​

​

2) Mapping AGN accretion in the SFR-M* plane

​

The advantage of using large galaxy samples is the opportunity to investigate the AGN/host connection as a function of multiple parameters at the same time.

In Delvecchio et al. (2015), for the first time, we explored the distribution of the average AGN accretion rate (BHAR) in the SFR-M* plane of star-forming galaxies. We selected a large sample of (~7200) Herschel-detected galaxies in the COSMOS, GOODS-N and GOODS-S fields, over the redshift range 0<z<2.5.

We split our sample in bins of redshift, SFR and M*, and "mapped" the average BHAR in each SFR-M*-z bin from the  X-ray emission obtained via individual detections and stacking of deep Chandra images. We found that the average BHAR correlates with both SFR and M* at z<0.8. At higher redshift, the dependence on the SFR seems the primary driver of the correlation. Our study nicely supports that accretion onto the central black hole and star formation broadly trace each other, in a statistical sense.

​​

Despite numerical simulations and semi-analytical models invoke AGN feedback to "quench" the galaxy star formation and reproduce several observables (local galaxy stellar mass function, Mbh-σ correlation), a compelling evidence for a shut-down of the galaxy star formation induced by AGN feedback is still limited to a few "smoking-gun" cases, but not as widespread as predicted by models. Even more intriguingly, the positive correlations found between BHAR and SFR or M* (e.g. Mullaney et al. 2012; Chen et al. 2013; Delvecchio et al. 2015) pose the question of whether AGN feedback hampers or triggers galaxy star formation.

Addressing this question requires a panchromatic analysis of large samples of AGN host-galaxies.

​

​

​

3) Unveiling elusive, weakly accreting AGN with deep radio data

​

Radio data can be crucial to unveil a elusive population of AGN, that might be missed by X-ray, mid-IR and optical surveys. In particular, radio observations can directly access a key phase of AGN feedback (e.g. ``radio-mode'') in which the AGN mechanical output can limit the M* growth of galaxies (e.g. Croton et al. 2006) in the form of radio-jets or hot bubbles that can extend out to ~Mpc scales (e.g. Fabian et al. 2012).

 

​I have conducted a multi-wavelength investigation of the properties of radio-selected sources in the COSMOS field, using radio data from the VLA-COSMOS 3 GHz Large Project (PI: V. Smolcic et al. 2016), which counts nearly 11,000 radio sources over 2.6 sq. degrees down to an unprecedented depth over an area like COSMOS. In my recent work (Delvecchio et al. 2017), I employed panchromatic diagnostics, including X-ray, mid-IR and radio selection criteria, that enabled me to identify a two-fold AGN population in radio:

- moderate to high radiative luminosity AGN (HLAGN). This population is defined as the collective sample of X-ray (Lx > 10^42 erg/s), mid-IR (Donley et al. 2012) and SED-identified AGN (Delvecchio et al. 2014). They constitute 21% of the sample.

- low to moderate radiative luminosity AGN (MLAGN). They are identified only via a strong excess in radio emission (relative to the galaxy SFR), but completely missed by X-ray, mid-IR and SED diagnostics. They show up only in radio, and constitute 17% of the sample.

​

While HLAGN predominantly lie around the main-sequence (MS) relation (blue contours in the right Figure), MLAGN (red contours in the right Figure) stand significantly below the MS, out to z~4.

​

 

​

​

​

​

​

​​​​​​​4) Toward a comprehensive AGN census through mid-IR, X-ray and radio data

​

It is widely accepted that a multi-wavelength approach is paramount to reach a more comprehensive understanding of the connection between AGN accretion and galaxy evolution.

Several studies (e.g. Hickox et al. 2009; Goulding et al. 2014) have shown that samples of X-ray, mid-IR and radio selected AGN overlap by a small percentage (10-30%, depending on the relative sensitivity of the survey at these frequencies), which suggests that samples of AGN based on a single-band selection are probably biased toward a specific AGN class. However, for the same reason, it becomes clear that different AGN selections give us complementary views on the AGN zoo and the co-evolution with their hosts.

In collaboration with David Alexander, I am working on the distribution of the above AGN populations as a function of their Eddington ratio. My results (Figure on the right) suggest that the identified samples of X-ray, IR and radio-based AGN display significantly different Eddington ratio distributions (Delvecchio et al., in prep.), with radio-identified AGN standing at systematically lower values, compared to those identified from X-ray or mid-IR selection.

Motivated by this, I have been studying the joint sample of X-ray, mid-IR and radio selected AGN in the COSMOS field, either detected or not in a given band. I will deal with non-detections by performing stacking, that consists of grouping non-detections together and piling up their photon counts, at a given frequency. This technique provides a mean signal associated to the undetected galaxy population at a given band. The stacked signals will be combined to those obtained from individual detections, in every band (X-ray, mid-IR or radio), in order to constrain the mean AGN radiative power (from X-ray or mid-IR data) and the mean AGN mechanical power (from radio data). This approach will be applied to every single galaxy detected in the COSMOS field (about one million galaxies in total, Laigle et al. 2016). This way, I will create a large control sample of galaxies without AGN emission, that could be compared to those showing signatures of AGN activity. By iterating this analysis as a function of the galaxy M* and cosmic epoch, I will be able, for the first time, to follow how galaxies with a given M* evolved through cosmic time, undergoing different stages of AGN feedback, from mostly radiative (e.g. higher X-ray-to-ratio output) to mostly mechanical (higher radio-to-X-ray output). My project will clarify how AGN feedback influences the M* behaviour of galaxies, and how it evolves with cosmic time, by means of a large, reliable, and highly complete census of AGN host-galaxies.

 

​

​

​

​

5) Bridging the gap: evolution of AGN populations from pencil-beam fields to all-sky surveys

​

Tracing the evolution of the full AGN population over a wide range of redshifts and luminosities benefits from a combination of deeper and shallower surveys.

​

The analysis that we carried out in the COSMOS field will be tied to those published in other surveys, in order to bridge the gap between the deepest pencil-beam fields (e.g. GOODS North and South, 0.1 deg^2) and the medium/large area surveys such as the XXL (North and South, 50 deg^2 in total).

​

This will ultimately be achievable with the EMU survey, that will cover 75% of the full sky at 1.4 GHz (like the NVSS and FIRST surveys) down to a sensitivity comparable to that reached by the VLA-COSMOS survey at the same frequency.

The enormous amount of data delivered by EMU will be a breakthrough in modern astronomy, as will exponentially increase the size of the galaxy samples available to the scientific community.

I am a member of the EMU consortium, working on multi-wavelength SED-fitting of AGN and galaxies. In this context, the deliverables of the EMU survey will provide an unprecedented step forward in AGN and galaxy evolution studies.

​

​

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

​

​

 

 

 

 

 

 

 

 

 

 

​

​

​

​