Different types of AGN feedback, and their imprint on the Large Scale Structure cross cosmic time

Sergio Martin-Alvarez, Vid Iršič, Sophie Koudmani, Martin A. Bourne, Leah Bigwood, and Debora Sijacki

Find the full publication here: https://ui.adsabs.harvard.edu/abs/2025MNRAS.539.1738M/abstract

For more context (Chisari et al. 2019): https://ui.adsabs.harvard.edu/abs/2019OJAp....2E...4C/abstract

What is this about?

Cosmology has a funny habit: we take something ridiculously complicated such as the full cosmic web, and then summarise its properties with a couple of lines in a plot.

One of the most important ones is the matter power spectrum (MPS): it tells us how “clumpy” matter is at different scales. It is the sort of statistic that next generation surveys (Rubin, Euclid, Roman) will measure with extraordinary precision.

However, a slightly uncomfortable fact is that to cash in on that precision, we need theoretical predictions at the ~1% level out to fairly small (non-linear) scales: roughly 10s of Mpc and below. And, as reviewed in Chisari et al. (2019), the uncertainty from galaxy formation physics (cooling, star formation, and especially feedback) can easily exceed that threshold and bias cosmological inferences if ignored.

But here comes the problem. The Universe is not made of collisionless dark matter alone. Gas cools, forms stars that explode, and in the centers of the biggest galaxies we find supermassive black holes that can inject enormous energy into their surroundings. This is what we call active galactic nucleus (AGN) feedback.

So… if AGN are busy kicking baryons around, do they also kick our precision cosmology measurements around?

That’s exactly what we set out to test.

What did we do?

We used the FABLE simulation suite: a set of cosmological hydrodynamical simulations of galaxy formation that includes AGN feedback from accreting supermassive black holes.

The key idea is a controlled experiment. We keep the same cosmology and initial conditions, and we only change how the AGN feedback couples to the gas. In practice, we tweak the efficiency of two commonly invoked “modes”: the quasar mode (high accretion, radiative or wind like), and the radio mode (low accretion, jet like). Then we measure how the matter power spectrum responds across cosmic time, and we also ask a very physical question: where does the change come from? Does the difference come from the regions inside haloes (the environments of galaxies, groups, and clusters), or outside haloes (the more diffuse cosmic web).

Figure: Matter power spectrum suppression in FABLE. The ratio between the full physics matter power spectrum and a dark matter only run, shown at different cosmic times and for different AGN feedback variants. The key feature is a scale dependent suppression, strongest around 1 Mpc, whose evolution depends on whether we modify the quasar or radio mode.

What do we find?

Here are the main takeaways.

AGN feedback leaves a clear, scale dependent imprint. In the fiducial FABLE model, it suppresses clustering most strongly around 1 Mpc, reaching the ~10% level by the present day (cosmic time ~13.8 Gyr) compared to the corresponding dark matter only simulation.

Different AGN “modes” matter in different ways. Changing the quasar mode and changing the radio mode do not produce the same time evolution: radio mode variations tend to be more effective on larger scales and at later epochs, while quasar mode changes imprint differently across time.

Finally, the origin of the suppression shifts with cosmic time. At the present day (cosmic time ~13.8 Gyr), the suppression is dominated by what AGN do inside haloes (where they can rearrange mass in deep potential wells), but at earlier epochs (roughly ~6 Gyr after the Big Bang and earlier), both the mass distribution inside haloes and outside haloes contribute significantly.

If you were wondering “can we just calibrate this away with one number?”, the answer is: not really. The effect is multi scale, time dependent, and it depends on the detailed physics of how AGN feedback couples to baryons.

Why does this matter?

The big picture message is that baryonic physics is not a tiny nuisance if you’re aiming for precision cosmology. Upcoming surveys will measure large scale structure extremely well, and interpreting those measurements means we need a better handle on feedback, especially AGN feedback.

One particularly exciting implication is that measurements at earlier times (roughly ~6 Gyr after the Big Bang and earlier) are not just “more data”. They are a lever arm: they help distinguish what kind of AGN feedback is at play, because the suppression pattern changes depending on whether the mass rearrangement happens predominantly inside haloes, outside haloes, or both.

Where can I read more?

If you’d like all the details (and, importantly, all the plots), I’d recommend reading the paper itself:

https://ui.adsabs.harvard.edu/abs/2025MNRAS.539.1738M/abstract

Feel free to contact me with any questions you might have!

Un saludo!