Eight new feedback tests, from wind metal loading to AGN demographics

The week of August 31 to September 6, 2026 puts the same question on several scales: where does feedback couple to gas, which phase carries the signal, and how long does the tracer remember the event? The eight papers below range from controlled Milky Way-mass simulations to DESI absorption, H-alpha emission, X-ray halo energetics, resolved AGN outflows, nearby-halo sightlines, and an AGN population database.

Authors and date. Prachi Khatri and collaborators, submitted September 2, 2026. 1

Study design. The authors compare five Auriga cosmological magnetohydrodynamic simulations. The runs vary wind metal loading and energy loading while keeping the z=0 stellar masses and morphologies within about 10% of one another. 1

Result. The extreme models differ by about an order of magnitude in median gas metallicity at |z| >= 30 kpc. Metal loading also changes cooling, the cool-CGM fraction, accretion, and the distribution of metals into the CGM and IGM. 1

Why read. The controlled setup isolates a feedback prescription that is often hidden inside an integrated wind-energy budget. The paper gives a direct test of coupling location and metal tracer in a Milky Way-mass halo.

Caveat. The result is a simulation comparison. Observation matching remains limited by differences between the simulated and observed measurements. 1

Authors and date. S. Weng and collaborators, submitted September 1, 2026. 2

Study design. The authors use DESI DR2 down-the-barrel Na I D absorption for approximately 6,000 star-forming galaxies and compare flow-selected systems with matched controls. 2

Result. Galaxies with outflow signatures have 0.25-0.40 dex higher specific star-formation rates and 0.04-0.06 dex higher central metallicities. Slow-inflow systems have 0.20-0.30 dex higher specific star-formation rates with no reported central-metallicity offset, while fast inflows show a weaker star-formation enhancement and a modestly lower metallicity. 2

Why read. The paper puts neutral inflow and outflow tracers beside two galaxy scaling relations. The sample size makes the comparison useful for deciding whether a flow signature should be read as a short-lived event or as part of a population offset.

Caveat. The design supports population-level associations. The offsets alone leave the causal direction between gas flow, star formation, and central enrichment unresolved. 2

Authors and date. Ivan Rapoport and collaborators, submitted August 31, 2026. 3

Study design. The authors compare Illustris, IllustrisTNG, and TNG300-1 after adding improved H-alpha emission, dust treatment, Lyman-series propagation, diffuse-ISM emission, and AGN photo-excitation models. They test the models against observed H-alpha luminosity functions at z ~ 1-1.5. 3

Result. Diffuse gas contributes about 1%-10% of the H-alpha luminosity. AGN photo-excitation dominates Illustris galaxies above . Low-feedback IllustrisTNG matches the observed luminosity functions, while high-feedback Illustris misses both the faint and bright ends. 3

Why read. H-alpha supplies an observable bridge between feedback prescriptions and galaxy populations at intermediate redshift. The paper also separates nebular emission sources instead of treating the luminosity function as a direct star-formation counter.

Caveat. The inference depends on emission and attenuation prescriptions. The H-alpha luminosity function and stacked kSZ signal favor different feedback regimes, so the observable pair does not select one model by itself. 3

Authors and date. Jennifer K. S. Friske, Filippo Fraternali, and Gabriele Pezzulli, submitted August 31, 2026. 4

Study design. The semi-analytic chemical-evolution model includes a time-evolving three-dimensional potential, inside-out disc formation, and a rotating isothermal CGM. The authors fit the model to Milky Way structure and metallicity-gradient data with Bayesian inference. 4

Result. The fit mildly favors a subvirial corona with . The model predicts present-day CGM rotation of about 100-160 km/s and reproduces the direction of gradient evolution over roughly 8 Gyr. 4

Why read. The paper connects hot-mode accretion and angular-momentum conservation to a tracer that observers can measure in the disc. It is a useful complement to direct outflow studies because the inferred CGM state enters through long-term chemical evolution.

Caveat. The corona properties are inferred indirectly within a semi-analytic model. The paper constrains an accretion and CGM scenario rather than directly measuring feedback energy or an outflow.

Authors and date. A. Cattaneo and S. Ettori, version 2 updated September 4, 2026. 5

Study design. The authors use a one-free-parameter hot-gas model calibrated to the observed X-ray luminosity-halo mass relation. They compare observed entropy with adiabatic simulations and connect heating to SMBH accretion histories. 5

Result. The model infers that SMBHs thermalize 1%-3% of their output in surrounding gas. Ellipticals and groups heat faster than they cool and can unbind halo gas, while clusters above approach heating-cooling self-regulation. 5

Why read. The paper places the feedback-coupling question on a mass sequence from galaxy halos to clusters. Its X-ray entropy framing is useful when comparing local heating with the longer-lived state of hot intragroup and intracluster gas.

Caveat. The coupling fraction is inferred from semi-analytic and semi-empirical assumptions. The paper provides a model-based energy accounting rather than a direct measurement of thermalization.

Authors and date. Pierpaolo Condò and collaborators, submitted September 3, 2026; the record also reports acceptance on September 1. 6

Study design. The authors analyze VLT-MUSE integral-field spectroscopy with custom point-spread-function subtraction. The inferred black-hole mass range is , and the dimensionless accretion rate spans 4.1-980. 6

Result. The resolved ionized outflow extends about 2 kpc, reaches a maximum velocity of roughly 340 km/s, and has an estimated mass outflow rate of ~4.2 x 10^-4 M_sun/yr. The unresolved nuclear outflow reaches about 1140 km/s and exceeds 1.5 M_sun/yr. 6

Why read. The two velocity and mass-flow scales make the coupling-location problem visible in one object. A high-rate nuclear wind and a much weaker host-scale signature can coexist in the same observation.

Caveat. The contrast may reflect weak coupling to the host or different activity episodes. The outflow rates depend on the adopted gas and geometry model. 6

Authors and date. Yong Zheng and collaborators, submitted August 31, 2026. 7

Study design. The sample contains 21 local spiral galaxies within 15 Mpc and 26 HST/COS quasar sightlines at 0.1-0.9 R_200c. The authors combine the sightlines with existing H I satellite surveys. 7

Result. The inferred cool-CGM mass is . Hosts with gaseous satellites reach metal-ion absorber detection rates of up to 50%. Column densities show no reported correlation with satellite count or total H I mass, while sightlines within half a massive satellite's virial radius have elevated detections, consistent with stripped debris. 7

Why read. The paper connects absorber statistics to a spatial scale that can separate satellite environments from the broader host-halo profile. It gives an observational test for environmental gas redistribution alongside feedback-driven CGM models.

Caveat. The intrinsic scatter is about 1-2 dex. The satellite contribution is small relative to host and profile scatter, and the stripping interpretation remains one explanation among several. 7

Authors and date. Alessandro Peca and collaborators, submitted September 3, 2026. 8

Study design. AGN-DB v1.0 combines more than 100 public AGN catalogs and samples through the end of 2025. The cross-match uses the Lyra Bayesian likelihood-ratio framework and preserves per-property provenance. 8

Result. The database contains about 8.1 million unique sources, about 7.8 million after stellar-contaminant flagging, and about 6.8 million classified AGN. The scale makes AGN demographics available as an input to population-level feedback comparisons. 8

Why read. Feedback studies often need an AGN selection function before they can compare host properties, duty cycles, or absorber incidence. The per-property provenance is as important here as the source count.

Caveat. The paper describes a database and catalog construction. The retrieved arXiv record does not expose a separate public download or repository endpoint, so this entry is a database paper rather than an independently verified data release. 8

The papers place feedback coupling at different locations. Khatri et al. vary wind composition and energy inside otherwise similar galaxy simulations. Rapoport et al. test how emission prescriptions translate feedback into an H-alpha population statistic. Cattaneo and Ettori infer long-term black-hole heating in hot halos, while Condò et al. resolve a nuclear-to-host contrast in one super-Eddington AGN.

The gas phases also differ. Na I D traces neutral material, H-alpha traces ionized emission, X-ray entropy traces hot gas, metal ions trace cool CGM structure, and metallicity gradients retain a longer chemical record of accretion. A tracer with a long memory can connect many episodes while losing the timing of any single outflow; a resolved velocity field keeps scale information while leaving geometry and mass-loading assumptions in the result.

The sampled scales run from the nucleus and inner kiloparsecs to disc gradients, tens of kiloparsecs in Milky Way-mass halos, fractions of R_200c around nearby spirals, and cluster-scale hot gas. The evidence therefore supports a comparison of where each prescription leaves an observable imprint. The set does not supply one common calibration target: simulations isolate inputs, observations measure associations or projected gas, and semi-analytic models infer coupling from integrated halo properties.

The targeted check for current-week open-source astronomy releases produced no qualifying CGM/IGM feedback code release. The same check produced no independently verified dataset release; AGN-DB is retained above as a database paper with its access limitation stated explicitly. 8