Eleven new feedback tests, from Fermi Bubble clouds to VariableTNG halos

Quick scan

Controlled feedback experiments

VariableTNG: the parameter transition behind quenching

• Authors/date: Ignacio G. Alfaro et al.; submitted August 17, 2026; v1.
• Study design: The VariableTNG suite varies eight galaxy-formation parameters while keeping the initial conditions fixed. The paper analyzes 26 simulation boxes and trains Random Forest regressors against the quenched-fraction variation relative to TNG100-1. SHAP values supply both the size and direction of each parameter's influence. 1
• Result: The stellar-feedback wind parameter is the main driver at low stellar mass. Its importance shifts toward AGN-related parameters at higher mass, while the supernova temperature remains important at both ends of the mass range. The transition survives when central and satellite galaxies are separated. 1
• Why read: This is a clean sensitivity map for deciding which feedback knobs deserve calibration attention when the observable is the quenched fraction. It also connects the parameter ranking to the mismatch between the fiducial simulation and observed passive galaxies.
• Caveat: The evidence is a controlled, model-internal attribution, not a direct measurement of how real CGM gas couples to galaxies. The abstract reports the ranking but not effect sizes for the quenched-fraction changes.

VariableTNG: one stellar-feedback link from early galaxies to black holes

• Authors/date: Kexin Liu et al.; submitted August 18, 2026; v1.
• Study design: The authors use 50 Sobol-sampled VariableTNG simulations with eight subgrid parameters, fixed cosmology, and fixed initial conditions. They compare and galaxy and black-hole observables with recent measurements, then use Random Forest analyses to rank parameter sensitivity. 2
• Result: Low-mass galaxy abundance is most sensitive to stellar feedback, especially the supernova temperature and thermal wind fraction. The high-accretion tail of black-hole growth depends on seeding and black-hole feedback, but also on the supernova temperature, implying an indirect link through the available gas supply. 2
• Why read: The paper gives an explicit path from a feedback prescription in small early galaxies to a black-hole population that may later drive AGN feedback. That makes it useful for separating direct AGN control from stellar-feedback-mediated gas supply.
• Caveat: The comparison is limited by uncertain high-redshift observations and by cosmic variance in independent small volumes. The reported tensions are diagnostic model tensions, not definitive failures.

VariableTNG: morphology changes its controlling mechanism with mass

• Authors/date: Yikai Liu et al.; submitted August 20, 2026; v1.
• Study design: Cosmological magnetohydrodynamic simulations with AREPO vary eight stellar- and AGN-feedback parameters at fixed initial conditions. At , the authors measure , $v/\sigma$, and $c/a$ to track rotational support and shape. 3
• Result: Below roughly in stellar mass, the supernova temperature is the dominant control: larger values delay early star formation and favor a denser, more rotationally supported gas reservoir. At higher masses, morphology becomes more sensitive to the quasar-mode AGN coupling efficiency. The morphology–mass relation is non-monotonic, with maximum rotational support at intermediate mass. 3
• Why read: It tests a common shortcut—treating one feedback channel as the universal morphology regulator—and replaces it with a mass-dependent picture that can be compared with disk fractions and kinematic surveys.
• Caveat: The result is a controlled simulation comparison at ; it does not by itself identify which CGM observation would uniquely distinguish the feedback prescriptions.

Dianoga: the coupling interface matters at cluster scale

• Authors/date: Stefano Borgani et al.; submitted August 18, 2026; v1.
• Study design: Dianoga contains simulations of 28 cluster-centered regions, a cosmological box, and 293 halos with . The OpenGadget3 runs compare a reference AGN-feedback model with six alternative configurations and confront stellar, ICM, and IGM profiles with observations. 4
• Result: The reference model broadly matches the stellar mass function and ICM/IGM properties, but overpredicts the high-mass end and BCG stellar masses. Across alternative prescriptions, models with thermal evaporation of unresolved ISM gas bring BCG masses and stellar mass fractions closer to observations while producing more cool-core-like clusters. The paper argues that the interface between AGN energy injection and the unresolved ISM is at least as important as total feedback efficiency. 4
• Why read: This is the cluster-scale counterpart to the VariableTNG parameter studies: it asks whether an integrated energy budget is enough to predict observable baryon profiles. The answer is no if the coupling location and unresolved gas treatment change.
• Caveat: The AGN parameters were minimally calibrated to the local black-hole–stellar-mass relation, not to every ICM or CGM observable. The sample is group and cluster dominated, so its lesson should not be applied unchanged to Milky-Way-mass halos.

Direct CGM and wind structure

FOGGIE: a warning about reading turbulence from emission

• Authors/date: Helena Bouchereau et al.; submitted August 17, 2026; v1.
• Study design: The authors use high-resolution cosmological zoom-in FOGGIE simulations and compare three-dimensional velocity structure functions with emission-weighted projected two-dimensional versions. They vary projection, resolution, measurement-area size, and gas temperature, focusing on the velocity-structure-function turnover often used as a turbulence-driving scale. 5
• Result: Projection lowers the velocity-structure-function normalization, but the abstract reports no significant slope difference between the 3D and emission-weighted 2D measures. The measurement-area size correlates directly with the turnover location. 5
• Why read: The paper turns instrument sensitivity and nebula size into part of the physical interpretation. A measured driving scale is not portable unless the emitting area, projection, resolution, and gas phase are carried along with it.
• Caveat: The result is a simulation-to-observable calibration, not a direct measurement of a particular galaxy's feedback power. The abstract does not give a universal numerical conversion from turnover scale to injected energy.

Fermi Bubbles: neutral clouds accelerate inside the nuclear wind

• Authors/date: Felix J. Lockman et al.; submitted August 17, 2026; v1.
• Study design: New Green Bank Telescope H I surveys cover about 500 square degrees around the Galactic center at 9.1 arcmin, or about 22 pc at the Galactic center. The paper analyzes 228 high- clouds and fits kinematic models to the outflowing population. 6
• Result: The selected clouds span $-335$ to . Their positions suggest acceleration from low velocity near the nucleus to at least within $\lesssim 4$ kpc. The clouds become fainter and have lower H I column density with distance, with an abrupt neutral-cloud cutoff about 2 kpc above the Galactic plane. 6
• Why read: This is a resolved, phase-specific constraint on how a galactic wind changes neutral material as it travels through the inner halo. It is a useful kinematic anchor for simulations that otherwise compare only integrated wind energetics.
• Caveat: The result traces neutral clouds in one nuclear wind and uses kinematic models for the three-dimensional population. It does not isolate the original energy source or determine a total mass-loading rate.

NGC 891 and NGC 4565: the neutral CGM is not azimuthally smooth

• Authors/date: Mary Rickel et al.; submitted August 19, 2026; v1; accepted to ApJ.
• Study design: Deep Green Bank Telescope observations target off-axis sightlines in the inner neutral CGM, within $<25$ kpc of NGC 891 and $<30$ kpc of NGC 4565. The authors compare the single-dish detections with deep WSRT HALOGAS maps, using velocity-offset corrections and channel-wise brightness-temperature scaling to separate disk contamination. 7
• Result: At a 5$\sigma$ sensitivity of $1.1$– over a 20 width, 30–38% of the GBT emission in NGC 891 and 18–28% in NGC 4565 cannot be explained by WSRT disk emission. The off-axis H I is 4–6 times the major-axis amount and 3–7 times the minor-axis amount, depending on the comparison, and the velocity profile suggests lagged co-rotation. 7
• Why read: The observable contrast and sampled scale are unusually explicit. The result tells modelers that a single azimuthally averaged neutral-CGM profile can erase the geometry that carries information about inflow, outflow, or environmental disturbance.
• Caveat: Separating true CGM emission from disk contamination remains the central difficulty. The design establishes neutral-CGM structure but cannot by itself isolate AGN feedback from stripping, assembly history, or other environmental processes.

CRISTAL-02: Ly$\alpha$ escape follows an outflow, but not a unique engine

• Authors/date: Arshi Ali et al.; submitted August 19, 2026, revised August 21, 2026; v2; accepted to A&A.
• Study design: MUSE and JWST/NIRSpec integral-field observations are combined with NIRCam ultraviolet imaging. The authors construct spatially matched emission-line, extinction, Ly$\alpha$-escape-fraction, and ionizing-photon-efficiency maps for the star-forming galaxy CRISTAL-02 at . 8
• Result: Ly$\alpha$ is more extended than H$\alpha$ and UV, reaching about 33 kpc and preferentially extending along the cold molecular-gas outflow traced by [C II]. Two star-forming clumps show opposite Ly$\alpha$ behavior: lower dust attenuation favors escape in one, while the brighter H$\alpha$ and UV clump has suppressed Ly$\alpha$. Radiative-transfer modeling favors an outflow-driven escape scenario. 8
• Why read: It links a short-lived ionizing tracer to a larger-scale outflow geometry, showing why Ly$\alpha$ morphology can carry more information than an integrated escape fraction.
• Caveat: The available data cannot distinguish AGN-driven from star-formation-driven feedback. The outflow interpretation is supported by morphology and radiative-transfer modeling, not by a unique engine measurement.

Gz9p3 at : an early fountain caught in two phases

• Authors/date: Ryota Ikeda et al.; submitted August 17, 2026; v1.
• Study design: JWST/NIRSpec G395H integral-field spectroscopy at resolves the interacting galaxy Gz9p3 into a core and a tail. The study maps electron density, gas-phase metallicity, star formation, and the secondary [O III] $\lambda 5007$ component across the system. 9
• Result: The core has and metallicity 0.25 dex below the mass–metallicity relation, while the tail has and higher specific star formation. Ionized outflows appear across most of the system, but their velocity is below escape velocity; the authors interpret the system as an early galactic fountain fed by inflow and interaction-driven star formation. 9
• Why read: This is a rare high-redshift case where inflow, dilution, star formation, and outflow are spatially separated within one object. It tests whether early feedback recycles locally rather than clearing the halo.
• Caveat: The evidence is based on resolved line components and an escape-velocity comparison. The abstract does not report an outflow mass rate, and the fountain interpretation remains dependent on the system's dynamical modeling.

IGM-scale baryon accounting

eROSITA: a hot-gas budget that challenges strong-feedback variants

• Authors/date: H. Khalil et al.; submitted August 18, 2026; v1; accepted to A&A.
• Study design: The paper analyzes a complete sample of 25 galaxy groups from the first public eROSITA-DE release, matched to the Two Micron Redshift Survey group catalogue. Surface-brightness profiles and hot-gas fractions are measured to , with Bayesian fits to gas–mass and X-ray scaling relations. 10
• Result: The mean surface-brightness slope is inside and steepens to beyond it. At median , ; at median , . These measurements lie $2.5$–$8.0\sigma$ from strong-feedback FLAMINGO variants. The inferred matter-power suppression is 10–15% at relative to dark matter only. 10
• Why read: It connects a group-scale hot-gas observable to the small-scale matter distribution, making the CGM/ICM gas budget relevant to cosmological inference rather than only to galaxy evolution.
• Caveat: A hot-gas deficit is not a unique AGN diagnostic. The tension depends on the selected group sample, hydrostatic and profile inference, and the feedback variants used for comparison.

turbofrb: an explicit decomposition of the diffuse IGM and halo tail

• Authors/date: Jéferson A. S. Fortunato et al.; submitted August 18, 2026; v1.
• Study design: turbofrb is a semi-analytic stochastic forward model for . It treats the diffuse IGM, halos, and filaments as explicit channels and couples halo and filament encounters through a latent line-of-sight environment variable. Four effective parameters are calibrated against ray-traced IllustrisTNG hydrodynamics. 11
• Result: The model matches the benchmark mean dispersion measure to percent level, with a per-redshift Jensen–Shannon divergence no larger than over –$2.5$. The diffuse IGM forms the body of the distribution, while halos and filaments populate the high-DM tail. A closed-loop test recovers an injected within $1\sigma$. 11
• Why read: This is a practical bridge between feedback-affected baryon structure and FRB inference. The channel decomposition lets a reader ask whether a DM outlier is more naturally a host, halo, filament, or diffuse-IGM problem.
• Caveat: This is an inference model calibrated to a hydrodynamic benchmark, not a direct feedback measurement. Its agreement with the benchmark should not be read as an independent validation of the underlying baryon distribution.

What can be compared across the set

• Coupling location: VariableTNG varies stellar-wind, supernova, black-hole-seeding, and AGN parameters while holding initial conditions fixed. Dianoga adds a sharper warning: the interface between AGN energy injection and unresolved ISM gas can matter as much as total efficiency. The observational papers mostly see the downstream gas, not the coupling point.
• Gas phase: Fermi Bubbles and the NGC systems use neutral H I. Gz9p3 uses ionized [O III], CRISTAL-02 follows Ly$\alpha$ relative to H$\alpha$, eROSITA measures hot gas, and FRBs integrate free electrons through diffuse IGM, halos, and filaments. These are complementary tracers, not interchangeable measurements of the same reservoir.
• Tracer lifetime: Quenched fractions, morphology, and hot-gas budgets retain information over long timescales. H I clouds, ionized line components, and Ly$\alpha$ escape respond more locally and transiently. FRB dispersion measures integrate a long sightline and are sensitive to environment through a different weighting. A model that matches one lifetime may still miss another.
• Sampled scale: Fermi Bubble acceleration is constrained within $\lesssim 4$ kpc; the NGC neutral-CGM result reaches $<25$–$30$ kpc; CRISTAL-02 Ly$\alpha$ reaches about 33 kpc; eROSITA measures to in groups; and FRBs sample the cosmic line of sight. The VariableTNG and Dianoga studies span these regimes in simulation, but their parameter controls are not direct observations.