arXiv subjects

G. Martin

Publications and source records attributed to G. Martin.

At least 19 recordsLinked to original sources

arXiv2026

Intracluster light and dark matter halo shapes reflect common assembly, not mutual coupling: shape correspondence in CDM but not SIDM

The intracluster light (ICL) has been proposed as a luminous tracer of the dark matter (DM) halo shape in galaxy clusters, with recent observational studies and $\Lambda$CDM simulations suggesting close alignment between the two components. We test whether this correspondence holds in a self-interacting dark matter (SIDM) cosmology using matched CDM and SIDM ($\sigma_\mathrm{DM}/m = 0.1$ and $0.5\;\mathrm{cm^{2}\,g^{-1}}$) hydrodynamical cluster simulations from TheThreeHundred project. We measure 3D axis ratios and major-axis orientations as a function of radius for both the DM halo and the stellar (BCG+ICL) distribution, and quantify the spatial correspondence using the Weighted Overlap Coefficient. In the SIDM simulations, DM haloes become significantly rounder in their inner regions with increasing DM cross-section, while the stellar distribution retains the same elongated morphology in both CDM and SIDM. This decoupling is robust across two baryonic physics models and two resolution levels, and is detected at $>5\sigma$ for $\sigma_\mathrm{DM}/m = 0.5\;\mathrm{cm^{2}\,g^{-1}}$. It arises because the stellar shape is governed by the accretion geometry of tidally stripped satellites along the large-scale structure, which is identical in CDM and SIDM, whereas, in SIDM, halo shapes respond additionally to local self-interactions. The ICL-DM shape correspondence demonstrated in CDM therefore breaks down in an SIDM universe.

astro-ph.CO↗

arXiv2026

Isophote shape analysis and the unfortunate subtlety of dwarf galaxy structure

Dwarf galaxies ($M_{*}/M_{\odot} \lesssim 10^{9.5}$), being sensitive to key evolutionary drivers like baryonic feedback and tidal perturbation, are crucial for understanding galaxy evolution as a whole. Their abundance and faintness, however, ensures that most will be studied primarily via broadband imaging for the foreseeable future. It is thus crucial to identify the most informative broadband-derivable quantities in the dwarf regime. As studies of widely used morphological parameters like concentration, asymmetry, and smoothness suggest these lack discriminatory power among dwarfs, we assess alternatives derived from isophotes: position angle twists, ellipticity, deviations from pure ellipses, and residuals to single-S\'{e}rsic profile fits. Using these parameters, we compare dwarf populations with massive galaxies of the same morphological class, and among themselves by morphological class. Only dwarf spirals may differ from their massive counterparts, being structurally simpler; dwarf and massive early type galaxy (ETG) isophotal similarity suggests all dwarf ETGs may be triaxial. Among only dwarfs, morphological classes are indistinguishable in this parameter space. A principal component analysis (PCA) using all available morphological, isophotal, and physical parameters expands on this: no PC explains more than $\sim$26% of the population variance, and no clear multimodality appears in any pairwise PC projection. We find similarly moderate spectral clustering, with a silhouette score of only 0.35. Given this self-similarity, parsing dwarf galaxy evolution from photometric parameters alone will likely require detailed statistical analysis of large dwarf populations in a high-dimensional parameter space, a task suitable for up-coming large-scale surveys like the Legacy Survey of Space and Time.

astro-ph.GA↗

arXiv2026

The major merger-active galactic nucleus connection up to the cosmic noon

Galaxy major mergers are a potential mechanism for triggering active galactic nuclei (AGN) activity, but their role remains debated, particularly beyond the local Universe. We aim to shed light on the merger-AGN connection at $z=0.5$-$2$, exploiting the multi-wavelength datasets and {\it James Webb Space Telescope} (JWST) observations in the COSMOS field. We construct a stellar mass-limited sample and identify AGN via mid-infrared (MIR) colours, X-ray detections, and spectral energy distribution (SED) fitting. We train convolutional neural networks to identify mergers with mock JWST observations. We create non-AGN and non-merger control samples matching the redshift, stellar mass, and star-formation rate distributions of the AGN and mergers. We find AGN to be moderately more frequent in mergers than in non-mergers, with excess ratios ranging from $\sim2.5$ (X-ray AGN) to $\sim1.3$ (MIR) and $\sim 1.1$-1.2 (SED AGN). Similarly, AGN galaxies show a higher merger fraction ($f_{merg}$) than non-AGN controls. We then study $f_{merg}$ as a function of relative and absolute AGN power, utilising the AGN fraction ($f_{AGN}$) and accretion disc luminosity (L$_{disc}$) parameters. We uncover a $f_{merg}$-$f_{AGN}$ relation with two regimes: $f_{merg}$ stays roughly flat for less-dominant AGN ($f_{AGN}<0.8$) but increases at $f_{AGN}>0.8$ for the MIR and X-ray AGN, and more gently for SED AGN, where mergers appear to be the main triggering mechanism. Additionally, $f_{merg}$ increases monotonically as a function of L$_{disc}$, for all AGN types, reaching $f_{merg}>50\%$ for the most luminous AGN (L$_{disc} \gtrsim 10^{46}\,{erg\,s^{-1}}$). Overall, our results suggest that major mergers can trigger AGN out to the cosmic noon at $z\sim2$. Furthermore, the role of major mergers shows a clear dependence on AGN luminosity and remains the principal mechanism for fuelling the most powerful AGN.

astro-ph.GA↗

arXiv2026

AGN in massive galaxies identified via optical broadband variability: lessons from VST-COSMOS for future LSST science

We study the properties of 56 massive (M$_{\rm{\star}}$ > 10$^{10}$ M$_{\odot}$) galaxies at $z<1$ that host AGN, detected via their broadband optical variability in the VST-COSMOS survey. VST-COSMOS provides a nearly-identical single visit depth ($r$ $\sim$ 24.6 mag) and temporal baseline (eleven years) as the forthcoming Legacy Survey of Space and Time (LSST), albeit in a much smaller 1 deg$^2$ footprint (four orders of magnitude smaller than that of the LSST). We compare the properties (morphologies, the presence of interactions, rest-frame colours and environment) of our AGN to galaxies in a control sample, which are drawn from the non-variable population and matched in redshift and stellar mass to their AGN counterparts. The fraction of AGN with early-type morphology ($\sim$55 per cent) and the fraction that is interacting ($\sim$23 per cent) are similar to what is observed in the controls, suggesting that these AGN are not primarily triggered by interactions. Similarly, the AGN and controls do not show strong differences in their rest-frame $(u-z)$ colours or local environment, suggesting that neither the recent star formation histories nor the surroundings of the AGN are strongly atypical of the general galaxy population. This study provides a glimpse into forthcoming AGN science using the LSST. With vastly improved statistics, LSST will offer unprecedented insights into AGN demographics, host-galaxy evolution and the processes that fuel supermassive black holes, potentially reshaping our understanding of their place in the Universe.

astro-ph.GA↗

arXiv2026

Multi-wavelength morphology and dust emission in low-redshift dwarf galaxies in COSMOS-Web with HST and JWST

Low-mass or dwarf galaxies (M$_{\ast}<10^{9}$ M${\odot}$) are abundant in the Universe, yet their formation and evolution remain poorly understood. Their enhanced sensitivity to feedback from star formation and active galactic nuclei (AGN) make them excellent laboratories to test whether feedback prescriptions in cosmological simulations accurately reproduce their interstellar medium (ISM) properties. We present JWST/NIRCam and MIRI imaging of nine dwarf galaxies from COSMOS-Web survey at redshift $z<0.08$, with star formation rates ranging from 0.003-0.3 M${\odot}$ yr$^{-1}$ and stellar masses of log M$_{\ast}\sim8-9$ M$_{\odot}$. The detection rate with both NIRCam and MIRI is 100\%, indicating that these dwarfs possess substantial ISM content. The detected sample includes a roughly equal mix of early-type and late-type dwarfs, suggesting that it is representative of the broader dwarf galaxy population in low-density environments. We find that the observed MIRI flux distributions are comparable to forward-modelled flux distributions of mass-matched simulated galaxies in TNG50. We further conduct a multi-wavelength morphological analysis complementing the JWST NIRCam and MIRI imaging with archival HST/ACS data, employing the CAS (concentration, asymmetry, smoothness) framework. Among the multi-wavelength images, MIRI exhibits the largest variation in CAS parameters, likely due to dust lanes and clumps in several galaxies, also suggested by Spectral Energy Distribution (SED) fitting. This suggests that the dust content in these systems may be higher than those implied by rest-frame optical or near-infrared observations alone. Upcoming UV/optical and mid-infrared spectroscopic follow-up will be critical for constraining the gas kinematics and dust grain properties of dwarf galaxies in low-density environments such as COSMOS.

astro-ph.GA↗

arXiv2025

Variability-selected AGN in dwarf galaxies: the incidence of AGN in dwarf and massive galaxies is similar

We use the VST-COSMOS survey to identify, via their optical broadband variability, 30 AGN in nearby (z<0.4) dwarf (10^8 MSun < M < 10^10 MSun) galaxies. VST-COSMOS offers a 1 deg^2 survey footprint, a single visit depth of 24.6 mag and 68 r-band visits spanning an eleven-year temporal baseline. Compared to a control sample matched in stellar mass and redshift, the dwarf AGN population shows an elevated fraction of early-type galaxies but a similar fraction of interacting objects, suggesting that interactions do not play a significant role in triggering these AGN. Dwarf AGN hosts do not show strong differences in their distances to nodes, filaments and massive galaxies compared to the controls, which indicates that AGN triggering, at least in this sample, is not strongly correlated with environment. Finally, by combining the true number of galaxies, the detectability of AGN and the measured numbers of variable sources in dwarf and massive (M > 10^10 MSun) galaxies, we estimate the relative frequency of AGN in these two mass regimes. Our results suggest that the incidence of AGN in dwarfs and massive galaxies is similar (within less than a factor of 2 of each other), with some evidence that the AGN fraction increases with stellar mass in the dwarf population.

astro-ph.GA↗

arXiv2025

The origin of the intra-cluster light in The Three Hundred simulations

We investigate the origin and formation mechanisms of the intra-cluster light (ICL) in The Three Hundred simulations, a set of 324 hydrodynamically resimulated massive galaxy clusters. The ICL, a diffuse component of stars not bound to individual galaxies, serves as a key tracer of cluster formation and evolution. Using two hydrodynamics implementations, Gadget-X and Gizmo-Simba, we identify the stellar particles comprising the ICL at z=0 and trace them back in time to the moments when they were formed and accreted into the ICL. Across our 324 clusters, half of the present-day ICL mass is typically in place between $z \sim $ 0.2 and 0.5. The main ICL formation channel is the stripping of stars from subhalos after their infall into the host cluster. Within this channel, 65-80 per cent of the ICL comes from objects with stellar (infall) masses above $10^{11}$ M$_\odot$, i.e., massive galaxies, groups and clusters. Considering the ratio of the infalling halo to the total cluster mass, a median of 35 per cent of the mass is brought in major merger events, though this percentage varies significantly across clusters (15-55 per cent). Additional contributions come from minor mergers (25-35 per cent) and smooth accretion (20-50 per cent). The primary contributors typically infall at $z \leq 1$, with smaller fractions arriving at redshifts between 1 and 2. Regarding other formation channels, we find minor contributions from stars formed in subhalos after their infall and stars stripped while their contributing halo remains outside the host cluster (and can eventually fall inside or stay outside). Finally, for our two sets of simulations, we find medians of 12 (Gadget-X) and 2 (Gizmo-Simba) per cent of the ICL mass formed in-situ, that is, directly as part of the diffuse component. However, this component can be attributed to stripping of gas in high-velocity infalling satellite galaxies.

astro-ph.CO↗

arXiv2025

Nearby dwarf galaxies with extreme star formation rates: a window into dwarf-galaxy evolution in the early Universe

We study a sample of nearby (z~0.2) low-luminosity dwarf (10^7 MSun < M* < 10^8 MSun) galaxies which have extreme (0.1 - 3 MSun/yr) star formation rates (SFRs) for this mass regime, making them plausible analogues of dwarfs at z~5.5. We compare the properties of these analogues to control samples of 'normal' dwarfs, which reside on the star formation main sequence (SFMS) at z~0.2 and are matched in their stellar mass and redshift distributions to the analogue population. The analogue and normal populations do not show differences, either in their half-light radii or the projected distances to nodes, filaments and massive galaxies. This suggests that the comparatively extreme SFRs in the analogues are not driven by them being anomalously compact or because they reside in specific environments which might provide a larger gas supply. However, the fractions of interacting galaxies and those that have early-type morphology are significantly elevated (by factors of ~5.6 and ~9 respectively) in the analogues compared to the normal population. Extrapolation of the redshift evolution of the star formation main sequence into our mass range of interest appears to underestimate the SFRs of observed dwarfs at z~5.5. Since current SFMS measurements remain dominated by low and intermediate redshift data (especially at low stellar masses), our study suggests that this underestimation may be driven by interactions (which are more frequent at earlier epochs) boosting the SFRs in the high-redshift dwarf population. Our results are consistent with a picture where higher gas availability, augmented by interactions, drives much of the stellar mass assembly of dwarf galaxies in the early Universe.

astro-ph.GA↗

arXiv2025

Cosmic reflections I: the structural diversity of simulated and observed low-mass galaxy analogues

Dwarf galaxies serve as powerful laboratories for investigating the underlying physics of galaxy evolution including the impact of baryonic feedback processes and environmental influences. We compare the visual and structural properties of dwarf galaxies in ultra-deep HSC-SSP imaging of the COSMOS field with those measured from realistic HSC-like synthetic observations of dwarfs generated by the Illustris TNG50 and NewHorizon simulations. Using S\'ersic profile fitting and non-parametric morphological metrics (Gini, $M_{20}$, asymmetry, and concentration), we evaluate the diversity of structural properties in observed and simulated galaxies. Our analysis shows that NewHorizon and TNG50 galaxies lie at opposite extremes of observed structural trends: NewHorizon produces diffuse, extended galaxies with shallow S\'ersic indices, while TNG50 yields compact, concentrated systems with steep indices. Both simulations reproduce observed structural trends more closely at higher stellar masses ($M_{\star}\sim10^{9.5} {\rm M_{\odot}}$) but fail to capture the full diversity of COSMOS dwarfs at lower masses. Non-parametric metrics further show that NewHorizon galaxies exhibit more uneven, clumpy light distributions while TNG50 galaxies have smoother but excessively concentrated profiles. These structural differences reflect underlying differences in their physical prescriptions and are likely driven by differing approaches to ISM physics, supernova feedback and star formation in addition to differences in numerical resolution. Our findings highlight the unique power of low-mass galaxies to constrain differences in simulation physics, especially star formation and feedback. Upcoming surveys from facilities like the Vera C. Rubin Observatory and Euclid will enable more rigorous comparisons with simulations, offering deeper insights into the physical processes shaping galaxy evolution.

astro-ph.GA↗

arXiv2025

Intracluster light is a biased tracer of the dark matter distribution in clusters

The diffuse stellar component of galaxy clusters known as intracluster light (ICL) has been proposed as an observable tracer of the cluster's dark matter (DM) halo. Assessing its reliability as a DM tracer requires understanding how the intracluster stars are energetically linked to the underlying DM distribution, which we investigate at $z\approx0$ in 12 galaxy clusters with $M_{178} = 1.18 - 3.71 \times 10^{14}\,\textrm{M}_\odot$ from the Horizon-AGN simulation. We quantify the orbital energies of these components by their mean specific energies ${\langle \varepsilon \rangle}$, and find that this quantity is $\approx$ 25 per cent lower for the intracluster stars than the DM, whilst the energetics of the satellite galaxies (a standard DM tracer) are only marginally ($\approx$ 5 per cent) higher than the DM. Importantly, the lower ${\langle \varepsilon \rangle}$ of the intracluster stars compared to the DM is robust against the precise separation between the brightest cluster galaxy (BCG) and the ICL. The specific energy distribution of ICL stars is concentrated towards lower energies and poorly samples the higher energies, where much of the DM resides. Consequently, the intracluster stars have velocity distributions with lower typical speeds and a more centrally-concentrated density profile than the DM. We also find that intracluster stars have more radially-biased orbits than the DM, indicating these components have distinct orbital distributions. This study demonstrates that although the morphology of the ICL may match the DM halo, the ICL is a biased tracer of DM, and these biases must be understood in order to infer properties of the DM from the ICL.

astro-ph.GA↗

arXiv2025

The quenching of star formation in dwarf galaxies: new perspectives from deep-wide surveys

Dwarf galaxies dominate the galaxy number density, making them critical to our understanding of galaxy evolution. However, typical dwarfs are too faint to be visible outside the very local Universe in past surveys like the SDSS, which offer large footprints but are shallow. Dwarfs in such surveys have relatively high star formation rates, which boost their luminosity, making them detectable in shallow surveys, but also biased and potentially unrepresentative of dwarfs as a whole. Here, we use deep data to perform an unbiased statistical study of ~7,000 nearby (z<0.25) dwarfs (10^8 MSun < M < 10^9.5 MSun) in the COSMOS field which, at these redshifts, is a relatively low-density field. At z~0.05, ~40 per cent of dwarfs in low-density environments are red/quenched, falling to ~30 per cent by z~0.25. Red dwarfs reside closer to nodes, filaments and massive galaxies. Proximity to a massive galaxy appears to be more important in determining whether a dwarf is red, rather than simply its distance from nodes and filaments or the mean density of its local environment. Interestingly, around half of the red dwarfs reside outside the virial radii of massive galaxies and around a third of those also inhabit regions in the lower 50 per cent in density percentile (i.e. regions of very low ambient density). Around half of the red dwarf population is, therefore, quenched by mechanisms unrelated to environment, which are likely to be internal processes such as stellar and AGN feedback.

astro-ph.GA↗

arXiv2025

Morphological Classification of Galaxies Through Structural and Star Formation Parameters Using Machine Learning

We employ the XGBoost machine learning (ML) method for the morphological classification of galaxies into two (early-type, late-type) and five (E, S0--S0a, Sa--Sb, Sbc--Scd, Sd--Irr) classes, using a combination of non-parametric ($C,\,A,\,S,\,A_S,\,\mathrm{Gini},\,M_{20},\,c_{5090}$), parametric (S\'ersic index, $n$), geometric (axial ratio, $BA$), global colour ($g-i,\,u-r,\,u-i$), colour gradient ($\Delta (g - i)$), and asymmetry gradient ($\Delta A_{9050}$) information, all estimated for a local galaxy sample ($z<0.15$) compiled from the Sloan Digital Sky Survey (SDSS) imaging data. We train the XGBoost model and evaluate its performance through multiple standard metrics. Our findings reveal better performance when utilizing all fourteen parameters, achieving accuracies of 88\% and 65\% for the two-class and five-class classification tasks, respectively. In addition, we investigate a hierarchical classification approach for the five-class scenario, combining three XGBoost classifiers. We observe comparable performance to the ``direct'' five-class classification, with discrepancies of only up to 3\%. Using SHAP (an advanced interpretation tool), we analyse how galaxy parameters impact the model's classifications, providing valuable insights into the influence of these features on classification outcomes. Finally, we compare our results with previous studies and find them consistently aligned.

astro-ph.GA↗

arXiv2024

New tools for studying planarity in galaxy satellite systems: Milky Way satellite planes are consistent with {\Lambda}CDM

We introduce a new concept -- termed "planarity" -- which aims to quantify planar structure in galaxy satellite systems without recourse to the number or thickness of planes. We use positions and velocities from the Gaia EDR3 to measure planarity in Milky Way (MW) satellites and the extent to which planes within the MW system are kinematically supported. We show that the position vectors of the MW satellites exhibit strong planarity but the velocity vectors do not, and that kinematic coherence cannot, therefore, be confirmed from current observational data. We then apply our methodology to NewHorizon, a high-resolution cosmological simulation, to compare satellite planarity in MW-like galaxies in a {\Lambda}CDM-based model to that in the MW satellite data. We demonstrate that kinematically supported planes are common in the simulation and that the observed planarity of MW satellites is not in tension with the standard {\Lambda}CDM paradigm.

astro-ph.GA↗

arXiv2024

Stellar stripping efficiencies of satellites in numerical simulations: the effect of resolution, satellite properties and numerical disruption

The stellar stripping of satellites in cluster haloes is understood to play an important role in the production of intracluster light. Increasingly, cosmological simulations have been utilised to investigate its origin and assembly. However, such simulations typically model individual galaxies at relatively coarse resolutions, raising concerns about their accuracy. Although there is a growing literature on the importance of numerical resolution for the accurate recovery of the mass loss rates of dark matter (DM) haloes, there has been no comparable investigation into the numerical resolution required to accurately recover stellar mass loss rates in galaxy clusters. Using N-body simulations of satellite galaxies orbiting in a cluster halo represented by a static external potential, we conduct a set of convergence tests in order to explore the role of numerical resolution and force softening length on stellar stripping efficiency. We consider a number of orbital configurations, satellite masses and satellite morphologies. We find that stellar mass resolution is of minor importance relative to DM resolution. Resolving the central regions of satellite DM halos is critical to accurately recover stellar mass loss rates. Poorly resolved DM haloes develop cored inner profiles and, if this core is of comparable size to the stellar component of the satellite galaxy, this leads to significant over-stripping. To prevent this, relatively high DM mass resolutions of around $m_{\rm DM}\sim10^{6}$ M$_{\odot}$, better than those achieved by many contemporary cosmological simulations, are necessary.

astro-ph.GA↗

arXiv2024

Dust and Power: Unravelling the merger - active galactic nucleus connection in the second half of cosmic history

Galaxy mergers represent a fundamental physical process under hierarchical structure formation, but their role in triggering AGNs is still unclear. We aim to investigate the merger-AGN connection using state-of-the-art observations and novel methods in detecting mergers and AGNs. We selected stellar mass-limited samples at redshift z<1 from KiDS, focusing on the KiDS-N-W2 field with a wide range of multi-wavelength data. Three AGN types, selected in the MIR, X-ray, and via SED modelling, were analysed. To identify mergers, we used convolutional neural networks trained on two cosmological simulations. We created mass and redshift-matched control samples of non-mergers and non-AGNs. We observe a clear AGN excess (a factor of 2-3) in mergers with respect to non-mergers for the MIR AGNs, and a mild excess for the X-ray and SED AGNs, indicating that mergers could trigger all 3 types but are more connected with the MIR AGNs. About half of the MIR AGNs are in mergers but it is unclear whether mergers are the main trigger. For the X-ray and SED AGNs, mergers are unlikely to be the dominant trigger. We also explore the relation using the continuous AGN fraction $f_{AGN}$ parameter. Mergers exhibit a clear excess of high $f_{AGN}$ values relative to non-mergers, for all AGNs. We unveil the first merger fraction $f_{merg}-f_{AGN}$ relation with two distinct regimes. When the AGN is not dominant, the relation is only mildly increasing or even flat, with the MIR AGNs showing the highest $f_{merg}$. In the regime of very dominant AGNs ($f_{AGN}\geq0.8$), $f_{merg}$ shows a steeply rising trend with increasing $f_{AGN}$ for all AGN types. These trends are also seen when plotted against AGN bolometric luminosity. We conclude that mergers are most connected with dust-obscured AGNs (linked to a fast-growing phase of the SMBH) and are the main or even the sole fuelling mechanism of the most powerful AGNs.

astro-ph.GA↗

arXiv2024

The properties of AGN in dwarf galaxies identified via SED fitting

Given their dominance of the galaxy number density, dwarf galaxies are central to our understanding of galaxy formation. While the incidence of AGN and their impact on galaxy evolution has been extensively studied in massive galaxies, much less is known about the role of AGN in the evolution of dwarfs. We search for radiatively-efficient AGN in the nearby (0.1 < z < 0.3) dwarf (10^8 MSun < M < 10^10 MSun) population, using SED fitting (via Prospector) applied to deep ultraviolet to mid-infrared photometry of 508 dwarf galaxies. Around a third (32 +/- 2 per cent) of our dwarfs show signs of AGN activity. We compare the properties of our dwarf AGN to control samples, constructed from non-AGN, which have the same distributions of redshift and stellar mass as their AGN counterparts. KS tests between the AGN and control distributions indicates that the AGN do not show differences in their distances to nodes, filaments and nearby massive galaxies from their control counterparts. This indicates that AGN triggering in the dwarf regime is not strongly correlated with local environment. The fraction of AGN hosts with early-type morphology and those that are interacting are also indistinguishable from the controls within the uncertainties, suggesting that interactions do not play a significant role in inducing AGN activity in our sample. Finally, the star formation activity in dwarf AGN is only slightly lower than that in their control counterparts, suggesting that the presence of radiatively-efficient AGN does not lead to significant, prompt quenching of star formation in these systems.

astro-ph.GA↗

arXiv2024

Galaxy merger challenge: A comparison study between machine learning-based detection methods

Various galaxy merger detection methods have been applied to diverse datasets. However, it is difficult to understand how they compare. We aim to benchmark the relative performance of machine learning (ML) merger detection methods. We explore six leading ML methods using three main datasets. The first one (the training data) consists of mock observations from the IllustrisTNG simulations and allows us to quantify the performance metrics of the detection methods. The second one consists of mock observations from the Horizon-AGN simulations, introduced to evaluate the performance of classifiers trained on different, but comparable data. The third one consists of real observations from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP) survey. For the binary classification task (mergers vs. non-mergers), all methods perform reasonably well in the domain of the training data. At $0.1<z<0.3$, precision and recall range between $\sim$70\% and 80\%, both of which decrease with increasing $z$ as expected (by $\sim$5\% for precision and $\sim$10\% for recall at $0.76<z<1.0$). When transferred to a different domain, the precision of all classifiers is only slightly reduced, but the recall is significantly worse (by $\sim$20-40\% depending on the method). Zoobot offers the best overall performance in terms of precision and F1 score. When applied to real HSC observations, all methods agree well with visual labels of clear mergers but can differ by more than an order of magnitude in predicting the overall fraction of major mergers. For the multi-class classification task to distinguish pre-, post- and non-mergers, none of the methods offer a good performance, which could be partly due to limitations in resolution and depth of the data. With the advent of better quality data (e.g. JWST and Euclid), it is important to improve our ability to detect mergers and distinguish between merger stages.

astro-ph.GA↗

arXiv2023

The formation of cores in galaxies across cosmic time -- the existence of cores is not in tension with the LCDM paradigm

The `core-cusp' problem is considered a key challenge to the LCDM paradigm. Halos in dark matter only simulations exhibit `cuspy' profiles, where density continuously increases towards the centre. However, the dark matter profiles of many observed galaxies (particularly in the dwarf regime) deviate strongly from this prediction, with much flatter central regions (`cores'). We use NewHorizon (NH), a hydrodynamical cosmological simulation, to investigate core formation, using a statistically significant number of galaxies in a cosmological volume. Halos containing galaxies in the upper (M* > 10^10.2 MSun) and lower (M* < 10^8 MSun) ends of the stellar mass distribution contain cusps. However, halos containing galaxies with intermediate (10^8 MSun < M* < 10^10.2 MSun) stellar masses are generally cored, with typical halo masses between 10^10.2 MSun and 10^11.5 MSun. Cores form through supernova-driven gas removal from halo centres, which alters the central gravitational potential, inducing dark matter to migrate to larger radii. While all massive (M* > 10^9.5 MSun) galaxies undergo a cored-phase, in some cases cores can be removed and cusps reformed. This happens if a galaxy undergoes sustained star formation at high redshift, which results in stars (which, unlike the gas, cannot be removed by baryonic feedback) dominating the central gravitational potential. After cosmic star formation peaks, the number of cores, and the mass of the halos they are formed in, remain constant, indicating that cores are being routinely formed over cosmic time after a threshold halo mass is reached. The existence of cores is, therefore, not in tension with the standard paradigm.

astro-ph.GA↗