AGN Feedback with updated sub-grid model


Research Project started from Jan. 2017 @ SHAO


Email from Prof. Ho
1. After Eq. (5). You say that v_W,C ~ 10^4 km/s is well observed. I think the wind velocity depends on the kind of tracer. Some, e.g. UFOs, can have very large velocities, but molecular outflows, for example, have lower velocity but carry most of the mass. Some of these caveats should be given.
2. After Eq. (6). What is the justification for epsilon_0 = 10^-3?
3. S2.4, and elsewhere. You discuss radiation and wind mostly, but occasionally you also mention "jet" as a separate component. This is somewhat confusing. I had thought that when the accretion rate is low, when we are in the RIAF regime, a positive Bernouille parameter cases material to want to escape, and that this occurs in the form of a not-so-well-collimated, not-very-fast "wind", which produces the compact radio emission we see (e.g., Sgr A*, M81*, etc). Some people might call this a "jet", or "base of a jet", but that this is equivalent to a "wind". Are you now saying that RIAFs can have a "wind" and a "jet" at the same time? Or in different phases? Please clarify.
4. S3.1. This work not only neglects the initial ISM, but you also neglect any subsequent halo accretion of gas and, of course, acquisition of gas from mergers, minor or major. These will occur, for any realistic scenario. That is, your galaxy is not only isolated, but it is a closed box. This is fine for parameter exploration, but then you should not be able to discuss evolutionary trends over cosmological timescales (e.g., Fig. 3, 4, 8, 11).
5. S3.1 and elsewhere. I suggest that you no longer call the correlation between BH mass and galaxy mass the "Magorrian relation", or, at least, do not use the scaling as originally given by Magorrian et al. (1998), which has long been shown to be inaccurate. Instead, the latest scaling between BH mass and bulge stellar mass is given by Kormendy & Ho (2013; Equation 10). Using this latest result (note, the BH mass/stellar mass ratio is not a simple constant), an elliptical with stellar mass 3 x 10^11 solar masses corresponds to a BH mass/stellar mass ratio of 0.0059, not 0.001 that you adopt. In other words, your BH mass is a factor of 6 too small.
6. One of the most interesting results of this study is that the light curve consists of a series of very short bursts, as shown in Figure 3. While this may be true *during* a single active episodes, I expect there to be long gaps between episodes with much less activity. It doesn't make sense to me that basically the AGN is almost always "on", from 0 to 12 Gyr. While there are dips when L/LEdd drops by several orders of magnitude, the BH quickly rises back up to L/LEdd ~ 0.01. If this is the case, basically all galaxies should be AGNs almost all the time, at all redshifts. This is not consistent with the luminosity function evolution of AGNs. For example, the most massive BHs (i.e. quasars), turn-off (downsize) below z ~ 2; and so does star formation in these massive galaxies. I don't see any such downturn in Fig. 3. This is probably due to the oversimplified assumptions, as mentioned above, that the galaxy is isolated.
I think this needs to be discussed.
You cite the paper of Schawinski et al. (2015) as support for the 10^5 yr AGN lifetime. Be careful. I know this paper. I was the editor who rejected it, before it eventually appeared in MNRAS. The paper has many many problems.
7. Figure 7 is extremely interesting. I think your most important conclusion, that AGN feedback basically has a negative effect on star formation, especially near the central region, is quite robust. However, I am less confident about there being star formation at >= 10 kpc.
Why? This is not observed. Giant ellipticals generally do not have young stars at large radii.
Besides, I was searching for, but could not find, any clear description of what is your prescription for star formation.
I am also confused whether Fig. 7 applies to the cold or hot feedback mode.
8. Fig. 8 shows a gradually declining specific star formation rate, from ~2 to 12 Gyr. I think this is problematic. Massive galaxies do *not* have this kind of star formation history. They are formed quickly (in a big burst) early (say z = 6-4) and then quickly get quenched by z = 2, after which their SFR stays very low.
9. I like Figure 9 very much. It is nice that it matches observational constraints quite well.
10. Fig. 11-13. I think I mentioned this before, in connection with Yaping's draft. It is nice that you can predict and match the hot gas content. However, what about the cold gas content? If you trust your simulations enough to include star formation, you should have similar predictions for the cold gas.
11. S1. I suggest that you do not cite the paper by Graham et al. (2011). This person has a terrible reputation, and his work is really of very questionable quality.
I suggest that you drop this: "and the observed exponential cutoff … (Schechter 1976)." The usual argument involves the discrepancy between the mass function of dark matter halos and galaxies, which is supposed to be due to AGN feedback for massive galaxies. It is out of place here.


Email from Prof. Feng
In the paper: “In the region 300 pc <r <2 kpc, the suppression of SF is mainly attributed to the wind feedback which pushes the gas away by its momentum interaction with ISM. Beyond 2 kpc, star formation is enhanced.” These conclusions are from your Fig.7, attached below.
I wonder how accurate the value of this 2kpc. Kexin Guo(my postdoc, in cc) are working with the latest IFU data from MaNGA. We find many massive galaxies (M_star ~ 10^11 M_sun) in the process of being quenched, have a H_alpha ring (which indicates star formation) with weak central AGN activity (LINER emission in the center). They typically look like:
The greenish color indicate H_hapha emission (i.e. star formation). The pinkish color in the center indicates both H_hapha and [N II] emission (i.e. LINER emission). We believe these are strong observational evidence that the star formation is quenched from inside, very likely due to the feedback from the central low luminosity AGN. From my students Chengpeng Zhang and Jing Dou’s work, we even know that this quenching mechanism acts through preventing atomic gas convert to molecular gas (by whatever physical mechanism, such as heating, blow gas out etc.).
Hence we are very interested in your theoretical explanation and predictions from your simulation. One thing puzzling me is that the scale of the H_hapha ring (as indicated in the above images) is about 5-10 kpc, larger than the 2 kpc scale from the “fullFB” in your simulation. However, it’s roughly consistent with your “windFB” and “fullFBem03” models. Since the value of 2 kpc is basically the intersection point of the black dashed line (i.e. “noFB”) and blue solid line (i.e. “fullFB”), its precise value could be quite uncertain. What do you think? Do you think the wind feedback could push the gas further way out to 5 or even 10 kpc?
Kexin, can you please send your SFR profiles plot to Feng, with some brief descriptions of the plots? These plots nicely illustrates the star formation is gradually suppressed in the center regions of the galaxies during the quenching process. However, we do not find the evidence of star formation boost (i.e. the positive feedback) in the outer regions.
The results in your fig.8 can also be explained by a self-regulation process. For instance, if the wind feedback blows the gas out, then SFR drops, gas accumulates with time (due to gas inflow) and SFR will rise again. The system (i.e. the galaxy) is roughly in the equilibrium state (i.e. dynamical balance between gas inflow, SF and gas outflow). The average SFR or sSFR is primarily determined by the gas inflow (which is determined by cosmological halo accretion).

unnamed.jpg unnamed2.jpg


  • email from Prof. Yuan

Email from Dr. Kexin
Thanks for the inspiring paper very much! Attached is the SFR profile plot that Yingjie mentinoed based on 131 MANGA face-on disk galaxies. The galaxies are split into 4 bins, with decreasing global star-formation activity from starbursts to red sequence color-coded from purple to red. Actually starbursts here are normal star-forming galaxies, not mergers. The SFR profiles shown in each color are stacked (median) profiles of galaxies in each bin, with the shaded area indicating a 1-sigma scatter. Galaxies selected here has stellar mass mostly between 10^{10.5-11}. Mass distribution difference in each bin is accounted already, so there should be no systematic bias. SFR measured here are based on dust-corrected Ha emission. However for liner hosts, which is a dominant population of red sequence, we did a AGN-SF decomposition based on the line ratios on BPT diagram. For comparison, for the case where we believe Ha in liner regions are exclusively from AGN, we plot the SFR profiles in dashed lines.
Following the decrease in global SFR, i.e., from purple to red, the inside-out damping in star formation could be seen clearly from this plot. Together with Chengpeng's result that AGN rate increases with decreasing SSFR, we believe that the inside-out suppression is very possibly caused by AGN feedback, which may lead to a final quenching of a disk galaxy. We also found there is a connection between central SFR damping and compaction (bulge formation), I am still working on some plots to see if a quantified conclusion could be made.



updated sub-grid AGN physics by Prof. Yuan