arXiv Science⌕ Search

arXiv subjects

Christian Kragh Jespersen

Publications and source records attributed to Christian Kragh Jespersen.

At least 19 recordsLinked to original sources
arXiv2026

On the importance of medium-resolution continuum spectroscopy for JWST's extragalactic legacy

We argue for a significant investment of JWST time into an extremely deep medium-resolution NIRSpec MSA survey targeting a large, representative sample of high-redshift $(z>1.5)$ galaxies, specifically intended to obtain high-SNR continuum detections. This is only possible with JWST, and no planned future instrumentation has similar capabilities, meaning that obtaining such observations with JWST must be urgently prioritised. The vast majority of existing archival NIRSpec observations are not deep enough for continuum science, being focused on the detection of emission lines. Deeper continuum data opens up the ability to constrain a much wider range of key galaxy physical parameters, including stellar population ages and metallicities, stellar kinematics and neutral gas physical properties, inflows and outflows. Such measurements will provide crucial insight into quenching physics, galaxy dynamical masses, merger histories, pristine early star-formation, reionization and the physics of Little Red Dots, as well as enabling all of the emission-line science that can be achieved with shallower observations. To facilitate future survey planning, we present an empirically derived relationship between rest-optical continuum SNR in JWST NIRSpec medium-resolution data, rest-optical apparent magnitude and exposure time, which we show to be independent of redshift (and hence the grating used). We then provide a 1000-hour sketch survey plan based on achieving the continuum SNR $=20$/pixel necessary to constrain key physical properties and scaling relations for a large galaxy sample at $z>1.5$. We finally discuss NIRSpec sky subtraction strategies, which are of critical importance for continuum SNR.

astro-ph.IM↗
arXiv2026

A Large Diversity in the Quiescent Fractions of Massive Galaxies in Proto-clusters at $z\sim4-6$ in the FRESCO GOODS fields

To constrain the emergence of the first quiescent galaxies (QGs) and their connection to large-scale structure, we search for rare, high-redshift QGs in proto-clusters using spectral energy distribution modeling, multiple rest-frame color diagnostics, and visual inspection. Our sample includes 12 QG candidates with stellar masses log(M*/Msun) ~ 9-11 in four proto-clusters at z ~ 4-6 within GOODS-North (GOODS-N) and GOODS-South (GOODS-S). The selection uses 11 NIRCam filters in GOODS-N and 17 NIRCam plus 2 MIRI filters in GOODS-S, combined with 9 ancillary HST filters, covering 0.4-7.7 microns. We measure an increasing quiescent fraction with stellar mass, with averages of 1.0%(+0.7/-0.4; 5/502), 3.5%(+2.8/-1.7; 4/115), and 33.3%(+22.2/-17.5; 3/9) for log(M*/Msun) = 9 - 9.75, 9.75 - 10.5, 10.5 - 11.25, respectively, across four proto-clusters. Notably, the two GOODS-N proto-clusters at z=4.406 and z=5.194 have a quiescent fraction of 100% (2/2 at z=4.406; 1/1 at z=5.194) in their highest mass bin (log(M*/Msun) > 10.5), whereas the proto-clusters in GOODS-S at similar redshifts have zero massive quiescent galaxies. Contrary to the expectation that massive galaxies occupy cluster centers at z~0, our QGs are found primarily on the outskirts of overdensity peaks. We also find fewer robust QGs at z>5 (1/8) and none in the z=5.384 proto-cluster, consistent with simulations suggesting that efficient quenching mechanisms had not yet matured. Our results highlight the pronounced variations in quiescent fractions across proto-clusters at z>4. The emerging diversity of proto-cluster QG populations calls for a stronger understanding of the role of environment in early galaxy evolution.

astro-ph.GA↗
arXiv2026

Graph Neural Networks for Astrophysics and Cosmology

Physics seldom makes breakthroughs when we consider objects only in isolation. Instead we model and measure how objects sit among their neighbours, spatial, temporal, or otherwise: a galaxy in its environment, a halo in its history, a photon hit in a detector. Graphs are ideally suited for encoding exactly these relations, and Graph Neural Networks (GNNs) learn the geometric functions that respect them. This review presents the use of GNNs for astrophysics and cosmology, and argues that GNNs often suit physics and physical data better than any alternative: they fit the native geometry of the domain and the signal, enable the incorporation of the symmetries and prior knowledge we already trust and have toiled to build, and stay flexible enough to capture what we do not yet understand. GNNs' physically intuitive workings also render them more interpretable than other deep-learning architectures. We show how we can begin from general principles of the geometry of deep learning and build the graph and the graph neural network from the ground up, and how other architectures, like the convolutional network and the transformer, can be seen as special cases of one unified geometric blueprint. We survey applications across cosmology, galaxy formation, dynamics, and instrumentation; and we chart the frontier. From this survey, we find that only one in six articles outside instrumentation applies GNNs to data rather than just simulations. We also distil a set of best practices, including when not to reach for a graph. Throughout, we will prioritise physical intuition and foundational principles over mathematical machinery whenever possible.

astro-ph.IM↗
arXiv2026

The ionised interstellar medium of DSFGs revealed by JWST/NIRSpec and ALMA: Super-solar metallicity, low ionisation parameters and, typical electron densities

We present a detailed study of near-infrared (2-4$\rm μ$m) JWST/NIRSpec spectra of 48 high-redshift ($z=2.53^{+1.32}_{-0.70}$) galaxies detected with ALMA at $>3σ$. From a multi-wavelength SED analysis we establish the sample has a a median stellar mass of $\rm\log_{10}(M_\ast/M_\odot)=10.8\pm0.1$ and dust mass of $\rm\log_{10}(M_{\rm d}/M_\odot)=8.7\pm0.1$, covering a broad range of far-infrared luminosity $\rm (\log_{10}(L_{FIR}/L_\odot)=10.9-12.7)$. The majority of sources show no signs of AGN activity, with 40% having either X-ray counterparts $(\rm L_{Xc}>10^{42}erg/s)$, elevated optical line ratios, or broad (FWHM>800 km/s) H$α$ profiles, although we note this is a lower limit due to the stochastic placement of NIRSpec slits. We establish the sample has a median gas-phase metallicity of $12+\log({\rm O/H})=8.71\pm0.02$, as derived from the [NII]/H$α$ ratio, with the most FIR-luminous galaxies ($\rm\log_{10}(L_{\rm FIR}/L_\odot)>12$) falling $0.15\pm0.03$dex above the fundamental metallicity relation. From the [SII] emission-line doublet ratio, we measure a median electron density of $\log_{10}(n_{\rm e}/{\rm cm}^{-3})=2.53\pm0.07$ consistent with less-massive, star-forming, galaxies at the same epoch. For nine galaxies with [OII] and H$β$ detections (median $\rm\log_{10}(L_{\rm FIR}/L_\odot)=11.81\pm0.15$), we derive a median observed (dust-uncorrected) ionisation parameter of $\rm\log_{10}(U)=-2.84\pm0.06$. Our results indicate that luminous far-infrared galaxies are massive, chemically evolved systems that appear to deviate from the standard dust and metal production equilibrium observed in less obscured galaxies. This study demonstrates the synergy of JWST and ALMA in unveiling the nature of DSFGs, and highlights the need for a NIRSpec survey of uniformly selected, massive, dust-obscured, galaxies to fully characterise their interstellar medium.

astro-ph.GA↗
arXiv2026

PANORAMIC: The Dawn of Massive Quiescent Galaxies I. Number Density and Cosmic Variance from 1000 arcmin$^2$ NIRCam Imaging

We measure the number density and field-to-field variance of massive quiescent galaxies at $z\sim3$ - 8 using the JWST/NIRCam pure-parallel imaging survey PANORAMIC together with archival observations, covering an area of 0.28 deg$^2$ ($\sim1000$ arcmin$^2$) in at least six filters. We identify quiescent galaxy candidates at $z\gtrsim3$ with $M_\ast \gtrsim 10^{10}\,M_\odot$, comprising 101 galaxies in a gold sample of high-confidence candidates and 137 in a more inclusive silver sample. We measure their evolving comoving number density, finding $(1.5$ vs. $3.1)\times10^{-5}\,\mathrm{Mpc}^{-3}$ at $z=3$ - 4 for the gold and silver samples, respectively, and a decline by more than a factor of 20 by $z\sim6$. Comparisons with empirical models and cosmological simulations show that widely used frameworks underpredict the abundance of massive quiescent galaxies at $z\gtrsim4$ by $\gtrsim1$ dex, indicating that current implementations of early star formation, feedback, and quenching do not produce enough early quenched systems. With 34 independent sightlines, we present the first direct empirical measurement of field-to-field variance for quiescent galaxies at $z>3$, finding a high cosmic variance of $σ_{\rm CV}\approx0.7\pm0.3$. This exceeds predictions from abundance-matched mock catalogs, suggesting that early quiescent galaxies are more strongly clustered, and more likely to be found near one another or in more biased regions, than expected in current galaxy-formation models. Any successful model for the emergence of early massive quiescent galaxies must reproduce both their abundance evolution and their imprint on the large-scale distribution.

astro-ph.GA↗
arXiv2026

DeepDive: Simultaneous Formation of Massive Quiescent Galaxies in High-Redshift Galaxy Overdensities

We report on the spectroscopic confirmation of overdense regions of massive quiescent galaxies (QGs) in the early Universe with JWST/NIRSpec. Based on data from the DeepDive NIRSpec program and archival data from the Dawn JWST Archive, we confirm three QGs in the vicinity of Jekyll & Hyde, a pair of massive QG and a dusty star-forming galaxy, at $z=3.71$ and two QGs around SXDS-27434 at $z=4.01$. According to the analysis of galaxy number density with photometric redshifts, Jekyll & Hyde (SXDS-27434) are in an overdense region, where the number density of galaxies is three (four) times higher than the average in the COSMOS (SXDS) field. SED fitting suggests that most of the QGs follow similar star formation histories and have consistent formation and quenching epochs. The same trend is observed in other proto-clusters hosting QGs that were already identified by ground-based telescopes, indicating that the large-scale environment plays an important role in the formation of QGs. In addition, JWST spectra reveal a broad H$α$ emission line from SXDS-27434 and faint emission lines from other three QGs, which are identified as AGN-driven based on their emission line ratios. The overdensity is also reproduced by the Illustris TNG300 simulation at $z=3.71$, in which the member QGs also have similar quenching epochs. These results are consistent with a scenario in which the large-scale structure enhances merger activity and/or gas accretion and triggers AGN feedback, thereby driving simultaneous formation and quenching of QGs in the overdensity.

astro-ph.GA↗
arXiv2026

The Roman eXtreme Deep Field (RXDF)

The Roman eXtreme Deep Field (RXDF) program is one of the five General Astrophysics Survey (GAS) programs approved for observing time with the Nancy Grace Roman Space Telescope in Cycles 1 and 2. It has been allocated 386.41 hours to carry out an imaging survey to AB = 30 mag (5-sigma) over ~140x larger area than the Hubble eXtreme Deep Field (HXDF) full-depth area (ACS+WFC3/IR). The RXDF will cover the full Roman wavelength range with 7 bands, reaching AB = 30 mag in RZYJH, 29 mag in F, and 28 mag in K, over a full-depth area of 678.75 arcmin^2 embedded in a total area of 1,243 arcmin^2, and far exceeding the depths of the Roman Core Community Surveys (CCS). The RXDF is within the Euclid Ultra Deep Field (EUDF) near the North Ecliptic Pole (NEP), a strategic long-term field for generational space facilities, with a wealth of multi-wavelength data including extensive coverage from the James Webb Space Telescope (JWST) NEXUS Treasury program. The observations will cover 3 epochs at a 1-year cadence, each epoch divided into 3 sub-epochs ~10 days apart, enabling time-domain studies on time baselines from ~10 days to over ~2 years. The RXDF is uniquely positioned to address critical questions in reionization, large scale structure (LSS), growth of supermassive black holes (SMBHs), little red dots (LRDs), and high-z supernovae (SNe); the volumes probed by HST+JWST are too small at these extreme depths, and even the deepest CCS tiers are too shallow. In addition to our key objectives, a wealth of additional science will be enabled by engaging the community with our rapidly released datasets, revolutionizing a wide range of science for a lasting legacy. This short document, which is converted from the approved RXDF proposal, aims to provide the community with a summary of the program.

astro-ph.GA↗
arXiv2026

Introducing sapphire: Towards Hybrid Physics-Informed, Data-Driven Modeling of Galaxy Formation

Semi-analytic models (SAMs) have been treating galaxy populations as dynamical systems for $\gtrsim50$ years, but their evolution equations remain poorly constrained. We introduce sapphire, a modular, automatically differentiable, GPU-accelerated SAM written in JAX. For the first time, we compute exact Jacobian and Hessian matrices of a galaxy formation SAM, using the Pandya et al. (2023) nonlinear differential equation system as an example. These allow efficient, interpretable local and global sensitivity analyses, which reveal that supernova energy loading is the key astrophysical parameter. We use gradient descent and Hamiltonian Monte Carlo (HMC) to perform comprehensive mock parameter recovery tests. These indicate that the $z=0$ stellar-to-halo-mass relation alone does not contain enough information to infer many astrophysical parameters. Using observations of star-forming galaxies from the MaNGA survey and the Behroozi et al. (2019) empirical model as one baseline, we derive multiple posteriors assuming different combinations of data, including $z=0$ interstellar medium gas fractions and metallicities. The inferred physical parameters suggest that galaxies self-regulate their star formation primarily through preventative rather than ejective feedback, though this remains uncertain due to the lack of satellite galaxies, black holes and multi-phase galactic atmosphere physics. Both Fisher and HMC forecasts demonstrate the potential of sapphire to enable precision inference for galaxy formation and cosmology in a hybrid physics-informed, data-driven way, but more work is needed to expand its library of models and methods. We make sapphire publicly available at https://github.com/virajpandya/sapphire.

astro-ph.GA↗
arXiv2026

Exploring Cosmic Dawn with PANORAMIC II: Cosmic Variance and Galaxy Clustering at $z\sim10$

Observational campaigns with JWST have revealed a higher-than-expected abundance of UV-bright galaxies at $z\gtrsim10$, with various proposed theoretical explanations. A powerful complementary constraint to break degeneracies between different models is galaxy clustering. In this paper, we combine PANORAMIC pure parallel and legacy imaging along 34 independent sightlines to measure the cosmic variance ($σ_{\rm CV}$) in the number counts of Lyman break galaxies at $z\sim10$ which is directly related to their clustering strength. We find $σ_{\rm CV}=0.96^{+0.20}_{-0.18}$, $1.46^{+0.54}_{-0.44}$, and $1.71^{+0.72}_{-0.59}$ per NIRCam pointing ($\sim9.7\,{\rm arcmin}^2$, $\lesssim1.5\,{\rm pMpc}$ at $z\sim10$) for galaxies with M$_{\rm UV}<-19.5$, $-20$, and $-20.5$. Comparing to galaxies in the fiducial UniverseMachine, we find that $σ_{\rm CV}$ is consistent with our measurements, but that the number densities are a factor $\gtrsim5$ lower. We implement simple models in the UniverseMachine that represent different physical mechanisms to enhance the number density of UV-bright galaxies. All models decrease $σ_{\rm CV}$ by placing galaxies at fixed M$_{\rm UV}$ in lower mass halos, but to varying degrees. Combined constraints on $σ_{\rm CV}$ and the UVLF thus tentatively disfavor models that globally increase the star formation efficiency (SFE) or the scatter in the M$_{\rm UV}$-$M_{\rm halo}$ relation, while models that decrease the mass-to-light ratio, or assume a power-law scaling of the SFE with $M_{\rm halo}$ agree better with the data. We show that with sufficient additional independent sightlines, robust discrimination between models is possible, paving the way for powerful constraints on the physics of early galaxy evolution through NIRCam pure parallel imaging.

astro-ph.GA↗
arXiv2026

A Black Hole Star at Cosmic Noon: Extreme Balmer break, photospheric continuum, and broad absorption by thick winds in a Little Red Dot at z=1.7

Recent studies at high redshift have revealed an enigmatic class of Little Red Dots (LRDs) with extreme Balmer breaks, stronger than in any stellar atmosphere. However, it is unclear whether such objects exist at lower redshift, especially given the low number of LRDs reported at $z\lesssim 2$. Here we report the discovery of PAN-BH*-1, an LRD with an extreme Balmer break at $z=1.73$, identified from JWST/NIRCam pure-parallel imaging taken by the PANORAMIC survey, and confirmed by deep VLT/X-Shooter spectroscopy. The rest-optical to near-infrared spectral energy distribution of PAN-BH*-1 is consistent with a photospheric continuum with effective temperature $T_{\rm eff}\approx 4800$ K. The broad H$α$ emission line shows remarkably deep absorption, stronger than previously measured in any LRD. The absorption trough spans from $-520$ km/s to $+267$ km/s with respect to the systemic redshift. The presence of blue- and red-shifted absorption suggests complex dynamics of the obscuring gas along the line of sight. We speculate that the absorption trough can be produced by a thick wind launched from a thick, rotating photospheric disk, the latter being the source of the red optical continuum. While the source is unresolved in the rest-optical JWST data ($r_{\rm eff,UV}<47$ pc), the rest-NUV HST imaging shows an extended morphology with $r_{\rm eff,opt}=1.0^{+0.5}_{-0.3}$ kpc, that we interpret as a host galaxy with a stellar mass $\sim 10^8$ $M_\odot$, in line with the narrow H$α$ emission. The discovery of this object at cosmic noon highlights the feasibility of systematic searches for extreme LRDs with wide-area facilities such as Euclid and Roman.

astro-ph.GA↗
arXiv2026

Black Hole Stars Across the Universe: Identifying Central Engine Dominated Little Red Dots at $z\sim1.5-9.5$

Photometric selections of Little Red Dots (LRDs) largely rely on identifying their ``V-shaped'' spectral energy distribution (SED). Recent work suggests this V-shape stems from a combination of a central engine -- also referred to as a Black Hole Star (BH*) -- and a star-forming host galaxy. We present a new and highly complementary photometric selection that is based on incorporating BH* templates in the \texttt{eazy} redshift fitting code. Selecting compact sources where a BH* template contributes $>80$\% to the best fitting SED in the rest-optical, we compile a sample of 241 BH*-dominated candidates from $\sim1000\,{\rm arcmin}^2$ of legacy and pure parallel JWST imaging. Our selection does not require a blue UV-component, and it successfully identifies objects that resemble the paradigmatic sources ``MoM-BH*-1'' and ``The Cliff''. We find that BH*-dominated sources exist across a wide range of redshifts ($z\sim1.7-9.3$) and optical luminosities (log$(L_{5100}/{\rm erg}\,{\rm s}^{-1})\sim42-44.5$), and we measure a median Balmer break strength of $\sim3$, with some breaks reaching values $>10$. We estimate bolometric luminosities in the range log$(L_{\rm bol}/{\rm erg}\,{\rm s}^{-1})\sim42-45$, which, assuming accretion at the Eddington-limit, would translate to black hole masses of $M_{\rm BH}\sim10^4-10^7{\rm M_\odot}$, spanning the intermediate mass black hole to the quasar regime. The number density of BH*-dominated candidates peaks at $z\sim5-6$ ($\sim10^{-5}\,{\rm Mpc}^{-3}$) and it declines by an order of magnitude down to $z\sim2$. Tentatively, comparing to V-shaped LRD samples suggests that the fraction of BH*-dominated sources among the broader LRD population does not decrease towards lower redshift. Crucially, our work demonstrates that BH*-dominated sources are not merely an early-Universe phenomenon but rather persist at least until cosmic noon.

astro-ph.GA↗
arXiv2026

Exploring Cosmic Dawn with PANORAMIC I: The Bright End of the UVLF at $z\sim9 -17$

In its first two years of operation, the James Webb Space Telescope has enabled the discovery of a surprising number of UV-bright galaxies at $z\sim10-14$. Their number density is still relatively uncertain, due to cosmic variance effects, and the limited survey area with deep imaging. Here, we combine pure parallel imaging from the PANORAMIC survey with data from legacy fields to constrain the bright end (M$_{\rm UV}<-18.5$) of the UV luminosity function (UVLF) over $0.28\,$deg$^2$ of NIRCam imaging in 6 or more filters, and along 35 independent lines of sight. Using conservative color selections, we compile robust dropout samples at $z\sim10$, $z\sim13$, and $z\sim17$, and identify 16 new candidates from PANORAMIC. Our inferred UVLFs at $z\sim10$ are consistent with literature results and we confirm the high abundance of galaxies at the bright end (M$_{\rm UV}\lesssim-21$) with better number statistics. We find somewhat lower number densities at $z\sim13$ compared to previous studies, and no robust candidates at $z\sim17$, indicating a rapid evolution of the galaxy population from $z\sim10-17$. The improved upper limits at $z\sim17$ imply that the cosmic UV luminosity density drops by at least a factor $\sim50$ from $z\sim10$ to $z\sim17$. Comparing our results to models proposed to explain the abundance of UV-bright galaxies at $z\gtrsim10$, we conclude that a modest increase in the star formation efficiency, or in the burstiness of star formation, a more top-heavy initial mass function, a lack of dust attenuation, or a combination of these effects at $z\gtrsim10$, is sufficient to match our observational constraints.

astro-ph.GA↗
arXiv2026

There and back again? Neutral outflows in z~3.5 quiescent galaxies

Neutral gas outflows play a crucial role in the baryon cycle of galaxies, their properties provide key insights into the transition from star formation to quiescence. In this work, we investigate the neutral gas outflow of 23 massive ($M_\star = 10^{10.1-11.6}\,\rm M_\odot$) quiescent galaxies (QGs) at z=2.82--4.61, selected from the JWST NIRSpec (R~1000) and NIRCam program DeepDive. We trace the neutral gas outflows using the NaI Doublet absorption lines and detect excess NaI D in 13/23 (57%) targets, of which 7/23 (30%) show blueshifted absorption with velocity offsets $|Δv|$ >~ 150 km/s. The z ~ 3.5 targets have $Δv$ similar to those of their local counterparts; they are also equivalent when compared in SFR--$Δv$ space. We derive mass outflow rates and identify the most extreme neutral gas outflow rate $\log(\dot M_{\rm out} / \mathrm{M_\odot \, yr}^{-1})=2.68\pm0.27$ beyond the local Universe, coincident with an X-ray AGN. For all NaI D detected systems, the inferred mass outflow rate can, in principle, suppress ongoing star formation; however, the outflows are unlikely to escape their hosts, suggesting recycling on relatively short timescales (~3--180 Myr), depending on the assumed potential and launching radius. All NaI D detected targets occupy the LI(N)ER region of the BPT diagram and/or are X-ray detected, but we find no strong correlation between ongoing AGN and the neutral outflow: 2/4 broad-line/X-ray AGNs are NaI D undetected -- yet, the outflows can be powered by fossil/episodic AGNs, and one broad-line target shows a possible P-Cygni profile that indicates strong outflows. As neutral outflows alone are not able to permanently quench star formation by removing gas in our sample at z ~ 3.5, the presence of gas cycling in and out of massive passive systems may instead be the signature of feedback-regulated quenching-maintenance processes.

astro-ph.GA↗
arXiv2026

Winding Back the Clock: Recent Star Formation Histories of Massive Quiescent Galaxies Are Consistent With Their Rapid Number Density Evolution Since $\mathbf{z\sim7}$

Massive quiescent galaxies have been identified out to $z\sim7$ in early JWST data in a substantial excess ($\rm \gtrsim 1\,dex$ at $z>4$) of number densities from most theoretical predictions. We investigate whether the number densities implied by the star formation histories of quiescent galaxies at $2 7$. For this work, we rely on stellar population synthesis modeling of JWST NIRCam photometry (from CEERS and PRIMER) and NIRSpec/PRISM spectra of massive ($\rm M_{*} > 10^{10.5}M_{\odot}$) quiescent galaxies in the RUBIES survey. We infer their star-formation histories through Bayesian spectro-photometric fitting with Prospector, exploring the sensitivity of our results to stellar libraries and SFH priors. For each source, we compute a timescale over which it would be identified as quiescent -- leveraging the recent and most robust SFH timescale -- and deduce the number density of the quiescent population at previous epochs. These reconstructed number densities are then compared to existing observational constraints, including a new measurement from the PANORAMIC pure parallel survey, whose wide-area and independent sightlines reduce sensitivity to cosmic variance. We find striking agreement between reconstructed and observed number densities up to $z\sim7$, a self-consistency that lends credence to stellar population synthesis modeling of distant quiescent galaxies. Furthermore, by connecting the recent ($\rm \sim 1\,Gyr$) star-formation histories and number densities of quiescent galaxies and their implied progenitors, we reinforce the known tension between observations and model predictions at $3<z<7$.

astro-ph.GA↗
arXiv2026

A PANORAMIC of UV-optical morphologies of "Little Red Dots": Two groups of LRDs distinguished by UV half-light radius

Among the most remarkable results from JWST is the discovery of abundant, compact, and very red sources in the early Universe known as "Little Red Dots" (LRDs). The relative degree to which starlight and active galactic nuclei (AGN) drive the rest-frame UV and optical emission from LRDs remains unclear. With a large sample of LRDs selected photometrically from the pure-parallel PANORAMIC survey, we study their morphology as a function of rest-wavelength and find that the rest-UV light is typically more extended than the rest-optical. This result holds both when measuring LRD sizes with a single Sérsic profile and when comparing the fraction of light from a point source via joint PSF+Sérsic modeling. A shift occurs at the Balmer break, with LRDs becoming highly compact and unresolved ($R_{50,\rm{opt}}\lesssim100\;\rm{pc}$) in the rest-optical relative to the rest-UV. When splitting the sample at the Balmer break into those that are resolved and unresolved, a stacking analysis demonstrates that the latter are compact ($R_{50}\lesssim100\;\rm{pc}$) on average across the full rest-UV-optical spectrum. Conversely, those LRDs resolved at the break show extended UV emission ($R_{50,\rm{UV}}>200\;\rm{pc}$) on average. We find a similar dichotomy when repeating with a spectroscopic sample. Altogether, these results are consistent with the rest-UV emission driven by a combination of emission from starlight and a dense, dust-poor cloud of hydrogen gas enveloping an AGN. Differences between LRDs in the relative contribution from the AGN and starlight could reflect an ensemble of black hole seed masses, where a heavier seed produces an LRD of smaller $R_{50,\rm{UV}}$.

astro-ph.GA↗
arXiv2025

CosmoBench: A Multiscale, Multiview, Multitask Cosmology Benchmark for Geometric Deep Learning

Cosmological simulations provide a wealth of data in the form of point clouds and directed trees. A crucial goal is to extract insights from this data that shed light on the nature and composition of the Universe. In this paper we introduce CosmoBench, a benchmark dataset curated from state-of-the-art cosmological simulations whose runs required more than 41 million core-hours and generated over two petabytes of data. CosmoBench is the largest dataset of its kind: it contains 34 thousand point clouds from simulations of dark matter halos and galaxies at three different length scales, as well as 25 thousand directed trees that record the formation history of halos on two different time scales. The data in CosmoBench can be used for multiple tasks -- to predict cosmological parameters from point clouds and merger trees, to predict the velocities of individual halos and galaxies from their collective positions, and to reconstruct merger trees on finer time scales from those on coarser time scales. We provide several baselines on these tasks, some based on established approaches from cosmological modeling and others rooted in machine learning. For the latter, we study different approaches -- from simple linear models that are minimally constrained by symmetries to much larger and more computationally-demanding models in deep learning, such as graph neural networks. We find that least-squares fits with a handful of invariant features sometimes outperform deep architectures with many more parameters and far longer training time. Still there remains tremendous potential to improve these baselines by combining machine learning and cosmology to fully exploit the data. CosmoBench sets the stage for bridging cosmology and geometric deep learning at scale. We invite the community to push the frontier of scientific discovery by engaging with this dataset, available at https://cosmobench.streamlit.app

cs.LG↗
arXiv2025

Transfer Learning Beyond the Standard Model

Machine learning enables powerful cosmological inference but typically requires many high-fidelity simulations covering many cosmological models. Transfer learning offers a way to reduce the simulation cost by reusing knowledge across models. We show that pre-training on the standard model of cosmology, $Λ$CDM, and fine-tuning on various beyond-$Λ$CDM scenarios -- including massive neutrinos, modified gravity, and primordial non-Gaussianities -- can enable inference with significantly fewer beyond-$Λ$CDM simulations. However, we also show that negative transfer can occur when strong physical degeneracies exist between $Λ$CDM and beyond-$Λ$CDM parameters. We consider various transfer architectures, finding that including bottleneck structures provides the best performance. Our findings illustrate the opportunities and pitfalls of foundation-model approaches in physics: pre-training can accelerate inference, but may also hinder learning new physics.

astro-ph.CO↗