🔭 astrophysics

Bound or blown: the fate of hot gas in galaxy groups

By comparing forward-modeled XMM-Newton observations of galaxy groups with FLAMINGO hydrodynamical simulations, this study reveals that intermediate-strength AGN feedback best explains the observed thermodynamic properties of hot gas, ruling out both weak and extreme feedback scenarios.

Original authors: R. Seppi, D. Eckert, J. Schaye, J. Braspenning, M. Schaller, B. D. Oppenheimer, E. O'Sullivan, F. Gastaldello, L. Lovisari, M. A. Bourne, M. Sun, A. Finoguenov, H. Khalil, G. Gozaliasl, K. Kolokythas

Published 2026-06-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: R. Seppi, D. Eckert, J. Schaye, J. Braspenning, M. Schaller, B. D. Oppenheimer, E. O'Sullivan, F. Gastaldello, L. Lovisari, M. A. Bourne, M. Sun, A. Finoguenov, H. Khalil, G. Gozaliasl, K. Kolokythas, Y. E. Bahar, R. Santra

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Cosmic Tug-of-War

Imagine the Universe as a giant construction site. Dark matter acts as the invisible scaffolding that holds everything together. Inside these scaffolding structures (called "halos"), gas tries to cool down and form stars. But there's a problem: the gas is too hot and wants to escape.

Enter the Active Galactic Nuclei (AGN). Think of these as super-powered, cosmic blowtorches located at the center of galaxies. They shoot out massive amounts of energy, heating up the gas and trying to blow it out of the structure entirely.

The big question this paper asks is: How strong is this blowtorch?

Is it a gentle breeze that just warms the gas a little?
Is it a hurricane that blows all the gas away, leaving the galaxy empty?
Or is it a Goldilocks scenario—just strong enough to keep things in balance?

The Players: Real Data vs. Computer Simulations

To answer this, the scientists played a game of "Spot the Difference" between two things:

The Real World (X-GAP): They looked at 44 actual groups of galaxies using the XMM-Newton space telescope. These are like "neighborhoods" of galaxies, not the massive "cities" (clusters) usually studied. They measured how much hot gas was there and how hot it was.
The Virtual World (FLAMINGO): They used a massive supercomputer simulation called FLAMINGO. This simulation runs the Universe over and over again, but with different settings for the "blowtorch" (AGN feedback). Some runs have weak blowtorches; others have extreme, galaxy-destroying blowtorches.

The Challenge: The "Cosmic Lottery"

You can't just look at one galaxy in the simulation and compare it to one in real life. Why? Because of Cosmic Variance.

Imagine you are trying to guess the average height of people in a country. If you only measure 44 people in one small town, your result might be weird just by chance (maybe that town has a lot of basketball players). The universe is the same. The specific 44 groups the scientists saw might just be a "lucky" or "unlucky" sample.

To fix this, the scientists didn't just compare numbers. They built virtual telescopes.

They took the computer simulation.
They programmed a virtual XMM-Newton telescope to look at it.
They made the computer "see" the simulation exactly how the real telescope sees the real sky, including all the blurry edges, noise, and limitations.
Then, they analyzed the fake data exactly the same way they analyzed the real data.

This is like taking a photo of a painting, then taking a photo of a digital copy of that painting using the same camera, and comparing the two photos to see if the digital artist got the colors right.

The Results: Finding the "Goldilocks" Feedback

When they compared the real photos to the virtual ones, they tested several "feedback recipes":

The "Too Weak" Models: These models didn't blow enough gas out. The galaxies in the simulation were too full of gas and too cool compared to reality.
The "Too Extreme" Models: These models were like a firehose set to maximum. They blew almost all the gas out of the galaxy groups. The simulation showed empty, very hot groups. This was ruled out with high confidence (over 4 times the standard deviation).
The "Just Right" Model: The winner was a model called fgas −2σ.
- This model represents a feedback strength that is stronger than the standard "default" setting used in most simulations, but weaker than the most extreme "blowtorch" settings.
- It predicted that about 6% of the mass in these galaxy groups is hot gas.
- The "extreme" model predicted only about 3.8% gas, which was too low.

The Key Takeaway

The scientists found that the "blowtorch" in the real Universe is stronger than we thought, but not as violent as some extreme theories suggest.

The "Blowtorch" is effective: It successfully pushes gas out of small galaxy groups, making them less dense and hotter than gravity alone would allow.
But it's not a vacuum cleaner: It doesn't strip the groups completely bare. There is still a significant amount of hot gas left behind.

Why This Matters

Understanding this balance is crucial because gas is the fuel for making stars. If the blowtorch is too weak, galaxies make too many stars too fast. If it's too strong, they make none at all. By pinning down exactly how strong this feedback is in galaxy groups, the scientists are helping us understand the "recipe" for how the Universe builds its structures.

In short: The Universe's central engines are powerful enough to clear out the neighborhood, but they leave just enough gas behind to keep the cosmic party going.

Technical Summary: Bound or blown: the fate of hot gas in galaxy groups

Problem and Motivation
The thermodynamic state of hot gas in galaxy groups remains a critical uncertainty in understanding galaxy formation and the large-scale structure (LSS) of the Universe. While cosmological hydrodynamical simulations have advanced significantly in modeling star formation and Active Galactic Nucleus (AGN) feedback, the precise impact of feedback on the intra-group medium (IGrM) is still debated. Observations suggest that gas in groups is less dense and more extended than predicted by purely gravitational collapse, implying significant baryon expulsion. However, recent constraints from kinetic Sunyaev-Zel'dovich (kSZ) measurements favor extremely ejective feedback models, whereas other analyses, including XRISM observations, suggest lower velocity dispersions that may indicate overly ejective models. There is a need to directly compare simulation predictions with observational data while rigorously accounting for selection effects and measurement systematics to determine the correct strength of AGN feedback in low-mass haloes ( $M_{500c} \sim 10^{13}–10^{14} M_\odot$ ).

Methodology
This study employs a forward-modelling approach to compare the XMM-Newton Group AGN Project (X-GAP) sample with the FLAMINGO suite of cosmological hydrodynamical simulations. The methodology proceeds through several key stages:

Sample Selection and Analogues: The authors utilize the FLAMINGO simulation suite, which includes various AGN feedback prescriptions ranging from weak to extreme ejection (parameterized by shifts in the gas fraction calibration, $f_{gas} \pm N\sigma$ , and jet models). To create realistic analogues of the X-GAP sample, the authors apply the X-GAP selection function (based on X-ray flux, redshift, and velocity dispersion) to the simulation data. This selection accounts for the specific completeness and purity of the X-GAP sample derived from the AXES catalogue.
Forward Modelling of Observations: A critical innovation is the generation of end-to-end XMM-Newton mock observations. Starting from the 3D distribution of hot gas particles in the simulations, the authors:
- Generate idealized photon lists using the pyXSIM software, incorporating individual particle temperatures, metallicities, and abundances.
- Simulate realistic instrument effects (Point Spread Function, effective area variations, and background) using the xmm_simulator package to produce mock event lists.
- Analyze these mocks using the exact same imaging and spectral analysis pipelines (including point source masking, background modeling, and spectral fitting) applied to the real X-GAP data.
Observable Extraction: From the analyzed mocks, the authors derive core-excised X-ray luminosities ( $L_X$ ), gas masses ( $M_{gas}$ ), temperatures ( $T$ ), and velocity dispersions ( $\sigma_v$ ). They also account for cosmic variance by generating multiple independent light cones and uncertainties in the selection function.
Statistical Comparison: The study compares the distributions of these observables between the X-GAP data and the various FLAMINGO models. Crucially, the authors use a Gaussian copula approach to combine multiple observables (scaling relation normalizations, mean temperature, group counts, and velocity dispersion) while accounting for their covariance, rather than treating them as independent constraints.

Key Contributions

End-to-End Forward Modelling: The paper establishes a rigorous framework for comparing simulations to observations by generating mock XMM-Newton data that includes instrumental effects and analyzing them identically to real data. This eliminates biases arising from comparing "true" simulation quantities directly to "observed" quantities.
Multi-Observable Constraints: By combining the normalization of the $L_X-T$ and $M_{gas}-T$ scaling relations with group counts, mean temperature, and velocity dispersion, the study provides a more robust discriminator for feedback models than any single observable alone.
Quantification of Systematics: The work explicitly quantifies the impact of cosmic variance and selection function uncertainties, demonstrating that group counts alone are a weak discriminator due to large fluctuations (>20%), whereas thermodynamic properties offer tighter constraints.

Results
The analysis yields the following findings regarding the AGN feedback strength in galaxy groups:

Best-Fit Model: The FLAMINGO model with intermediate feedback strength ( $f_{gas} - 2\sigma$ ) provides the best agreement with the X-GAP data, showing a global tension of only $0.8\sigma$ . This model yields a typical gas fraction of $\sim 6.1\%$ for the selected groups.
Ruled-Out Models: The most extreme ejective feedback scenario ( $f_{gas} - 8\sigma$ ) is ruled out at $>4\sigma$ (specifically $4.4\sigma$ ). This model predicts gas fractions as low as $\sim 3.8\%$ and is inconsistent with the observed scaling relations and temperature distributions.
Fiducial vs. Data: The fiducial FLAMINGO calibration ( $L1\_m8$ ) shows a tension of $3.1\sigma$ , indicating that the standard feedback strength is likely too weak to reproduce the observed hot gas content in groups.
Scaling Relations: The normalization of the $L_X-T$ and $M_{gas}-T$ relations serves as the primary discriminator. Stronger feedback shifts systems to lower luminosities and higher temperatures at fixed mass, moving them away from the X-GAP distribution.
Cosmic Variance: The number of detected groups is heavily influenced by cosmic variance, making it a poor standalone metric for distinguishing between feedback models in the low-redshift, low-volume X-GAP sample.

Significance and Claims
The paper claims that the thermodynamic properties of galaxy groups favor an AGN feedback strength that is stronger than the fiducial FLAMINGO calibration but significantly weaker than the most ejective models currently proposed. This result highlights the necessity of combining forward modelling with multi-observable constraints to accurately probe the fate of hot baryons in low-mass haloes.

The authors note that their findings are in tension with recent kSZ measurements that favor the most ejective models ( $f_{gas} - 8\sigma$ ). They suggest this discrepancy may arise from differences in the probed mass and redshift ranges, selection effects, or the specific scales at which feedback operates. The study concludes that while current simulations can reproduce the galaxy stellar mass function, their predictions for the hot gas content in galaxy groups remain sensitive to the details of feedback implementation, and the "fate" of hot gas in these systems is likely intermediate between the extremes tested.