Eleven new feedback tests, from converging galactic winds to a quasar halo denser than models predict (2026)

The week of September 28 to October 4, 2026 produced eleven new arXiv results on gas inside, leaving and returning to galaxy halos. That is everything the arXiv listing carried by Monday morning, which reaches submissions logged through Friday, October 2. The weekend's submissions are announced after this hour and were not screened.

Two of the eleven look at the same gas with different telescopes and reach the same verdict: the cold circumgalactic medium around quasars holds far denser clumps than cosmological simulations produce. The rest of the week is a wind simulation that finally reports convergence, an X-ray measurement that caps the supernova progenitor mass in a starburst nucleus, and three analyses that use the Lyman-α forest to bound how neutral the intergalactic medium was.

An item earns a place below when the original reports a quantitative result on gas inside, leaving or returning to a galaxy halo or the intergalactic medium, states a method another group can repeat, and links to a public source.

At a glance

Item	Design / method	Main result	Scale or tracer
Li et al., wind convergence 1	Hydrodynamic simulations of an M82-like starburst at 4, 8, 16 and 32 pc, with star formation and supernova feedback re-tuned to a physical control volume and cooling radius	Stellar mass and kinetic energy converge to about 20 per cent across 4–32 pc; errors in the cool-phase outflow rate and X-ray luminosity fall below about 40 per cent, against about 50 and 75 per cent for the warm and hot phases	A starburst disc out to its halo; cool, warm and hot outflow phases
Attard et al., MEGATRON 2	Four star-formation and stellar-feedback prescriptions evolved from identical initial conditions in the MEGATRON zoom suite	The prescription sets how often a dwarf goes quiet; stronger or more clustered feedback quenches more dwarfs, and duty cycles at fixed mass differ by factors of several between models	M* ≈ 10⁴–10⁸ M☉ at z = 8.5
Hervella-Seoane et al., low-power jets 3	Six three-dimensional relativistic-hydrodynamic simulations of 10⁴⁴–10⁴⁵ erg s⁻¹ FRI-like jets in intracluster medium extracted from cosmological GADGET-3 runs	About 80 per cent of the injected kinetic power becomes internal energy of the ambient medium; the jets decelerate to Mach ≈ 2.3 and heat mostly through weak bow shocks	20 Myr of jet activity; the inner tens of kpc of clusters
Liu et al., NGC 3079 4	XMM-Newton RGS and EPIC spectroscopy of the nuclear superbubble and the galactic-scale superbubble	O/Fe = 0.47 and Ne/Fe = 0.77 in solar units imply an upper progenitor-mass cut-off of 13–15 M☉ for pure core-collapse enrichment; superbubble ages of 0.92 and 29.6 Myr point to episodic feedback	Central 4 kpc out to 25 kpc; diffuse X-ray gas
Genc et al., KiDS-DR5 5	Joint modelling of multiband magnitude shifts and galaxy–galaxy lensing for a mass-selected sample, validated on the MICE2 simulation	Circumgalactic dust mass scales linearly with stellar mass but more flatly than the Dusty-SAGE semi-analytic prediction, which the authors read as a gap in dust-related feedback	Lens galaxies below photometric redshift 0.4, in five stellar-mass bins
Lopez et al., ALPINE-CRISTAL-JWST 6	Spatially resolved (~kpc) [C II] and dust-continuum ALMA data combined with JWST/NIRSpec IFU spectroscopy	Galaxies carrying outflow signatures have lower dust-to-stellar surface density ratios and lower gas mass fractions, consistent with gas and dust expelled into the circumgalactic medium	kpc scales in galaxies at z ≈ 4–5
Travascio et al., quasar CGM density 7	CLOUDY photoionisation models of log-normal density distributions compared with stacked Keck/KCWI, Keck/MOSFIRE and VLT/MUSE observations	Stacked ratios Lyα/Hα = 9.1, HeII/Hα = 0.54 and HeII/Lyα = 0.06; single-density models fall short of the HeII/Hα ratio, and broad log-normal distributions need a high-density tail corresponding to a clumping factor of 10³–10⁴	Out to about 35 pkpc around quasars at z ≈ 2.1–4.5
Lazeyras et al., RePhyNe II 8	MUSE observations of quasar nebulae at z ≈ 3.8 compared with mock observations from the DaLya suite, COLIBRE and HELLO	Observed Lyα surface brightness covers a much larger fraction of the field than the mocks; the gas must reach at least 10 cm⁻³ between 50 and 200 ckpc, while simulated distributions have medians near 10⁻² cm⁻³ and clumping factors of 16–102	50–200 ckpc around quasars at z ≈ 3.8
Kundu et al., Lyα fractions 9	JWST Lyα-fraction measurements at z ≈ 6–14 compared with Lyα transmission from the THESAN simulations and a flexible semi-analytic model	THESAN's reionisation history and bubble sizes reproduce the measurements with a reduced chi-squared of 0.4–1.2, giving 1σ bounds on the mass-weighted neutral fraction of 0.38 (+0.14/−0.38) and 0.52 (+0.09/−0.21) at z ≈ 7, 0.57 (+0.23/−0.57) and 0.68 (+0.12/−0.39) at z ≈ 8, and 0.93 (+0.05/−0.17) and 0.89 (+0.11/−0.15) at z ≈ 11	IGM neutral fraction across z ≈ 6–14
Helton et al., Lyman-α forest at z ≈ 7.57 10	JWST/NIRSpec PRISM spectroscopy of JADES-GS-z7-LAF, confirmed with JWST/NIRCam F090W imaging	Flux blueward of Lyα is detected at about 7σ in the spectrum and about 10σ in the image, implying a transmitted sightline with x_HI ≪ 0.1 per cent through z ≈ 5.5–7.5	One massive (M* ≈ 10⁹·¹ M☉) galaxy at z ≈ 7.57
Shimizu et al., IGM transmission 11	JWST/NIRCam F470N imaging of an Hα-emitter overdensity in CEERS, with Subaru/HSC NB872 Lyα fluxes	The Lyα-emitter fraction is 0.41 (+0.15/−0.14) within 2 arcmin of the density centre against 0.27 (+0.19/−0.15) outside it, and stacked Lyα escape fractions are 0.100 (+0.091/−0.053) and 0.063 (+0.130/−0.063)	z ≈ 6.2; IGM transmission across an overdense region

Paper briefs

Galactic winds converge once the sub-grid model is rebuilt around physical scales

Authors and date. Xue-Fu Li, Weishan Zhu, Tian-Rui Wang, Huan Chen and Long-Long Feng, submitted September 28, 2026; accepted for publication in the Astrophysical Journal. 1

Study design. The authors simulate an M82-like low-mass starburst galaxy at resolutions of 4, 8, 16 and 32 pc and rebuild three sub-grid prescriptions to reduce their dependence on the grid. Star formation is fed by gas accretion onto sink particles inside a fixed control volume of about 20 pc, close to the size of a giant molecular cloud, and uses a resolution- and density-dependent efficiency. Supernova feedback injects its energy over the physical cooling radius of the remnant, and splits that energy between thermal and kinetic channels according to resolution. 1

Result. Across 4 to 32 pc the total stellar mass formed and the total kinetic energy vary by about 20 per cent, and the relative error in the cool-phase mass outflow rate and in the integrated X-ray luminosity falls below about 40 per cent. The warm and hot outflow phases converge less tightly, at about 50 and 75 per cent respectively. The upgraded model also improves convergence in the mass-loading factor and in the radial structure of the outflow. 1

Why read. Every galactic-wind prediction carries a resolution dependence, and this paper quantifies what that dependence is for one well-controlled galaxy. The cool phase, which most absorption and emission measurements actually trace, is the phase that converges best here. A reader comparing simulations at different resolutions can use the residual errors as the floor on any wind rate they compare to data. 1

Caveat. One galaxy and one feedback implementation carry the result. The convergence floor is set by the re-tuned prescriptions rather than derived, and the warm and hot phases, which carry most of the outflow's energy, remain the least converged. 1

The feedback prescription, not the halo, sets how often a dwarf goes quiet

Authors and date. David Attard, Ilian T. Iliev, Martin P. Rey, Harley Katz, Corentin Cadiou and Nicholas Choustikov, submitted September 29, 2026; preprint. 2

Study design. MEGATRON is a suite of high-redshift zoom simulations in which four star-formation and stellar-feedback prescriptions are run from identical initial conditions of the same haloes. The authors classify galaxies between 10⁴ and 10⁸ M☉ in stellar mass at z = 8.5 as star-forming, mini-quenched (no significant star formation in the last 10 Myr) or quenched (none in the last 100 Myr), and compare how each class depends on mass between models. 2

Result. The feedback prescription controls both how often a galaxy is quiescent and how that fraction depends on mass: models with stronger or more clustered feedback produce larger quiescent fractions than the fiducial and varying-IMF models, and duty cycles at fixed mass differ by factors of several. The cold-gas profiles of quenched galaxies split into two distinct modes, mechanical evacuation that leaves compact gas-poor systems and thermodynamic suppression that leaves a cold but inactive reservoir. Quiescent dwarfs at these masses fall below un-lensed JWST/NIRCam detection limits while star-forming dwarfs of the same mass do not, so flux-limited samples select the active phases. 2

Why read. Reionisation-era dwarf quenching is a regime where stellar feedback is the only plausible driver, and this suite isolates the feedback choice as the free parameter. The two quenching modes give observers a way to separate systems that lost their gas from systems that kept it, and the detection-limit argument explains why a complete census is hard from imaging alone. 2

Caveat. Four prescriptions, four outcomes, from the same initial conditions, so the suite maps a modelling uncertainty rather than settling it. The classifications rest on fixed 10 and 100 Myr time windows, and the detection-limit statement is a prediction about JWST rather than a measured completeness. 2

Low-power jets turn four fifths of their energy into heat

Authors and date. K. Hervella-Seoane, M. Perucho, S. Planelles, J. M. Martí and V. Quilis, submitted September 30, 2026; accepted in Astronomy & Astrophysics, doi:10.1051/0004-6361/202659737. 3

Study design. Six three-dimensional relativistic-hydrodynamic simulations follow FRI-like jets of 10⁴⁴–10⁴⁵ erg s⁻¹ through the inner regions of realistic intracluster medium. The environments are extracted from cosmological GADGET-3 runs rather than built from an idealised profile, and each run tracks 20 Myr of continuous jet activity with radiative cooling switched on. 3

Result. The jets decelerate quickly from mildly relativistic injection speeds to trans-sonic flow at an average Mach number of about 2.3, and weak bow shocks dominate the energy transfer. Roughly 80 per cent of the injected kinetic power is converted into internal energy of the ambient medium. Bremsstrahlung cooling measurably lowers the temperature of the expanding shocked shells. 3

Why read. The cooling-flow problem is usually posed as a question about powerful radio sources, and this paper supplies the heating efficiency for the low-power jets that actually dominate the AGN population. A coupling efficiency near 80 per cent, with weak shocks behind it, is the number a feedback model needs when it asks whether the lowest-power jets can hold a cool core in balance. 3

Caveat. The simulations cover the first 20 Myr of activity and the inner tens of kiloparsecs, so they say nothing about the long-term balance of a cluster or about the outer halo. The environments come from cosmological runs, but the jets themselves are injected in post-processing, and magnetic fields are left out. 3

NGC 3079's superbubble caps the supernova progenitor mass

Authors and date. Jiejia Liu, Junjie Mao, Shuinai Zhang, Guan-Fu Liu, Rui Huang and Wei Cui, submitted September 29, 2026; accepted in Astronomy & Astrophysics. 4

Study design. The authors model the diffuse X-ray emission of the starburst galaxy NGC 3079 with XMM-Newton. The reflection grating spectrometer covers the central 4 kpc, which contains a nuclear superbubble about 1 kpc across, and EPIC covers the galactic-scale superbubble out to about 25 kpc. They fit multi-temperature plasma models together with a charge-exchange component. 4

Result. Charge exchange contributes about 13 per cent of the total energy flux in the 0.2–2 keV band, so models that leave it out will bias the abundances. The best-fit abundance ratios are O/Fe = 0.47 and Ne/Fe = 0.77 in solar units. Compared with IMF-weighted core-collapse supernova yields, they imply an upper progenitor-mass cut-off of 13–15 M☉ for pure core-collapse enrichment. Adding a supernova Type Ia contribution raises the inferred cut-off, and its 1σ upper limit stays below 25 M☉ for Type Ia fractions up to 10 per cent. The characteristic ages of the nuclear and galactic-scale superbubbles are 0.92 (+0.39/−0.46) Myr and 29.6 (+1.6/−2.9) Myr. 4

Why read. Chemical enrichment of the halo depends on which stars explode, and the alpha-element-to-iron ratios in a starburst's hot gas are one of the few direct handles on the upper mass limit of the core-collapse population. The two superbubble ages, an order of magnitude apart, argue for episodic feedback in this galaxy. 4

Caveat. The progenitor-mass limit follows from comparing measured ratios with a chosen yield table, so it moves with the choice of yields and with the assumed Type Ia fraction. One galaxy and one pair of sightlines carry the abundance measurement. 4

Circumgalactic dust tracks stellar mass, more flatly than the models

Authors and date. Eray Genc, Angus H. Wright, Hendrik Hildebrandt, Dian P. Triani, Dominik J. Bomans, Benjamin Stölzner and nine co-authors, submitted September 28, 2026; submitted to Astronomy & Astrophysics. 5

Study design. The authors measure how much dust sits in the halos of galaxies, using a sample drawn from the final data release of the Kilo-Degree Survey with magnitude, photometric-redshift and stellar-mass cuts. They jointly model the multiband magnitude shifts that dust extinction produces and the galaxy–galaxy lensing signal that fixes the halo mass, and they validate the framework on the MICE2 simulation with a realistic circumgalactic dust model. 5

Result. The data require both a satellite fraction and dust extinction contributed by neighbouring halos. The circumgalactic dust mass scales linearly with stellar mass, in line with earlier observational work, but the slope is flatter than the Dusty-SAGE semi-analytic model predicts. The authors read that gap as evidence that dust-related feedback in galaxy formation models needs further refinement. 5

Why read. Dust in the halo is one of the few cheap tracers of how far outflows carry material away from a galaxy, and this measurement turns thousands of low-redshift lenses into a stellar-mass–dust-mass relation. The comparison against a semi-analytic model points at a concrete quantity for modellers to match. 5

Caveat. The dust mass is inferred through a model rather than observed directly, and the inference depends on the assumed extinction law and on the treatment of neighbouring halos. The comparison against Dusty-SAGE rests on one semi-analytic implementation. 5

Galaxies with outflows are the ones missing gas and dust

Authors and date. F. Lopez, M. Relaño, I. De Looze, A. L. Faisst, R. Herrera-Camus, R. A. Jorgenson and 34 co-authors, submitted September 29, 2026; accepted in Astronomy & Astrophysics. 6

Study design. The survey combines ALMA observations of rest-frame [C II] and dust continuum with JWST/NIRSpec integral-field spectroscopy, resolving galaxies at z ≈ 4–5 on kiloparsec scales. The authors map atomic and molecular gas masses, dust and stellar mass surface densities, star formation rates and metallicity across each galaxy. 6

Result. Oxygen abundances fall between 12 + log(O/H) ≈ 8.1 and 8.3, varying by up to about 1 dex within a single galaxy. The calibration used to convert [C II] into atomic gas mass overestimates the atomic gas surface density by one to two orders of magnitude, and a spatially resolved conversion factor does better. Galaxies showing strong outflow signatures have lower dust-to-stellar surface density ratios and lower gas mass fractions than galaxies without them, which the authors read as gas and dust expelled into the circumgalactic medium, and as removal of dust by stellar feedback or radiation pressure. 6

Why read. The link between outflow signatures and a depleted dust and gas reservoir at kiloparsec resolution is the observational counterpart of the ejection assumed in nearly every feedback model. The [C II] calibration warning matters independently, because a one- to two-order-of-magnitude overestimate of atomic gas mass propagates straight into mass-loading factors. 6

Caveat. The sample spans z ≈ 4–5, and the comparison between outflow hosts and non-hosts is an association; the direction of cause remains open. The gas masses also depend on the conversion factor the authors are warning about. 6

The quasar halo's cold gas needs a broad, clumpy density distribution

Authors and date. Andrea Travascio, Sebastiano Cantalupo, Titouan Lazeyras, Naveen A. Reddy, Charles C. Steidel, Yongming Liang and 13 co-authors, submitted October 1, 2026; submitted to Astronomy & Astrophysics. 7

Study design. Extended Lyα emission around quasars traces cold circumgalactic gas, but it is a resonant line and therefore hard to convert into a density. The authors use CLOUDY photoionisation models to predict the HeII/Hα ratio for quasar-illuminated gas over a wide parameter range, including log-normal density distributions, and compare those predictions with observations. The observational sample is 13 quasars at z ≈ 2.2 observed with Keck/KCWI, with the strongest emission followed up in Keck/MOSFIRE, plus 26 VLT/MUSE quasars at 3.1 < z < 4.5. 7

Result. Inside matched apertures out to about 35 pkpc, the stacked spectra give Lyα/Hα = 9.1, HeII/Hα = 0.54 and HeII/Lyα = 0.06. A single-density model reproduces the low HeII/Hα ratio only with densities above 10 cm⁻³ or with an implausibly soft AGN spectrum. Broad log-normal density distributions fit at lower median densities, but they need a substantial high-density tail, formally corresponding to a clumping factor of 10³ to 10⁴ in the authors' parameterisation. In the MUSE sample, 14 of 26 systems show co-spatial emission in the relevant lines. 7

Why read. The clumping of cold halo gas sets how efficiently a quasar's radiation couples to its halo, and how much of the emission is fluorescence rather than recombination. This paper converts a line ratio into a constraint on the shape of the density distribution rather than a single number, which is what a simulation has to reproduce. 7

Caveat. The clumping factor is a parameter in the authors' log-normal description rather than a measured quantity, so its numeric value depends on that parameterisation. The stacks average over sightlines, and quasar haloes are a selected population. 7

Quasar nebulae come out denser and clumpier than three simulation suites

Authors and date. Titouan Lazeyras, Sebastiano Cantalupo, Andrea Travascio, Nicolas Ledos, Joop Schaye, Gabriele Pezzulli and seven co-authors, submitted October 1, 2026; submitted to Astronomy & Astrophysics. 8

Study design. The study compares MUSE observations of quasar Lyα nebulae at z ≈ 3.8 with mock observations generated from simulations under the assumption that the emission is maximally fluorescent. Three simulation sources supply the mocks: the new DaLya suite, COLIBRE at its m5 and m6 resolutions, and HELLO. The observable compared between them is the distribution of Lyα surface brightness, which under that assumption maps onto the cold-gas density distribution. 8

Result. The observations show a much larger covering fraction of high Lyα surface brightness than the simulations, with the single exception of the higher-resolution COLIBRE m5 run. Reproducing the observed surface brightness requires densities of at least 10 cm⁻³ in the quasar circumgalactic medium between 50 and 200 ckpc. The simulated cold-gas probability density functions are skewed log-normals with medians near 10⁻² cm⁻³ and clumping factors between 16 and 102, so the simulations fall well short of the required density. 8

Why read. This is a simulation-versus-observation test where both sides are computed on the same observable, and it locates a specific failure: the cold phase of the circumgalactic medium around quasars is missing from the models. The result points the same way as the independent line-ratio constraint published the same week. 8

Caveat. The whole comparison rests on the maximal-fluorescence assumption, so any contribution from recombination or from collisional excitation would weaken the inferred density requirement. Only one of the simulation suites, at its higher resolution, comes close to the data, and quasar-hosted haloes are a selected population. 8

Lyα fractions put the neutral fraction at 38 per cent at z ≈ 7

Authors and date. Aritra Kundu, Adam Lidz and Guochao Sun, submitted September 30, 2026; submitted to the Astrophysical Journal. 9

Study design. JWST measurements now give the fraction of ultraviolet-selected galaxies that show a detectable Lyα line between z ≈ 6 and z ≈ 14. Because Lyα scatters off neutral hydrogen in the surrounding intergalactic medium, that fraction carries information about the reionisation history and about the sizes of ionised bubbles. The authors compare the measurements with Lyα transmission calculated from the THESAN reionisation simulations and from a flexible semi-analytic model, calibrating the intrinsic Lyα emission emerging from the interstellar and circumgalactic medium against empirical equivalent-width distributions at z ≈ 6. 9

Result. THESAN's reionisation history and bubble sizes agree with the current JWST Lyα fractions, with a reduced chi-squared between 0.4 and 1.2. The semi-analytic transmission calculations agree with THESAN's, which lets the authors scan a wider family of reionisation models and place 1σ bounds on the average mass-weighted neutral fraction. Those bounds are 0.38 (+0.14/−0.38) for THESAN and 0.52 (+0.09/−0.21) for the semi-analytic fit at z ≈ 7, 0.57 (+0.23/−0.57) and 0.68 (+0.12/−0.39) at z ≈ 8, and 0.93 (+0.05/−0.17) and 0.89 (+0.11/−0.15) at z ≈ 11. 9

Why read. Lyα visibility is measured far more cheaply than the Lyα forest at these redshifts, so a model that converts the observed fraction into a neutral fraction widens the set of sightlines that can be used. The paper also reports the two model families side by side, so the reader can see how much of each bound comes from the data and how much from the model. 9

Caveat. Both the intrinsic Lyα emission and its transmission are modelled, and the calibration of the intrinsic emission leans on a single redshift. The bounds are quoted at three redshifts, and the two model families give noticeably different central values at z ≈ 7. 9

A single galaxy at z ≈ 7.57 looks through an almost fully ionised sightline

Authors and date. Jakob M. Helton, Joris Witstok, Marta Laska, Kevin N. Hainline, Nikko J. Cleri, Francesco D'Eugenio and 12 co-authors, submitted September 29, 2026; to be submitted. 10

Study design. At z ≈ 7.57 the intergalactic medium is expected to absorb essentially all flux blueward of Lyα, so a detection of that flux would mark an unusually transparent sightline. The authors report flux blueward of Lyα from the massive, star-forming galaxy JADES-GS-z7-LAF using JWST/NIRSpec PRISM spectroscopy, and check the same signal in JWST/NIRCam F090W imaging. 10

Result. The blueward flux is detected at about 7σ in the spectrum and about 10σ in the image, and the imaged flux is spatially coincident with the galaxy, emerging about 0.4 proper kpc from its ultraviolet centroid at a rest wavelength near 1500 Å. The authors find no evidence for a foreground interloper, and they note that they cannot formally reject one. The transmitted flux implies a neutral fraction well below 0.1 per cent along the sightline between z ≈ 5.5 and 7.5, and the sightline coincides spatially and kinematically with several galaxy overdensities. The turnover at the Lyα break is consistent either with an extreme damping wing at a neutral column of about 10²³·³ cm⁻² or with a strong nebular continuum contributing some 60 to 70 per cent of the flux. 10

Why read. Cosmic reionisation is expected to end somewhere near this redshift, and a sightline this transparent at z ≈ 7.57 is a direct look at the ionised bubbles that models put around galaxy overdensities. Whatever happens to the interloper question, the measurement sets the transparency a real sightline can reach. 10

Caveat. One galaxy and one sightline carry the result, and the authors state that a foreground interloper remains possible with the present data. The two readings of the turnover, a damping wing and a nebular continuum, carry different implications and stay degenerate for now. 10

Lyα survives more often inside a z ≈ 6.2 galaxy overdensity

Authors and date. Shunta Shimizu, Nobunari Kashikawa, Junya Arita, Ryo Emori, Kohei Inayoshi, Akio K. Inoue and ten co-authors, submitted September 28, 2026; submitted to the Astrophysical Journal. 11

Study design. Lyα visibility during reionisation depends on the emitting galaxy and on how much neutral hydrogen the photons cross. The authors identify a projected overdensity of Hα emitters at z ≈ 6.2 in the CEERS field from JWST/NIRCam F470N imaging, then measure Lyα fluxes with Subaru/HSC NB872 imaging and compare galaxies within 2 arcmin of the density centre against those outside it. 11

Result. The Lyα-emitter fraction is 0.41 (+0.15/−0.14) in the inner region and 0.27 (+0.19/−0.15) in the outer region, and the stacked Lyα escape fractions are 0.100 (+0.091/−0.053) and 0.063 (+0.130/−0.063). Both measures point to higher Lyα visibility inside the overdensity, at modest significance. Galaxies nearer the centre are systematically more massive with redder ultraviolet slopes, which on its own would predict lower escape fractions, so the environmental difference resists a purely stellar-population explanation. The authors read it as reduced Lyα attenuation through a more ionised intergalactic medium, as expected if large ionised bubbles form preferentially around such overdensities. 11

Why read. If ionised bubbles are larger around overdensities, Lyα visibility should vary with environment, and this paper measures that variation on the same galaxies with both an emission-line count and a stacked escape fraction. The two measures agree in direction, and the stellar-population trend pushes the other way, which makes the environmental signal harder to dismiss as a selection effect. 11

Caveat. The inner and outer Lyα-emitter fractions differ by less than 1σ, and the escape fractions carry errors large enough to cover the gap, so the environmental signal is suggestive rather than established. One field and one overdensity carry the measurement. 11

Code and data releases

Both release categories came back empty for this window. No public code release tied to circumgalactic or intergalactic feedback physics appeared among the week's papers: the abstracts and comment fields of the eleven items above, and of the wider astro-ph listing screened for the five days, carry no new public repository, package or tool announcement. Repository sweeps over the same window returned one already-covered feed-forward pipeline and one gallery of halo visualisations, and neither is a new release.

No separately deposited, openly downloadable dataset on circumgalactic or intergalactic feedback physics appeared either. The Zenodo astronomy records dated inside the window hold conference material, software from other fields, and asteroseismology data products, and nothing on CGM or IGM feedback. The most recent relevant deposition remains the DESI green-valley cool circumgalactic medium data release, which is embargoed until April 1, 2027 and still carries no public files.

What the week separates

The cold CGM comes out denser than the simulations, by two independent routes. Lazeyras et al. compare MUSE quasar nebulae with mock observations from three simulation suites and find that the observed Lyα surface brightness needs at least 10 cm⁻³ between 50 and 200 ckpc, against simulated medians near 10⁻² cm⁻³ 8. Travascio et al. reach the same place from line ratios rather than surface brightness, needing a high-density tail with a clumping factor of 10³ to 10⁴ 7. The two figures measure different things, a threshold density against a parameter in a log-normal description, so the agreement is one of direction rather than of value. Both rest on the fluorescence assumption, and both leave the density itself inferred.
Sub-grid choices set the answer in both directions. Li et al. rebuild star formation and supernova coupling around physical scales and pull the cool-phase outflow rate into a convergence band of 20 to 75 per cent across a factor of eight in resolution 1. Attard et al. run four feedback prescriptions from identical initial conditions and find that the quiescent fraction and the duty cycle of dwarf galaxies move by factors of several between them 2. The first result shows how much of a discrepancy is numerical and can be removed; the second shows how much remains once it has been, because it comes from the feedback model itself.
The heating budget holds up at both ends of the feedback scale. Hervella-Seoane et al. find that FRI-like jets convert about 80 per cent of their injected kinetic power into internal energy of the ambient medium, mostly through weak bow shocks at Mach ≈ 2.3 3. Liu et al. use abundance ratios in NGC 3079's hot gas to cap the mass of the supernovae supplying those metals at 13–15 M☉, and date its two superbubbles 30 times apart 4. The first supplies a coupling efficiency for the black hole channel; the second supplies the yield inventory for the stellar one.
Two sightlines agree that reionisation was patchy, one of them unusually so. Kundu et al. fit the JWST Lyα fractions with THESAN and with a semi-analytic model and place the neutral fraction near 0.38 to 0.52 at z ≈ 7, rising to roughly 0.9 at z ≈ 11 9. Helton et al. then report a sightline at z ≈ 7.57 with a neutral fraction below 0.1 per cent, two orders of magnitude below that average 10. Shimizu et al. supply the middle case, with Lyα surviving more often inside a z ≈ 6.2 overdensity than outside it 11. A mean neutral fraction of that size and a single fully transparent sightline fit together only if the ionised regions around overdensities are large, which is the picture the third result measures directly.

References

At a glance

Item	Design / method	Main result	Scale or tracer
Li et al., wind convergence 1	Hydrodynamic simulations of an M82-like starburst at 4, 8, 16 and 32 pc, with star formation and supernova feedback re-tuned to a physical control volume and cooling radius	Stellar mass and kinetic energy converge to about 20 per cent across 4–32 pc; errors in the cool-phase outflow rate and X-ray luminosity fall below about 40 per cent, against about 50 and 75 per cent for the warm and hot phases	A starburst disc out to its halo; cool, warm and hot outflow phases
Attard et al., MEGATRON 2	Four star-formation and stellar-feedback prescriptions evolved from identical initial conditions in the MEGATRON zoom suite	The prescription sets how often a dwarf goes quiet; stronger or more clustered feedback quenches more dwarfs, and duty cycles at fixed mass differ by factors of several between models	M* ≈ 10⁴–10⁸ M☉ at z = 8.5
Hervella-Seoane et al., low-power jets 3	Six three-dimensional relativistic-hydrodynamic simulations of 10⁴⁴–10⁴⁵ erg s⁻¹ FRI-like jets in intracluster medium extracted from cosmological GADGET-3 runs	About 80 per cent of the injected kinetic power becomes internal energy of the ambient medium; the jets decelerate to Mach ≈ 2.3 and heat mostly through weak bow shocks	20 Myr of jet activity; the inner tens of kpc of clusters
Liu et al., NGC 3079 4	XMM-Newton RGS and EPIC spectroscopy of the nuclear superbubble and the galactic-scale superbubble	O/Fe = 0.47 and Ne/Fe = 0.77 in solar units imply an upper progenitor-mass cut-off of 13–15 M☉ for pure core-collapse enrichment; superbubble ages of 0.92 and 29.6 Myr point to episodic feedback	Central 4 kpc out to 25 kpc; diffuse X-ray gas
Genc et al., KiDS-DR5 5	Joint modelling of multiband magnitude shifts and galaxy–galaxy lensing for a mass-selected sample, validated on the MICE2 simulation	Circumgalactic dust mass scales linearly with stellar mass but more flatly than the Dusty-SAGE semi-analytic prediction, which the authors read as a gap in dust-related feedback	Lens galaxies below photometric redshift 0.4, in five stellar-mass bins
Lopez et al., ALPINE-CRISTAL-JWST 6	Spatially resolved (~kpc) [C II] and dust-continuum ALMA data combined with JWST/NIRSpec IFU spectroscopy	Galaxies carrying outflow signatures have lower dust-to-stellar surface density ratios and lower gas mass fractions, consistent with gas and dust expelled into the circumgalactic medium	kpc scales in galaxies at z ≈ 4–5
Travascio et al., quasar CGM density 7	CLOUDY photoionisation models of log-normal density distributions compared with stacked Keck/KCWI, Keck/MOSFIRE and VLT/MUSE observations	Stacked ratios Lyα/Hα = 9.1, HeII/Hα = 0.54 and HeII/Lyα = 0.06; single-density models fall short of the HeII/Hα ratio, and broad log-normal distributions need a high-density tail corresponding to a clumping factor of 10³–10⁴	Out to about 35 pkpc around quasars at z ≈ 2.1–4.5
Lazeyras et al., RePhyNe II 8	MUSE observations of quasar nebulae at z ≈ 3.8 compared with mock observations from the DaLya suite, COLIBRE and HELLO	Observed Lyα surface brightness covers a much larger fraction of the field than the mocks; the gas must reach at least 10 cm⁻³ between 50 and 200 ckpc, while simulated distributions have medians near 10⁻² cm⁻³ and clumping factors of 16–102	50–200 ckpc around quasars at z ≈ 3.8
Kundu et al., Lyα fractions 9	JWST Lyα-fraction measurements at z ≈ 6–14 compared with Lyα transmission from the THESAN simulations and a flexible semi-analytic model	THESAN's reionisation history and bubble sizes reproduce the measurements with a reduced chi-squared of 0.4–1.2, giving 1σ bounds on the mass-weighted neutral fraction of 0.38 (+0.14/−0.38) and 0.52 (+0.09/−0.21) at z ≈ 7, 0.57 (+0.23/−0.57) and 0.68 (+0.12/−0.39) at z ≈ 8, and 0.93 (+0.05/−0.17) and 0.89 (+0.11/−0.15) at z ≈ 11	IGM neutral fraction across z ≈ 6–14
Helton et al., Lyman-α forest at z ≈ 7.57 10	JWST/NIRSpec PRISM spectroscopy of JADES-GS-z7-LAF, confirmed with JWST/NIRCam F090W imaging	Flux blueward of Lyα is detected at about 7σ in the spectrum and about 10σ in the image, implying a transmitted sightline with x_HI ≪ 0.1 per cent through z ≈ 5.5–7.5	One massive (M* ≈ 10⁹·¹ M☉) galaxy at z ≈ 7.57
Shimizu et al., IGM transmission 11	JWST/NIRCam F470N imaging of an Hα-emitter overdensity in CEERS, with Subaru/HSC NB872 Lyα fluxes	The Lyα-emitter fraction is 0.41 (+0.15/−0.14) within 2 arcmin of the density centre against 0.27 (+0.19/−0.15) outside it, and stacked Lyα escape fractions are 0.100 (+0.091/−0.053) and 0.063 (+0.130/−0.063)	z ≈ 6.2; IGM transmission across an overdense region

Paper briefs

Galactic winds converge once the sub-grid model is rebuilt around physical scales

Authors and date. Xue-Fu Li, Weishan Zhu, Tian-Rui Wang, Huan Chen and Long-Long Feng, submitted September 28, 2026; accepted for publication in the Astrophysical Journal. 1

The feedback prescription, not the halo, sets how often a dwarf goes quiet

Authors and date. David Attard, Ilian T. Iliev, Martin P. Rey, Harley Katz, Corentin Cadiou and Nicholas Choustikov, submitted September 29, 2026; preprint. 2

Low-power jets turn four fifths of their energy into heat

Authors and date. K. Hervella-Seoane, M. Perucho, S. Planelles, J. M. Martí and V. Quilis, submitted September 30, 2026; accepted in Astronomy & Astrophysics, doi:10.1051/0004-6361/202659737. 3

NGC 3079's superbubble caps the supernova progenitor mass

Authors and date. Jiejia Liu, Junjie Mao, Shuinai Zhang, Guan-Fu Liu, Rui Huang and Wei Cui, submitted September 29, 2026; accepted in Astronomy & Astrophysics. 4

Circumgalactic dust tracks stellar mass, more flatly than the models

Galaxies with outflows are the ones missing gas and dust

Authors and date. F. Lopez, M. Relaño, I. De Looze, A. L. Faisst, R. Herrera-Camus, R. A. Jorgenson and 34 co-authors, submitted September 29, 2026; accepted in Astronomy & Astrophysics. 6

The quasar halo's cold gas needs a broad, clumpy density distribution

Quasar nebulae come out denser and clumpier than three simulation suites

Lyα fractions put the neutral fraction at 38 per cent at z ≈ 7

Authors and date. Aritra Kundu, Adam Lidz and Guochao Sun, submitted September 30, 2026; submitted to the Astrophysical Journal. 9

A single galaxy at z ≈ 7.57 looks through an almost fully ionised sightline

Authors and date. Jakob M. Helton, Joris Witstok, Marta Laska, Kevin N. Hainline, Nikko J. Cleri, Francesco D'Eugenio and 12 co-authors, submitted September 29, 2026; to be submitted. 10

Lyα survives more often inside a z ≈ 6.2 galaxy overdensity

Code and data releases

What the week separates

The cold CGM comes out denser than the simulations, by two independent routes. Lazeyras et al. compare MUSE quasar nebulae with mock observations from three simulation suites and find that the observed Lyα surface brightness needs at least 10 cm⁻³ between 50 and 200 ckpc, against simulated medians near 10⁻² cm⁻³ 8. Travascio et al. reach the same place from line ratios rather than surface brightness, needing a high-density tail with a clumping factor of 10³ to 10⁴ 7. The two figures measure different things, a threshold density against a parameter in a log-normal description, so the agreement is one of direction rather than of value. Both rest on the fluorescence assumption, and both leave the density itself inferred.
Sub-grid choices set the answer in both directions. Li et al. rebuild star formation and supernova coupling around physical scales and pull the cool-phase outflow rate into a convergence band of 20 to 75 per cent across a factor of eight in resolution 1. Attard et al. run four feedback prescriptions from identical initial conditions and find that the quiescent fraction and the duty cycle of dwarf galaxies move by factors of several between them 2. The first result shows how much of a discrepancy is numerical and can be removed; the second shows how much remains once it has been, because it comes from the feedback model itself.
The heating budget holds up at both ends of the feedback scale. Hervella-Seoane et al. find that FRI-like jets convert about 80 per cent of their injected kinetic power into internal energy of the ambient medium, mostly through weak bow shocks at Mach ≈ 2.3 3. Liu et al. use abundance ratios in NGC 3079's hot gas to cap the mass of the supernovae supplying those metals at 13–15 M☉, and date its two superbubbles 30 times apart 4. The first supplies a coupling efficiency for the black hole channel; the second supplies the yield inventory for the stellar one.
Two sightlines agree that reionisation was patchy, one of them unusually so. Kundu et al. fit the JWST Lyα fractions with THESAN and with a semi-analytic model and place the neutral fraction near 0.38 to 0.52 at z ≈ 7, rising to roughly 0.9 at z ≈ 11 9. Helton et al. then report a sightline at z ≈ 7.57 with a neutral fraction below 0.1 per cent, two orders of magnitude below that average 10. Shimizu et al. supply the middle case, with Lyα surviving more often inside a z ≈ 6.2 overdensity than outside it 11. A mean neutral fraction of that size and a single fully transparent sightline fit together only if the ionised regions around overdensities are large, which is the picture the third result measures directly.

Eleven new feedback tests, from converging galactic winds to a quasar halo denser than models predict

At a glance

Paper briefs

Galactic winds converge once the sub-grid model is rebuilt around physical scales

The feedback prescription, not the halo, sets how often a dwarf goes quiet

Low-power jets turn four fifths of their energy into heat

NGC 3079's superbubble caps the supernova progenitor mass

Circumgalactic dust tracks stellar mass, more flatly than the models

Galaxies with outflows are the ones missing gas and dust

The quasar halo's cold gas needs a broad, clumpy density distribution

Quasar nebulae come out denser and clumpier than three simulation suites

Lyα fractions put the neutral fraction at 38 per cent at z ≈ 7

A single galaxy at z ≈ 7.57 looks through an almost fully ionised sightline

Lyα survives more often inside a z ≈ 6.2 galaxy overdensity

Code and data releases

What the week separates

References

Related content

Eleven new feedback tests, from converging galactic winds to a quasar halo denser than models predict

At a glance

Paper briefs

Galactic winds converge once the sub-grid model is rebuilt around physical scales

The feedback prescription, not the halo, sets how often a dwarf goes quiet

Low-power jets turn four fifths of their energy into heat

NGC 3079's superbubble caps the supernova progenitor mass

Circumgalactic dust tracks stellar mass, more flatly than the models

Galaxies with outflows are the ones missing gas and dust

The quasar halo's cold gas needs a broad, clumpy density distribution

Quasar nebulae come out denser and clumpier than three simulation suites

Lyα fractions put the neutral fraction at 38 per cent at z ≈ 7

A single galaxy at z ≈ 7.57 looks through an almost fully ionised sightline

Lyα survives more often inside a z ≈ 6.2 galaxy overdensity

Code and data releases

What the week separates

References

Related content