but now with more accurate figures. These findings are crucial for constraining our understanding of star formation processes in environments characterized by extremely low molecular gas densities and for refining models of galaxy formation, thereby improving predictions concerning the formation, evolution, and distribution of these giant, elusive galaxies.

A trifecta of modelling tools: a Bayesian binary black hole model selection combining population synthesis and galaxy formation models

ABSTRACT Gravitational waves (GWs) have revealed surprising properties of binary black hole (BBH) populations, but there is still mystery surrounding how these compact objects evolve. We apply Bayesian inference and an efficient method to calculate the BBH merger rates in the Shark host galaxies, to determine the combination of COMPAS parameters that outputs a population most like the GW sources from the LIGO, Virgo, and KAGRA (LVK) transient catalogue. For our COMPAS models, we calculate the likelihood with and without the dependence on the predicted number of BBH merger events. We find strong correlations between hyper-parameters governing the specific angular momentum (AM) of mass lost during mass transfer, the mass-loss rates of Wolf–Rayet stars via winds and the chemically homogeneous evolution (CHE) formation channel. We conclude that analysing the marginalized and unmarginalized likelihood is a good indicator of whether the population parameters distribution and number of observed events reflect the LVK data. In doing so, we see that the majority of the models preferred in terms of the population-level parameters of the BBHs greatly overpredict the number of events we should have observed to date. Looking at the smaller number of models that perform well with both likelihoods, we find that those with no CHE, AM loss occurring closer to the donor during the first mass-transfer event, and/or higher rates of mass-loss from Wolf–Rayet winds are generally preferred by current data. We find these conclusions to be robust to our choice of selection criteria.

The Impact of Positive AGN Feedback on the Properties of Galaxies in a Semianalytic Model of Galaxy Formation

We introduce the state-of-the-art semianalytic model Formation and Evolution of GAlaxies (FEGA), which incorporates updated prescriptions for key physical processes in galaxy formation. Notably, FEGA features an unprecedented semianalytic modeling of positive active galactic nucleus (AGN) feedback. The model combines the latest prescriptions for gas infall and cooling, a revised star formation recipe that incorporates the extended Kennicutt–Schmidt relation, disk instability, updated supernova feedback, reincorporation of ejected gas, hot gas stripping from satellite galaxies, and the formation of diffuse light. A novel description of AGN feedback is introduced, describing the positive mode as a burst of star formation from a cooling gas fraction. FEGA is rigorously calibrated using a Markov Chain Monte Carlo procedure to match the evolution of the stellar mass function from high redshift to the present. Subsequently, the model is tested against several observed and predicted scaling relations, including the star formation rate (SFR)–mass, black hole–bulge and stellar mass, stellar-to-halo mass, and red fraction–mass relations. Additionally, we test FEGA against other galaxy properties, such as the distribution of specific SFRs, stellar metallicity, and morphology. Our results demonstrate that the inclusion of positive AGN feedback can coexist with its negative counterpart without drastic alterations to other prescriptions. Importantly, this inclusion improves the ability of the model to describe the primary scaling relations observed in galaxies.

Percolation Statistics in the MillenniumTNG Simulations

The statistical analysis of cosmic large-scale structure is most often based on simple two-point summary statistics, like the power spectrum or the two-point correlation function of a sample of galaxies or other types of tracers. In contrast, topological measures of clustering are also sensitive to higher-order correlations and thus offer the prospect to access additional information that may harbor important constraining power. We here revisit one such geometric measure of the cosmic web in the form of the so-called percolation analysis, using the recent MillenniumTNG simulation suite of the ΛCDM paradigm. We analyze continuum percolation statistics both for high-resolution dark matter particle distributions and for galaxy mock catalogs from a semianalytic galaxy formation model within a periodic simulation volume of 3000 Mpc on a side. For comparison, we also investigate the percolation statistics of random particle sets and neutrino distributions with two different summed particle masses. We find that the percolation statistics of the dark matter distribution evolves strongly with redshift and thus clustering strength, yielding a progressively lower percolation threshold toward later times. However, there is a sizable residual dependence on numerical resolution, which we interpret as a residual influence of different levels of shot noise. This is corroborated by our analysis of galaxy mock catalogs, whose results depend on sampling density more strongly than on galaxy selection criteria. While this limits the discriminative power of percolation statistics, our results suggest that it still remains useful as a complementary cosmological test when controlled for sampling density.

Linking High-z and Low-z: Are We Observing the Progenitors of the Milky Way with JWST?

The recent JWST observation of the Firefly Sparkle at z = 8.3 offers a unique opportunity to link the high- and the low-z Universe. Indeed, the claim of it being a Milky Way (MW) type of assembly at the cosmic dawn opens the possibility of interpreting the observation with locally calibrated galaxy-formation models. Here, we use the a state-of-the-art MW-evolution model to perform forward modeling of our Galaxy's progenitors at high-z. We build a set of mock spectra for the MW building blocks to make predictions for JWST and to interpret the Firefly Sparkle observation. First, we find that the most massive MW progenitor becomes detectable in a deep survey like JADES from z ≈ 8.2, meaning that we could have already observed MW analogs that still need interpretation. Second, we provide predictions for the number of detectable MW progenitors in lensed surveys like the CAnadian NIRISS Unbiased Cluster Survey, and interpret the Firefly Sparkle as a group of MW building blocks. Both the number of detections and the observed NIRCam photometry are consistent with our predictions. By identifying the MW progenitors whose mock photometry best fits the data, we find bursty and extended star formation histories, lasting >150–300 Myr, and estimate their properties: M h ≈ 108−9 M ⊙, M ⋆ ≈ 106.2−7.5 M ⊙, SFR ≈ 0.04–0.20 M ⊙ yr−1, and Z gas ≈ 0.04–0.24 Z ⊙. Uncovering the properties of MW analogs at cosmic dawn by combining JWST observations and locally constrained models will allow us to understand our Galaxy's formation, linking the high- and low-z perspectives.

On the Cosmic Variance of the Merger Rate Density of Binary Neutron Stars

The cosmic variance on the star formation history may lead to bias in the merger rate density estimation of binary neutron star (BNS) mergers by compact binary population synthesis. In this paper, we take advantage of the large box size of the Millennium Simulation combined with the semianalytic galaxy formation model GABE and the parameterized population binary star evolution model to examine how much effect the cosmic variance will introduce on the estimation of the merger rate density of BNS mergers. We find that for subbox sizes of 100 and 200 Mpc, the variance of merger rate density σ R /R at different redshifts is about 23%–35% and 13%–20%, respectively. On the one hand, as for the variance of the detection rate on BNS mergers with the current LIGO–Virgo–KAGRA (LVK) detector network, this value is very small at ≲10%, which indicates ignoring the cosmic variance is reasonable for estimating the merger rate density from current LVK observations. On the other hand, with next-generation gravitational wave detectors, it is possible to localize BNS mergers within subboxes possessing a length of 40 Mpc for a source redshift z s < 0.2. In such a small box, the cosmic variance of the merger rate density is significant, i.e., the value of σ R /R is about ∼55%. This hints that estimating the merger rate density of BNS in different sky areas may provide useful information on the cosmic variance.

Uncovering tidal treasures: automated classification of faint tidal features in DECaLS data

ABSTRACT Tidal features are a key observable prediction of the hierarchical model of galaxy formation and contain a wealth of information about the properties and history of a galaxy. Modern wide-field surveys such as LSST and Euclid will revolutionize the study of tidal features. However, the volume of data will prohibit visual inspection to identify features, thereby motivating a need to develop automated detection methods. This paper presents a visual classification of ∼2000 galaxies from the DECaLS survey into different tidal feature categories: arms, streams, shells, and diffuse. We trained a convolutional neural network (CNN) to reproduce the assigned visual classifications using these labels. Evaluated on a testing set where galaxies with tidal features were outnumbered $\sim 1:10$, our network performed very well and retrieved a median $98.7\pm 0.3$, $99.1\pm 0.5$, $97.0\pm 0.8$, and $99.4^{+0.2}_{-0.6}$ per cent of the actual instances of arm, stream, shell, and diffuse features respectively for just 20 per cent contamination. A modified version that identified galaxies with any feature against those without achieved scores of $0.981^{+0.001}_{-0.003}$, $0.834^{+0.014}_{-0.026}$, $0.974^{+0.008}_{-0.004}$, and $0.900^{+0.073}_{-0.015}$ for the accuracy, precision, recall, and F1 metrics, respectively. We used a gradient-weighted class activation mapping analysis to highlight important regions on images for a given classification to verify the network was classifying the galaxies correctly. This is the first demonstration of using CNNs to classify tidal features into sub-categories, and it will pave the way for the identification of different categories of tidal features in the vast samples of galaxies that forthcoming wide-field surveys will deliver.

Reconciling M/L Ratios Across Cosmic Time: a Concordance IMF for Massive Galaxies

The stellar initial mass function (IMF) is thought to be bottom heavy in the cores of the most massive galaxies, with an excess of low-mass stars compared to the Milky Way. However, studies of the kinematics of quiescent galaxies at 2 < z < 5 find M/L ratios that indicate lighter IMFs. Light IMFs have also been proposed for the unexpected populations of luminous galaxies that JWST has uncovered at z > 7 to reduce tensions with galaxy formation models. Here we explore “ski slope” IMFs that are simultaneously bottom heavy, with a steep slope at low stellar masses, and top heavy, with a shallow slope at high masses. We derive a form of the IMF for massive galaxies that is consistent with measurements in the local universe and yet produces relatively low M/L ratios at high redshift. This concordance IMF has slopes γ 1 = 2.40 ± 0.09, γ 2 = 2.00 ± 0.14, and γ 3 = 1.85 ± 0.11 in the regimes 0.08 M ⊙ – 0.5 M ⊙, 0.5 M⊙ – 1 M ⊙, and >1 M ⊙, respectively. The IMF parameter α, the mass excess compared to a Milky Way IMF, ranges from log(α)≈+0.3 for present-day galaxies to log(α)≈−0.1 for their star-forming progenitors. The concordance IMF applies only to the central regions of the most massive galaxies, with velocity dispersions σ ∼ 300 km s−1, and their progenitors. However, it can be generalized using a previously measured relation between α and σ. We arrive at the following modification to the Kroupa (2001) IMF for galaxies with σ ≳ 160 km s−1: γ1≈1.3+4.3logσ160 , γ2≈2.3−1.2logσ160 , and γ3≈2.3−1.7logσ160 , with σ 160 = σ/160 km s−1. If galaxies grow primarily inside out, so that velocity dispersions are relatively stable, these relations should also hold at high redshift.

Challenges in modeling the dark matter halo of NGC 1052–DF2: Cored versus cuspy halo models

Aims. The discovery of NGC 1052−DF2 and subsequent modeling have shown that NGC 1052−DF2 is deficient in dark matter and is in conflict with the standard stellar-to-halo mass ratio. In this work, we aim to resolve the degeneracy between the dynamical models on the mass estimate of the NGC 1052−DF2. Methods. We constructed mass models of NGC 1052−DF2 using an anisotropic distribution function with a radially varying anisotropy parameter and studied the effect of the various model parameters on the dark matter estimates. We used the observed stellar photometry as an input parameter to construct the distribution function and employed a Markov chain Monte Carlo (MCMC) method to estimate the dark matter model parameters. Results. We find that mass models with a cuspy dark matter halo have comparable χ2 to models with zero dark matter. Moreover, the cuspy dark matter halo fails to consistently account for the observed velocity dispersion in the inner and outer regions of the galaxy. Consequently, we rule out the possibility of a cuspy dark matter halo for describing the mass models of NGC 1052−DF2. Our study shows that the cored dark matter halo model with a total mass of log(MDM/M⊙) = 10.5 explains the observed kinematics but requires an extraordinarily large scale length (20 kpc) and an outer cutoff radius (26 kpc). While the cored mass model provides a comparatively better fit, our findings emphasize that the mass models are largely unconstrained by the available kinematic data. Our results suggest that NGC 1052−DF2 may not only have an ultra-diffuse stellar distribution but that it can, within uncertainties in the available kinematic data, potentially host an ultra-diffuse dark matter distribution compatible with the standard stellar-to-halo mass relation (SHMR) predicted by galaxy formation and evolution models.

5–25 μm Galaxy Number Counts from Deep JWST Data

Galaxy number counts probe the evolution of galaxies over cosmic time and serve as a valuable comparison point to theoretical models of galaxy formation. We present new galaxy number counts in eight photometric bands between 5 and 25 μm from the Systematic Mid-infrared Instrument Legacy Extragalactic Survey and the JWST Advanced Deep Extragalactic Survey deep MIRI parallel, extending to unprecedented depth. By combining our new MIRI counts with existing data from Spitzer and AKARI, we achieve counts across 3–5 orders of magnitude in flux in all MIRI bands. Our counts diverge from predictions from recent semianalytical models of galaxy formation, likely due to their treatment of mid-IR aromatic features. Finally, we integrate our combined JWST−Spitzer counts at 8 and 24 μm to measure the cosmic infrared background (CIB) light at these wavelengths; our measured CIB fluxes are consistent with those from previous mid-IR surveys but larger than predicted by models based on TeV blazar data.

FLORAH: a generative model for halo assembly histories

ABSTRACT The mass assembly history (MAH) of dark matter haloes plays a crucial role in shaping the formation and evolution of galaxies. MAHs are used extensively in semi-analytic and empirical models of galaxy formation, yet current analytic methods to generate them are inaccurate and unable to capture their relationship with the halo internal structure and large-scale environment. This paper introduces florah (FLOw-based Recurrent model for Assembly Histories), a machine-learning framework for generating assembly histories of ensembles of dark matter haloes. We train florah on the assembly histories from the Gadget at Ultra-high Redshift with Extra Fine Time-steps and vsmdplN-body simulations and demonstrate its ability to recover key properties such as the time evolution of mass and concentration. We obtain similar results for the galaxy stellar mass versus halo mass relation and its residuals when we run the Santa Cruz semi-analytic model on florah-generated assembly histories and halo formation histories extracted from an N-body simulation. We further show that florah also reproduces the dependence of clustering on properties other than mass (assembly bias), which is not captured by other analytic methods. By combining multiple networks trained on a suite of simulations with different redshift ranges and mass resolutions, we are able to construct accurate main progenitor branches with a wide dynamic mass range from $z=0$ up to an ultra-high redshift $z \approx 20$, currently far beyond that of a single N-body simulation. florah is the first step towards a machine learning-based framework for planting full merger trees; this will enable the exploration of different galaxy formation scenarios with great computational efficiency at unprecedented accuracy.

Gas-rich "ultra-diffuse" galaxies are consistent with the baryonic Tully-Fisher relation and with Milgromian dynamics

Some gas-rich ``ultra-diffuse'' galaxies (UDGs), which are extreme examples of low surface brightness (LSB) dwarf galaxies, have been reported to lack dark matter and to be offset from the baryonic Tully-Fisher relation (BTFR). If confirmed, these UDGs would represent a serious challenge for both Lambda CDM galaxy-formation models and Milgromian dynamics. Here I demonstrate that these conclusions are very dubious due to underestimated uncertainties on inclinations and/or distances. First, I show that UDGs are offset from the BTFR in the same way as usual face-on LSB dwarfs due to systematic biases at low inclinations. Next, I analyze the two UDGs with the best available rotation-curve data. The first (AGC\,242019) is ideally inclined for kinematic studies; MOND can fit the observed rotation curve with a distance of 12.5pm 0.6 Mpc, which is consistent with Virgocentric flow models. The second UDG (AGC\,114905) is close to face-on, so not ideal for kinematic studies; MOND can fit the observed rotation curve with a distance of 68pm 13 Mpc and inclination of circ circ $, which are consistent with existing data. In particular, I show that the disk inclination is more uncertain than previously estimated due to significant asymmetries (lopsidedness) in the stellar distribution. In conclusion, there is no strong evidence that gas-rich UDGs and gas-rich LSB dwarfs are distinct galaxy populations with different dynamical properties; instead, UDGs seem to be a subset of LSB dwarf galaxies biased toward face-on systems.

Debiasing with Diffusion: Probabilistic Reconstruction of Dark Matter Fields from Galaxies with CAMELS

Galaxies are biased tracers of the underlying cosmic web, which is dominated by dark matter (DM) components that cannot be directly observed. Galaxy formation simulations can be used to study the relationship between DM density fields and galaxy distributions. However, this relationship can be sensitive to assumptions in cosmology and astrophysical processes embedded in galaxy formation models, which remain uncertain in many aspects. In this work, we develop a diffusion generative model to reconstruct DM fields from galaxies. The diffusion model is trained on the CAMELS simulation suite that contains thousands of state-of-the-art galaxy formation simulations with varying cosmological parameters and subgrid astrophysics. We demonstrate that the diffusion model can predict the unbiased posterior distribution of the underlying DM fields from the given stellar density fields while being able to marginalize over uncertainties in cosmological and astrophysical models. Interestingly, the model generalizes to simulation volumes ≈500 times larger than those it was trained on and across different galaxy formation models. The code for reproducing these results can be found at https://github.com/victoriaono/variational-diffusion-cdm ✎.

Spectroscopic confirmation of two luminous galaxies at a redshift of 14

The first observations of the James Webb Space Telescope (JWST) have revolutionized our understanding of the Universe by identifying galaxies at redshift z ≈ 13 (refs. 1–3). In addition, the discovery of many luminous galaxies at Cosmic Dawn (z > 10) has suggested that galaxies developed rapidly, in apparent tension with many standard models4–8. However, most of these galaxies lack spectroscopic confirmation, so their distances and properties are uncertain. Here we present JWST Advanced Deep Extragalactic Survey–Near-Infrared Spectrograph spectroscopic confirmation of two luminous galaxies at z=14.32−0.20+0.08\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$z={14.32}_{-0.20}^{+0.08}$$\\end{document} and z = 13.90 ± 0.17. The spectra reveal ultraviolet continua with prominent Lyman-α breaks but no detected emission lines. This discovery proves that luminous galaxies were already in place 300 million years after the Big Bang and are more common than what was expected before JWST. The most distant of the two galaxies is unexpectedly luminous and is spatially resolved with a radius of 260 parsecs. Considering also the very steep ultraviolet slope of the second galaxy, we conclude that both are dominated by stellar continuum emission, showing that the excess of luminous galaxies in the early Universe cannot be entirely explained by accretion onto black holes. Galaxy formation models will need to address the existence of such large and luminous galaxies so early in cosmic history.

The COS-Holes Survey: Connecting Galaxy Black Hole Mass with the State of the CGM

We present an analysis of Hubble Space Telescope COS/G160M observations of C IV in the inner circumgalactic medium (CGM) of a novel sample of eight z ∼ 0, L ≈ L ⋆ galaxies, paired with UV-bright QSOs at impact parameters (R proj) between 25 and 130 kpc. The galaxies in this stellar-mass-controlled sample (log10 M ⋆/M ⊙ ∼ 10.2–10.9 M ⊙) host supermassive black holes (SMBHs) with dynamically measured masses spanning log10 M BH/M ⊙ ∼ 6.8–8.4; this allows us to compare our results with models of galaxy formation where the integrated feedback history from the SMBH alters the CGM over long timescales. We find that the C IV column density measurements (N C IV; average log10 N C IV,CH = 13.94 ± 0.09 cm−2) are largely consistent with existing measurements from other surveys of N C IV in the CGM (average log10 N C IV,Lit = 13.90 ± 0.08 cm−2), but do not show obvious variation as a function of the SMBH mass. By contrast, specific star formation rate (sSFR) is highly correlated with the ionized content of the CGM. We find a large spread in sSFR for galaxies with log10 M BH/M ⊙ > 7.0, where the CGM C IV content shows a clear dependence on galaxy sSFR but not M BH. Our results do not indicate an obvious causal link between CGM C IV and the mass of the galaxy’s SMBH; however, through comparisons to the EAGLE, Romulus25, and IllustrisTNG simulations, we find that our sample is likely too small to constrain such causality.

A warm dark matter cosmogony may yield more low-mass galaxy detections in 21-cm surveys than a cold dark matter one

ABSTRACT The 21-cm spectral line widths, $w_{50}$, of galaxies are an approximate tracer of their dynamical masses, such that the dark matter halo mass function is imprinted in the number density of galaxies as a function of $w_{50}$. Correcting observed number counts for survey incompleteness at the level of accuracy needed to place competitive constraints on warm dark matter (WDM) cosmological models is very challenging, but forward-modelling the results of cosmological hydrodynamical galaxy formation simulations into observational data space is more straightforward. We take this approach to make predictions for an ALFALFA-like survey from simulations using the EAGLE galaxy formation model in both cold (CDM) and WDM cosmogonies. We find that for WDM cosmogonies more galaxies are detected at the low-$w_{50}$ end of the 21-cm velocity width function than in the CDM cosmogony, contrary to what might naïvely be expected from the suppression of power on small scales in such models. This is because low-mass galaxies form later and retain more gas in WDM cosmogonies (with EAGLE). While some shortcomings in the treatment of cold gas in the EAGLE model preclude placing definitive constraints on WDM scenarios, our analysis illustrates that near-future simulations with more accurate modelling of cold gas will likely make strong constraints possible, especially in conjunction with new 21-cm surveys such as WALLABY.

A two-phase model of galaxy formation – II. The size–mass relation of dynamically hot galaxies

ABSTRACT In Paper-I, we developed a two-phase model to connect dynamically hot galaxies (such as ellipticals and bulges) with the formation of self-gravitating gas clouds (SGCs) associated with the fast assembly of dark matter haloes. Here, we explore the implications of the model for the size–stellar mass relation of dynamically hot galaxies. Star-forming sub-clouds resulting from the fragmentation of the turbulent SGC inherit its spatial structure and dynamical hotness, producing a ‘homologous’ relation, $r_{\rm f}\approx \, 100\, r_{\rm bulge}$, between the size of a dynamically hot galaxy ($r_{\rm bulge}$) and that of its host halo assembled in the fast regime ($r_{\rm f}$), independent of redshift and halo mass. This relation is preserved by the ‘dry’ expansion driven by dynamical heating when a galaxy becomes gas-poor due to inefficient cooling, and is frozen due to the stop of bulge growth during the slow assembly regime of the halo. The size–stellar mass relation is thus a simple combination of the galaxy–halo homology and the non-linear stellar mass–halo mass relation. Using a set of halo assembly histories, we reproduce all properties in the observed size–mass relation of dynamically hot galaxies, including the flattening in the low-mass end and the upturn in the massive end. The prediction matches observational data currently available to $z \approx 4$, and can be tested in the future at higher z. Our results indicate that the sizes of dynamically hot galaxies are produced by the dissipation and collapse of gas in haloes to establish SGCs in which stars form.

A two-phase model of galaxy formation: I. The growth of galaxies and supermassive black holes

ABSTRACT We develop a model for galaxy formation and the growth of supermassive black holes (SMBHs), based on the fact that cold dark matter haloes form their gravitational potential wells through a fast phase with rapid change in the potential, and that the high universal baryon fraction makes cooled gas in haloes self-gravitating and turbulent before it can form rotation-supported discs. Gas fragmentation produces subclouds so dense that cloud–cloud collision and drag on clouds are not significant, producing a dynamically hot system of subclouds that form stars and move ballistically to feed the central SMBH. Active galactic nucleus (AGN) and supernova feedback is effective only in the fast phase, and the cumulative effects are to regulate star formation and SMBH growth, as well as to reduce the amount of cold gas in haloes to allow the formation of globally stable discs. Using a set of halo assembly histories, we demonstrate that the model can reproduce a number of observations, including correlations among SMBH mass, stellar mass of galaxies and halo mass, the number densities of galaxies and SMBH, as well as their evolution over the cosmic time.

Estimation of line-of-sight velocities of individual galaxies using neural networks – I. Modelling redshift–space distortions at large scales

ABSTRACT We present a scheme based on artificial neural networks (ANNs) to estimate the line-of-sight velocities of individual galaxies from an observed redshift–space galaxy distribution. We find an estimate of the peculiar velocity at a galaxy based on galaxy counts and barycentres in shells around it. By training the network with environmental characteristics, such as the total mass and mass centre within each shell surrounding every galaxy in redshift space, our ANN model can accurately predict the line-of-sight velocity of each individual galaxy. When this velocity is used to eliminate the RSD effect, the two-point correlation function (TPCF) in real space can be recovered with an accuracy better than 1 per cent at s &amp;gt; 8 $\, h^{-1}\, \mathrm{Mpc}$, and 4 per cent on all scales compared to ground truth. The real-space power spectrum can be recovered within 3 per cent on k&amp;lt; 0.5 $\, \mathrm{Mpc}^{-1}\, h$, and less than 5 per cent for all k modes. The quadrupole moment of the TPCF or power spectrum is almost zero down to s = 10 $\, h^{-1}\, \mathrm{Mpc}$ or all k modes, indicating an effective correction of the spatial anisotropy caused by the RSD effect. We demonstrate that on large scales, without additional training with new data, our network is adaptable to different galaxy formation models, different cosmological models, and mock galaxy samples at high-redshifts and high biases, achieving less than 10 per cent error for scales greater than 15 $\, h^{-1}\, \mathrm{Mpc}$. As it is sensitive to large-scale densities, it does not manage to remove Fingers of God in large clusters, but works remarkably well at recovering real-space galaxy positions elsewhere. Our scheme provides a novel way to predict the peculiar velocity of individual galaxies, to eliminate the RSD effect directly in future large galaxy surveys, and to reconstruct the three-dimensional cosmic velocity field accurately.

LtU-ILI: An All-in-One Framework for Implicit Inference in Astrophysics and Cosmology

This paper presents the Learning the Universe Implicit Likelihood Inference (LtU-ILI) pipeline, a codebase for rapid, user-friendly, and cutting-edge machine learning (ML) inference in astrophysics and cosmology. The pipeline includes software for implementing various neural architectures, training schema, priors, and density estimators in a manner easily adaptable to any research workflow. It includes comprehensive validation metrics to assess posterior estimate coverage, enhancing the reliability of inferred results. Additionally, the pipeline is easily parallelizable, designed for efficient exploration of modeling hyperparameters. To demonstrate its capabilities, we present real applications across a range of astrophysics and cosmology problems, such as: estimating galaxy cluster masses from X-ray photometry; inferring cosmology from matter power spectra and halo point clouds; characterising progenitors in gravitational wave signals; capturing physical dust parameters from galaxy colors and luminosities; and establishing properties of semi-analytic models of galaxy formation. We also include exhaustive benchmarking and comparisons of all implemented methods as well as discussions about the challenges and pitfalls of ML inference in astronomical sciences. All code and examples are made publicly available at https://github.com/maho3/ltu-ili.