Astrophysics papers — 2026-09-07

We are learning how much gas actually stays within galaxy groups, which is a fundamental question for understanding how energy from stars and black holes pushes matter around the universe. By stacking cosmic microwave background maps from the Atacama Cosmology Telescope around galaxy groups identified in the DESI Legacy Survey, researchers have found that low-mass groups are significantly depleted of gas compared to their massive counterparts.

While massive groups hold a baryon fraction consistent with the universal average, smaller groups appear to have lost much of their content. This is likely because energetic feedback has pushed the gas out into extended profiles far beyond the virial radius.

This missing gas might be easier to track in the future using a new estimator designed to map the full ionized electron field directly. By using large-scale velocity reconstruction from galaxy surveys, this new method bypasses the need for uncertain models of how galaxies cluster, allowing us to see the electron distribution without being biased by where the light is.

It is expected to provide high-significance tomographic measurements of these gas distributions when combined with upcoming data from the Simons Observatory and DESI. Moving from large-scale structures to individual stellar systems, new simulations are reshaping how we think about galaxy evolution in dense environments.

By tracing the histories of dwarf galaxies in the IllustrisTNG simulation, researchers found that some compact dwarfs actually begin their lives as ultra-diffuse galaxies with massive amounts of gas. Rather than being stripped remnants of a pre-existing core, these compact stars appear to be built through intense starbursts triggered by the cluster environment itself during the transformation process.

On a much smaller scale, we are seeing unexpected behavior in how neutron stars move through the envelopes of massive stars. New general-relativistic hydrodynamical simulations show that as a neutron star is engulfed, it forms nested bow shocks that can actually reverse the direction of the drag force or increase it by up to two orders of magnitude compared to standard predictions.

This suggests our current models for how these binaries evolve and eventually merge might need significant recalibration. Finally, we are seeing some fascinating outliers in our own Galaxy's stellar census.

A comprehensive dynamical study of Cepheid variables has identified eighteen "rogue" stars that follow highly inclined or even retrograde orbits, which is very unusual for stars that should be settled in the thin disc. Most of these appear to be genuine runaway stars rather than misclassifications, providing a new way to map the chaotic history of stellar motions in the Milky Way.

We finally have a clearer picture of why massive objects sometimes stop or even reverse their inward migration in galaxies rather than sinking straight to the center. By looking at the distribution function of a host galaxy through high-resolution N-body simulations, researchers found that core stalling and dynamical buoyancy are driven by plateaus or inflections in phase space rather than just central density gradients.

This means structurally similar galactic cores can behave radically differently depending on their underlying distribution, which has huge implications for how black holes coalesce in dwarf galaxies. On a much larger scale, the ODIN survey has confirmed six massive protoclusters at cosmic noon by combining Ly α imaging with spectroscopy across the COSMOS and XMM-LSS fields.

These structures show that environment plays a massive role in galaxy evolution, as galaxies in these dense cores exhibit higher median line fluxes and a notable deficit of faint emitters compared to the field. The search for better ways to measure the universe's expansion continues with new models attempting to fix the Hubble tension.

By testing a cascade decaying dark matter sector that affects both early and late times, researchers found they could reach an H0 value of 68.76 km s-1 Mpc-1, though achieving higher values requires accepting much larger uncertainties in other cosmological parameters. We are getting much closer to understanding the very first moments of our universe by using machine learning to look backward through time.

Researchers have successfully used a three-dimensional U-Net to reconstruct the initial density field from simulated maps of 21-cm and CO line emissions, effectively peeling back layers of non-linear structure formation. By combining these two tracers—one looking at low-density neutral gas and the other at overdense star-forming regions—the reconstruction achieved a high cross-correlation coefficient even when accounting for realistic instrumental noise from future surveys like SKA1-Low.

This process essentially recovers lost cosmological information, tightening constraints on parameters like sigma eight and the spectral index by about two times. This ability to recover hidden information is equally vital for resolving the current tension in our measurements of how fast the universe is expanding.

New work shows that most proposed solutions to the Hubble tension that act before recombination actually create a new problem by pushing preferred baryon density levels into direct conflict with Big Bang Nucleosynthesis data. Specifically, when researchers include primordial deuterium measurements in their models, these early-time solutions struggle to recover the high Hubble constant we observe today.

While we work to understand the large-scale evolution of the cosmos, we are also refining how we use gravitational waves as "dark sirens" to map out distances. A new methodology has been developed to handle incomplete galaxy catalogues by using a sampled redshift prior, allowing researchers to jointly infer the properties of host galaxies alongside cosmological parameters.

This approach yielded an updated Hubble constant measurement of 71.9 with a median value that helps bridge some of the gaps in our current models. On a much smaller scale, we are getting better at identifying the specific signatures left behind by inflation through the study of primordial non-Gaussianity.

A new strategy for multitracer analysis suggests that instead of relying on difficult-to-observe secondary halo properties, we can split tracer samples by their large-scale dark matter environment to mitigate cosmic variance. Using this environmental approach with upcoming surveys like DESI could improve our constraints on these inflationary signatures by a factor of two to three.

Looking toward the search for life, we are learning exactly what to expect from the upcoming PLATO mission's observations of M-dwarf stars. Simulations of the mission's long-duration observation phase show that we can expect to find roughly four Earth-sized planets and seven super-Earths in habitable zones within the target field.

The study provides a critical forecast for completeness and false-alarm rates, proving that our ability to detect these worlds depends heavily on how we handle instrument systematics at the detection frontier. In our own solar neighborhood, new 3D magnetohydrodynamic simulations are finally matching the temperature jumps observed by the Voyager 2 spacecraft at the termination shock.

By distributing shock heating more realistically among protons, electrons, and pickup-ions—rather than channeling it all into cold protons—the models now reproduce observed conditions within a factor of ten to twenty. This is a major step forward because these hot ions significantly influence the overall shape of our heliosphere.

Finally, we are getting much clearer views of distant radio sources thanks to new X-ray data from the Swift Observatory. By adding deeper observations to existing catalogs, researchers have detected several previously unseen X-ray counterparts and provided more precise positions for known ones.

These detections reveal soft, diffuse emission that points toward the presence of hot galaxy coronae or group cores, helping us map the energetic environments of these distant objects. We finally have proof that we can use Webb to hunt for tiny moons around other stars, which is a massive leap forward for finding habitable worlds.

By analyzing twelve transits of the temperate planet LP 890-9c, researchers were able to rule out any moons larger than 0.1 Earth radii across its entire Hill region. This means we can now effectively exclude moons like Europa or Rhea from our search parameters, proving that the telescope is sensitive enough to detect objects nearly as small as our own Moon.

While we are looking for moons, we are also refining what we know about the giants in our own backyard. By using vertical mixing and chemical models to look at carbon monoxide in Uranus and Neptune, researchers found that Neptune is significantly more enriched in oxygen than Uranus.

This suggests these two ice giants likely followed very different evolutionary paths during their formation. The precision of our observations depends heavily on how well we calibrate our tools, a challenge highlighted by recent work on the Hayabusa2 mission.

After correcting for systematic offsets and the physical impact of touchdowns, new analysis of asteroid Ryugu shows that its eastern hemisphere is actually rougher and more covered in fine-grained dust than previously thought. This need for precision extends to how we interpret the light from distant, violent events like tidal disruption events.

New models show that as a supermassive black hole tears apart a star, the magnetic field can cause the resulting accretion disk to precess, creating predictable oscillations in X-ray and radio signals. These patterns could eventually allow us to measure black hole spin and magnetic flux directly from the timing of their flares.

The complexity of these environments is echoed in the study of supernovae, where observations of SN 2023ufx revealed a triple-peaked oxygen emission profile. This asymmetry suggests the explosion came from a massive, heavily stripped red supergiant in an extremely metal-poor galaxy.

Even when we look at more stable objects like white dwarfs, we are finding that our old ways of fitting models to data need an upgrade. By incorporating Gaia's precise distance measurements and using finer model grids, astronomers can finally untangle the "degenerate" solutions that used to make it impossible to tell if a star was one size or another.

This drive for clarity is also essential for the ongoing hunt for life in exoplanet atmospheres. A new, massive survey of ten different planets using Webb’s MIRI instrument suggests that the chemical signals we've seen in temperate sub-Neptunes are actual molecules rather than just instrumental noise.

While we haven't confirmed life yet, these results confirm that these specific types of planets are the most promising places to keep looking. We finally have a clearer picture of how small galaxies build their mass, which is vital for understanding if dark matter behaves as we expect in low-density environments.

Using DESI Data Release 1, researchers mapped the relationship between stellar and halo mass down to the dwarf galaxy scale without needing to guess based on larger galaxies. They found that while massive galaxies follow a predictable path, smaller ones are far more chaotic; the scatter in their mass relations jumps from 0.17 dex in Milky Way-sized systems to over 0.33 dex for LMC-like dwarfs, suggesting these tiny galaxies follow much more diverse evolutionary histories.

This complexity in small-scale structures is mirrored by the messy ways gas moves through them. In a study of nearby spiral galaxies, researchers found that dust-to-gas ratios increase with metallicity, but the ratio of dust to metals stays remarkably flat at about 30 percent.

This suggests that once a galaxy reaches a certain maturity, efficient grain growth in the interstellar medium creates a steady state where dust formation and destruction balance out. Looking further back in time, we are seeing how these processes shaped the very first massive structures.

A spectroscopic survey of a protocluster at redshift 3 discovered an unexpected, massive cloud of hydrogen gas offset from the cluster center by about 60 cMpc. This gas is surprisingly metal-rich and spans a huge velocity range, hinting that it might be part of a hidden protocluster or perhaps massive outflows driven by a single large galaxy.

The physics of how matter settles into these structures remains tricky to untangle, especially when things are spinning. When trying to map the stars orbiting Sagittarius A, scientists found that black hole spin and relativistic gravity effects are deeply tangled together in our current data.

To break this stalemate and actually measure the spin of the central black hole, we will likely need to observe multiple stars simultaneously to separate their individual orbits from the general curvature of spacetime. We are seeing some fascinating new ways to map the dark universe, particularly through how we use light from exploding stars.

By combining two massive datasets into a single unified sample called Unite, researchers have created the most comprehensive collection of Type Ia supernovae to date, totaling 2884 objects. This massive dataset is doing more than just refining our measurements; it is actually hinting at a potential tension in our cosmological models.

While the data shows some friction when trying to fit a constant dark energy model to both supernova and cosmic microwave background data, that tension seems to ease if we allow dark energy to evolve over time, suggesting the expansion of the universe might be more dynamic than we previously thought. This search for new physics in the cosmos is also being pushed into even more extreme scales through the study of microlensing.

Rather than looking for tiny black holes, scientists are proposing that we use cosmological supernovae to hunt for ultracompact minihalos—extended dark structures that could reveal secrets about the power spectrum of the early universe. On a much more local scale, we are getting a better look at how stars and galaxies interact to shape their evolution.

By looking at morphological disturbances in galaxy pairs, researchers have found that close encounters trigger intense bursts of star formation that peak during the encounter and leave behind a signature of intermediate-age stars long after the initial burst has faded. We are also seeing some high-stakes survival stories in distant solar systems.

Two sub-Neptunes, TOI-426 b and TOI-1839 b, have been confirmed orbiting very close to their stars, receiving massive amounts of radiation that should theoretically strip them of their atmospheres. Instead, they appear to be volatile-rich steam worlds, suggesting that these planets might be much more resilient to evaporation than our current models predict.

This resilience in planetary environments is mirrored by the complex chemistry we see in our own solar system's moons. New laboratory studies on Titan haze analogs show that while the specific way we make these particles in a lab can change their surface energy, the particles themselves are naturally quite cohesive and likely act as efficient seeds for hydrocarbon clouds.

Even within our own galaxy, we are gaining unprecedented clarity on its inhabitants through the new DESI Data Release 1. This release provides the largest spectroscopic catalogue of white dwarfs to date, giving us over 44,000 confirmed objects to study how these stellar remnants evolve.

Finally, as we prepare for massive surveys like Euclid, astronomers are working hard to ensure that "noise" from stars doesn't ruin our view of the distant universe. They have found that while imperfectly masking stars can create small-scale errors in our maps, the real danger comes from photometric persistence—a lingering light effect that could mimic large-scale cosmic structures if not carefully managed.

The tension in our current cosmological models might be much more localized than it looks. While many are debating whether dark energy is evolving, new model predictive scoring suggests that this preference for a w0waCDM model is almost entirely driven by a single Baryon Acoustic Oscillation data point at redshift 0.706.

This means the apparent tension isn't being fueled by supernova data, which actually supports the standard model, but rather by one specific observation in the BAO dataset. This uncertainty in our cosmic expansion history is compounded by new models suggesting dark energy might be an emergent phenomenon that only became active late in the universe's life.

When researchers tested scenarios where dark energy interacts with dark matter—specifically transferring energy from the former to the latter—they found that current data cannot yet pin down how fast this transition occurs. While these interactions can shift certain parameters like S8, they don't actually provide a better fit for the data than our standard cosmological constant model does.

Moving from the largest scales to the birth of galaxies, we are seeing how environments shape evolution in real time. In the core of a protocluster at redshift 2.2, observations from Hubble and JWST show that massive galaxies are undergoing rapid quenching about 500 million years before they were observed.

These galaxies appear to be running out of fuel, with gas fractions dropping below 7 percent as they transition into the red sequence. This depletion of gas is a fundamental part of the lifecycle in star-forming regions, much like how gravitational instability dictates the early stages of star formation itself.

We now understand that when a protoplanetary disc becomes massive enough, its own gravity creates large-scale spiral arms that can trap solids and potentially collapse into sub-stellar companions. This moves the theory of gravitational instability from a mere mathematical possibility to a framework we can actually test using high-resolution ALMA observations.

The most critical question for cosmologists right now is whether we need to rethink dark energy, and new Bayesian evidence suggests we might not need to change our fundamental models just yet. While some recent measurements have hinted at deviations from the standard cosmological model, a new comparison shows that when you account for prior-volume penalties, both early dark energy and late-time dynamical models are actually disfavored by the data.

This means that even if individual parameters look strange, the overall evidence still points toward the standard baseline. This tension in our understanding of cosmic evolution is mirrored by the difficulty of seeing what is actually out there, such as a newly discovered extremely low surface brightness galaxy candidate found in Rubin Observatory imaging.

This object, Rubin J122659.4+090236, is so diffuse that it could be either a local dwarf galaxy or a much more distant system, showcasing just how sensitive the upcoming LSST survey will be to these elusive structures. As we look closer at specific objects, we are finding that even the most massive systems have complex histories.

For example, observations of the blazar OJ 287 during its October 2022 flare support a model where a secondary black hole is orbiting a primary mass of roughly eighteen billion suns, with the flare likely caused by tidal effects rather than direct impact. Mapping these complex motions becomes easier as our catalogs grow, such as the new AGN-DB database which has compiled over six million confirmed active galactic nuclei from across the electromagnetic spectrum.

This massive scale allows us to move from studying single objects to understanding entire populations of black holes and their environments. We are also getting better at seeing through dust to measure the chemistry of distant worlds.

A new photometric method for M-dwarf microlensing lenses could soon allow us to measure metallicities for about 150 planetary systems in the Galactic Bulge using Roman Space Telescope data, providing a rare look at the composition of planets beyond the snow line. Even within a single galaxy, we are learning that different parts can have completely different origins.

By modeling the orbits and stellar populations of galaxies like LEDA 2220522, researchers have found counter-rotating disks that are younger and more metal-rich than the main body of the galaxy, suggesting they were formed by a fresh supply of gas rather than a violent merger. These same dynamical processes are at play in the early stages of solar systems as well.

New simulations show that if a protoplanetary disk disappears quickly enough, it can actually "freeze" planets into stable orbits, preventing them from being kicked out of resonance as the gas vanishes. Ultimately, all these observations require sophisticated ways to handle the data we collect.

A new guide for cosmological inference explains how researchers can choose between different compression methods—like MOPED or neural networks—to ensure they aren't losing vital information when they simplify massive datasets for analysis.

Today's papers

The papers

Important terms

Hubble tension
A major disagreement in astronomy regarding how fast the universe is expanding. Different measurement methods currently produce different values, leading scientists to search for new physics or better data to resolve the conflict.
Baryon fraction
The ratio of normal matter, like atoms and gas, to the total amount of matter in a cosmic structure. It helps scientists understand how much material stays within galaxy groups versus being pushed out by energy.
Dark energy
A mysterious force driving the accelerated expansion of the universe. Researchers are testing whether it stays constant or changes over time to see if it can explain current discrepancies in cosmological measurements.
Tidal disruption events
Occur when a supermassive black hole's gravity pulls apart a nearby star. This process creates an accretion disk that can emit predictable X-ray and radio signals, helping astronomers measure the black hole's properties.