Introduction

In recent years, remarkably tight correlations have been observed between the properties of supermassive black holes (SMBHs) residing in galaxy cores and those of the host galaxies themselves (Magorrian et al. 1998; Gebhhardt et al. 2000; Häring and Rix 2004). A growing body of evidence seems to support the idea that feedback from active galactic nuclei (AGN) provides a natural link between these. While every galaxy can potentially host a SMBH, only a relatively small fraction of these are observed in an active state. AGN activity manifests itself through powerful outflows observable right across the electromagnetic spectrum.

The central black holes are powered by accretion of surrounding cold gas. The resultant outflows, in turn, affect the cold gas supply by heating and/or transporting this gas away from dense inner regions with short cooling times. It is for this reason that feedback from radio sources is particularly interesting. Despite only contributing around one per cent of the AGN bolometric luminosity, radio-loud AGNs can profoundly affect their surroundings through such mechanical feedback. One piece of observational evidence supporting this view comes from studies of X-ray clusters. In the absence of feedback, large amounts of cold gas are expected in dense cluster cores (due to short cooling times), however no such gas has been found. This well-known ‘cooling flow problem’ points to the need for a central, powerful heating source.

Introduction

There are two families of luminous elliptical galaxies: cusp galaxies and core galaxies. Cusp galaxies have steep power-law surface-brightness profiles down to the centre (hence the name ‘power-law’ galaxies, often used as a synonym for cusp galaxies), corresponding to intrinsic stellar density profiles with inner logarithmic slope γ > 0.5; core galaxies have surface-brightness profiles with a flat central core, corresponding to γ < 0.3 (Faber et al. 1997; Lauer et al. 2007). Cusp galaxies are relatively faint in optical, rotate rapidly, have disky isophotes, host radio-quiet active galactic nuclei (AGN) and do not contain large amounts of X-ray-emitting gas; core galaxies are brighter in the optical, rotate slowly, have boxy isophotes, radio-loud AGN and diffuse X-ray emission (for a summary of these observational findings see Nipoti and Binney 2007; Kormendy et al. 2009, and references therein). The most popular explanation for the origin of such a dichotomy is that cusp galaxies are produced in dissipative, gas-rich (‘wet’) mergers, while core galaxies are produced in dissipationless, gas-poor (‘dry’) mergers (Faber et al. 1997), the cores being a consequence of core scouring by binary supermassive black holes (Begelman et al. 1980). The actual role of galaxy merging in the formation of elliptical galaxies is still a matter of debate (e.g. Naab and Ostriker 2009). What is reasonably beyond doubt is that cores must be produced by dissipationless processes, while cusps are a signature of dissipation (Faber et al. 1997; Kormendy et al. 2009, and references therein).

Introduction

AGN feedback is widely proposed as the solution to a number of otherwise difficult-to-explain problems in extra-galactic astrophysics. From an observational perspective, it is worth first dissecting the forms of ‘feedback’ that are under discussion, before embarking on any project to observe this potentially universal process. Figure 11.1 gives a short summary of the topic of feedback, which can broadly be split into two parts (column 2): heating of gas in situ, and outflows that remove matter from the host galaxy. Both processes may, or may not, be associated with jets, so jets have been placed separately. While outflows are assumed to predominantly affect the nuclear region and possibly the ISM of the host galaxy, in-situ heating of the gas must occur on very large scales within the IGM (column 3). The final column presents a selection of observed or yet-to-be-observed consequences of the physical mechanisms: the list is not meant to be exhaustive, but simply presents the range of the observations with which we must deal. While there is little argument that some aspects of AGN feedback have been directly detected, conclusive evidence for routine quenching of star formation and removal of the interstellar medium of the QSO host galaxy remains elusive.

Of particular relevance to this contribution are the narrow absorption line systems (NALs), which appear in every box on the right-hand side of Figure 11.1, and are arguably one of the best candidates for directly detecting ‘ubiquitous’ QSO feedback.

AGN feedback from a radio galaxy perspective

The relationship between black hole mass and bulge mass (Magorrian et al. 1998; Gebhardt et al. 2000a) indicates a symbiotic relationship between the formation of supermassive black holes and galaxy formation. Silk and Rees (1998) indicated how an isotropic wind from a black hole may interact with the infalling gas in a forming galaxy to provide a natural relationship between black hole mass and bulge mass. Saxton et al. (2005) also showed that jets propagating through an inhomogeneous interstellar medium generate an energy-driven, more or less spherical bubble, different from the bipolar structure that is usually associated with classical radio galaxies. Thus, from our viewpoint, when we consider the interaction between jets and the interstellar medium we naturally think of gigahertz peak spectrum (GPS) and compact steep spectrum (CSS) radio galaxies as well as high redshift radio galaxies. These sources appear to be radio galaxies in the early stages of their evolution in which there is abundant evidence for strong jet–ISM interaction in the form of shock-excited emission lines and anomalous gas velocities. Given that jet power and momentum can be isotropically distributed by an inhomogeneous medium, an important issue to address is the detailed interaction between clouds and outflows in such a medium. The nature of this interaction and in particular the momentum imparted to the gas surrounding the active nucleus is going to be quite different from that envisaged by Silk and Rees (1998) and many other papers since.

Introduction

Outflows from AGN are observed in a wide variety of forms: radio galaxies, broad absorption line quasars, Seyfert galaxies exhibiting intrinsic absorption in the UV, broad emission lines, warm absorbers and absorption lines in X-rays (e.g. Creenshaw et al. 2003; Everett 2007). There have been studies on the cosmological impact of quasar outflows on large scales (Furlanetto & Loeb 2001, here after FL01; Scannapieco and Oh 2004, hereafter SO04; Levine & Gnedin 2005, hereafter LG05). Barai (2008) investigated the cosmological influence of radio galaxies over the Hubble time. All these studies considered spherically expanding outflows.

On cosmological scales an outflow is expected to move away from the highdensity regions of large-scale structures, with the outflowing matter getting channelled into low-density regions of the universe (Martel and Shapiro 2001). We implement such anisotropic AGN outflows within a cosmological volume. The simulation methodology is given in Section 13.2, and the results are discussed in Section 13.3.

The numerical setup

N-body simulation and distribution

We simulate the growth of large-scale structures in a cubic cosmological volume of comoving size Lbox = 128h−1 Mpc. We use the particle-mesh (PM) algorithm, with 2563 equal-mass particles, on a 5123 grid. A particle has a mass of 1.32 × 1010M⊙, and the grid spacing is Δ = 0.25h−1 Mpc. We consider a concordance ∧CDM model with the cosmological parameters:ΩM = 0.268, Ω∧ = 0.732, H0 = 70.4 km s−1 Mpc−1, Ωb = 0.0441, ns = 0.947, and TCMB = 2.725.

Introduction

High-quality data from X-ray satellites have established a number of facts concerning the statistical thermo- and chemo-dynamical properties of the intracluster medium (ICM) in galaxy clusters. In particular, core regions of relaxed clusters show little evidence of gas cooler than a third of virial temperature (e.g. Peterson et al. 2001), temperature profiles display negative gradients outside the core region (e.g. Vikhlinin et al. 2005; Leccardi and Molendi 2008b) and the distribution of iron in the ICM shows a negative gradient, which is more pronounced for relaxed ‘cool core’ clusters (e.g. Vikhlinin et al. 2005; Leccardi and Molendi 2008a).

The above observational properties of the ICM arise from an interplay between the cosmological scenario of building up the large-scale structure and a number of astrophysical processes (e.g. star formation, energy and chemical feedback from supernovae and AGN) taking place on much smaller scales. Such issues can be addressed using cosmological hydrodynamical simulations where the complexity of relevant astrophysical processes can be described as the result of hierarchical assembly of cosmic structures (e.g. Borgani et al. 2008).

The generally accepted solution to the shortcomings of simulations is represented by AGN feedback. The presence of cavities in the ICM at the cluster centre is considered as the fingerprint of the conversion of the mechanical energy associated with AGN jets into thermal energy (and possibly in a non-thermal content of relativistic particles) through shocks (e.g. McNamara and Nulsen 2007).

Introduction

Halo assembly bias, the dependence of dark-halo clustering on their formation history, is becoming increasingly important. The reason for this is that a better understanding of the formation of galaxies is needed in order to fully exploit new measurements, which are being developed with increasing precision.

According to the standard cosmological scenario, galaxies are formed in high density regions consisting of virialized dark-matter particles. Such systems, termed dark-matter halos, form in a hierarchical and self-similar fashion, in which smaller objects form first and then continuously merge into ever larger objects. The merging process is not linear, in the sense that it doesn't arise from linear theory. Thus, it defines a time-dependent scale in which matter clustering becomes non-linear.

The theoretical framework describing the process of halo formation is the excursion set theory (Bond et al. 1991; Lacey & Cole 1993; Mo & White 1996). According to this theory, dark-matter density fluctuations, at a given scale, grow in the linear regime until they reach a critical value when they collapse. The collapse process is equivalent to a merging process of the smaller scales. When combined with the theory of random Gaussian fields, this framework explains the formation history of halos. A major result is that the history of a halo should not be correlated with the halo environment. This follows because density fluctuations are not correlated with larger scales. Another way of stating this is that halo history should not be correlated with the clustering of the halos themselves.

Introduction

This contribution aims to address the fundamental question, effectively highlighting the overall theme of the workshop, as to what processes are important for eventually suppressing the growth of supermassive black holes (SMBHs) and how is this related to the evolution of star formation from z ∼ 1 to the present. As illustrated in Figure 4.1, a global decline in mass accretion onto SMBHs and star formation rate density over the last 8 Gyr (Boyle and Terlevich 1998; Merloni 2004; Silverman et al. 2008b) is evident and may be driven by a mechanism such as feedback from AGN affecting the gas content of their hosts (Granato et al. 2004; Hopkins et al. 2008; Silk and Norman 2009). Such coupling may not only explain the local SMBH–bulge relations (see Shankar 2009 for an overview) but rectify the inconsistency between the observed distribution of high-mass galaxies and that predicted by semi-analytic models (Croton et al. 2006). Intriguingly, there is observational evidence for AGNs influencing their larger-scale environments that may lend support for the aforementioned feedback models. For example, radio jets are capable of impacting their intracluster medium, (Fabian et al. 2006) which may then in turn regulate further cluster cooling and inhibit star formation in the AGN host galaxy itself (Rafferty et al. 2008). Even at low power, radio-emitting outflows are capable of redistributing line-emitting gas in galactic nuclei (Whittle and Wilson 2004).

Introduction

In a ∧CDM cosmology, the growth of structure occurs hierarchically; small objects form first and undergomergers to form more massive objects. Modelling the formation and evolution of galaxies with numerical simulations is impossible because crucial aspects lack a complete physical model – notably, feedback and star formation.

An alternative to full numerical simulations is the semi-analytic models (Croton et al. 2006; De Lucia et al. 2006), so called because they use approximate prescriptions for physical processes that are poorly understood. These prescriptions contain parameters that are set by demanding that the model reproduces the observations of (typically) low-redshift galaxies. The process itself is often motivated by a result from a more detailed numerical simulation or from observations.

The cooling of gas is central to the process of galaxy formation, as it sets the rate at which gas becomes available for star formation. Feedback processes have the largest impact on the predictions for galaxy properties, as these processes affect the efficiency of galaxy formation by increasing the cooling time of hot gas and suppressing further star formation. Previous iterations of the semi-analytic models (Kauffmann 1996), which lacked a prescription for feedback, predicted that the galaxy population should continue to evolve at z < 1, at a rate greater than that due to passive evolution, as more massive galaxies are built up by continued mergers.

Introduction

Some galaxy evolutionary models postulate that powerful starburst galaxies at high-z yield local massive galaxies following the effects induced by an accreting supermassive black hole (SMBH) at their centre (e.g. Di Matteo et al. 2005). However, it is not clear on which spatial and temporal scales and through which physical processes this transition takes place (see Coppin et al. 2008). Here, we investigate this evolutionary scenario by comparing star formation rates (SFRs), AGN activity and stellar masses in high-z (z ∼ 2) active systems.

Spitzer selection of high-z luminous infrared galaxies

For this work, we selected a sample of IR luminous source candidates in a ∼s20 deg2 area obtained by combining the Lockman Hole field (LH, ∼11 deg2, α = 10h 45m, δ = + 58°), and the XMM-LSS field (XMM, ∼9 deg2, α = 02h 21m, δ = −04° 30′) of the Spitzer Wide Area Infrared Extragalactic Survey (SWIRE; Lonsdale et al. 2003). Both fields benefit from multi-band ground-based optical (Ugriz) and Spitzer IR bands (seven bands from 3.6 to 160 μm). IR luminous sources, powered by star formation or AGN activity, are expected to be bright mid-infrared (MIR) sources. Powerful starburst galaxies are characterised by spectral energy distributions (SEDs) that are bright throughout the MIR to millimetre range. Luminous AGNs are bright MIR sources because their emission from AGN-heated dust peaks in the MIR. We thus selected all sources with a 24 μm flux > 400 μJy (corresponding to ≳ 5σ).

Introduction

At redshifts above z ≳ 0.5 extragalactic jet sources are commonly associated with extended emission line regions (for a review see McCarthy 1993; Miley and De Breuck 2008). The most prominent emission line is the hydrogen Lyman α line, but other typical nebular emission lines have also been found. These regions are up to 100 kpc in extent, anisotropic and preferentially aligned with the radio jets (alignment effect). Their properties correlate with those of the radio jets: smaller radio jets (< 100 kpc) have more extended emission line regions with larger velocity widths (1000 km s−1) that are predominantly shock ionised, as diagnosed from their emission line ratios. Larger radio jets (> 100 kpc) have emission line regions even smaller than 100 kpc. Their turbulent velocities are typically about 500 km s−1 and the dominant excitation mechanism is photoionisation. The physical function of these emission line regions can be compared to a detector in a particle physics experiment: in both cases a beam of high-energy particles hits a target. Analysis of the interactions in the surrounding detector, or in astrophysics the emission line gas, provides information about the physical processes of interest. For the astrophysical jets, the information one would like to obtain from such analysis concerns two traditionally separate branches of astrophysics.

The considerable energy release that may be associated with the jet phenomenon is received by a large reservoir of gas surrounding the host galaxy.

Introduction

Only by incorporating various forms of feedback can theories of galaxy formation reproduce the present-day luminosity function of galaxies. It has also been argued that such feedback processes might explain the counterintuitive behaviour of ‘downsizing’ witnessed since redshifts z ≈ 1 − 2. To examine this question, observations spanning 0.4 < z < 1.4 from the DEEP2/Palomar survey (Bundy et al. 2006) are compared with a suite of equivalent mock observations derived from the Millennium Simulation, populated with galaxies using the Galform code (Bower et al. 2006).

Hierarchical assembly

The mock galaxy samples are generated from the population of dark matter halos in the Millennium Simulation (Springel et al. 2005). This simulation consists of approximately 10 billion dark matter particles each of mass 8.6 × 108h−1M⊙ evolving in a cubic volume of side 500h−1 Mpc, assuming a ∧CDM cosmology.

Dark matter halo merger trees are found from this 4-volume using the methods described by Harker et al. (2006). The lowest mass halos contained in these trees, of which there are about 20 million, consist of 20 particles corresponding to a total mass of 5 × 109h−1M⊙. Such halos could contain at most 9 × 108h−1M⊙ of baryonic material, which is well below the lower limit of the stellar mass functions to be considered in this work. Therefore we do not expect the resolution of the Millennium Simulation to affect our results.