A modeling study suggests that a galaxy can lose much of its gas even when its AGN activity has only a weak short-term relationship with the amount of gas present. In the simulation, the galaxy underwent pronounced gas depletion and star-formation quenching, with its cold-gas mass falling by nearly three orders of magnitude. The key distinction is between the flicker of the AGN and the slower change in the galaxy-wide reservoir.
That is the problem the study examines around gas-rich quasar hosts: a weak or null instantaneous link between AGN activity and host-galaxy gas content is not, by itself, incompatible with negative AGN feedback. The authors interpret substantial cold gas in quasar hosts as compatible with efficient, recurrent and time-integrated ejective feedback rather than as evidence against effective feedback. The proposed explanation rests on the long-term history of the reservoir, not on any one reading of AGN brightness.
What the model followed
The study uses high-resolution hydrodynamic MACER simulations of a disk galaxy. The model includes cosmological inflow, multiphase gas, star formation, stellar feedback, and self-consistent radiative and mechanical AGN feedback. The analysis follows time-varying AGN activity under cosmological gas inflow rather than a survey of observed galaxies.
Only one fiducial MACER run supplies the analyzed dataset. It represents a single idealized evolutionary pathway rather than a full cosmological sample, with comparisons made within that pathway across different apertures and activity ranges. The main diagnostics are the gas-to-stellar mass ratio and the cold-gas fraction, examined in relation to instantaneous AGN activity.
At 50 kpc, the cold-gas-to-stellar mass ratio spanned about three orders of magnitude, from roughly 10^-3 to roughly 1, over the model's 12 Gyr evolution. During approximately 8 to 9 Gyr, it fell from about 1 to about 0.01 while the Eddington ratio, the paper's activity measure, varied rapidly. The result is a large and smooth galaxy-scale change alongside much faster AGN variability.
The signal changed with scale
To test whether gas and activity moved together, the authors used Spearman and Kendall rank correlations. They evaluated full, low-Eddington and high-Eddington samples after the first 1 Gyr, across 5, 50 and 500 kpc apertures, with bootstrap error bars drawn from the 16th to 84th percentiles. The correlations were generally weak. The clearest signal was a mild positive association at 50 kpc, where the Spearman correlation was about 0.24; it was weaker at 5 kpc and essentially flat at 500 kpc. The low-Eddington subset tracked the full sample, and no strong anti-correlation was found.
The same mismatch appeared in the cold-gas fraction. The AGN Eddington ratio varied by more than five orders of magnitude, while the cold-gas fraction stayed near 0.1 to 1 at both analyzed apertures and did not closely follow short-timescale accretion variability. In this model, the fraction therefore had weak instantaneous dependence on AGN activity.
Star-formation rate tracked total cold-gas mass more closely than the cold-to-total gas ratio. That association was substantially stronger and approximately linear for total cold-gas mass. The simulated cold-gas mass predominantly spanned 10^8 to 10^11 solar masses, essentially matching the comparison range cited by the paper.
A long process, not a snapshot
The gas-density maps add a less tidy part of the story. During outflow phases, they showed progressive redistribution and removal of gas, but some material fell back or was replenished by cosmological inflow while the overall reservoir declined. The evolution was therefore nonmonotonic rather than a simple one-way loss. The authors read that pattern as support for a recurrent, time-integrated ejective interpretation.
The timescale gap makes the central result easier to see. AGN luminosity varied on roughly 10^5 to 10^6 years, while cumulative depletion of the galaxy-scale gas reservoir unfolded over about 1 Gyr. A gas-rich host and a rapidly changing AGN can therefore coexist in the model without that snapshot settling the fate of the reservoir.
That conclusion remains limited to the simulation. Because the analyzed dataset is one idealized pathway rather than a full cosmological sample, it does not establish population-level behavior for all real AGN or quasar hosts. The study's result is compatibility, not a demonstration that gas-rich quasar hosts undergo ejective quenching.
The authors consequently favor recurrent, time-integrated ejective feedback over a purely preventive scenario. The document is a preprint, arXiv:2608.25341v1, dated 26 August 2026. Its broader claim is conditional: within this single modeled pathway, substantial cold gas can coexist with long-term depletion and quenching.
Paper data and sources
Original title: A systematic study of AGN feedback in a disk galaxy using MACER. III. High Gas Fractions in AGN Hosts
Authors: Yuxuan Zou, Feng Yuan, Suoqing Ji et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text