The taxonomy of AGNs tends to be rather confusing as we do not yet understand the physics underlying the AGN phenomenon. Undoubtedly some of the differences we see between various types of AGNs are due more to the way we observe them than to fundamental differences between the various types; this is a theme that will be revisited in Chapter 7. We will introduce the various types of AGNs that are generally recognized, and try to make clear as we proceed how these various types may or may not be related.

Seyfert Galaxies

Seyfert galaxies are lower-luminosity AGNs, with MB > -21.5 + 51og/h0 for the active nucleus the generally accepted criterion, due originally to Schmidt and Green (1983), for distinguishing Seyfert galaxies from quasars. A Seyfert galaxy has a quasar-like nucleus, but the host galaxy is clearly detectable. The original definition of the class (Seyfert 1943) was primarily morphological, i.e., these are galaxies with high surface brightness nuclei, and subsequent spectroscopy revealed unusual emission-line characteristics. Observed directly through a large telescope, a Seyfert galaxy looks like a normal distant spiral galaxy with a star superimposed on the center. The definition has evolved so that Seyfert galaxies are now identified spectroscopically by the presence of strong, high-ionization emission lines. Morphological studies indicate that most if not all Seyferts occur in spiral galaxies (Chapter 8).

Khachikian and Weedman (1974) were the first to realize that there are two distinct subclasses of Seyfert galaxies which are distinguished by the presence or absence of broad bases on the permitted emission lines.

As a direct result of advances in the QSO survey techniques described in Chapter 10, the number of QSOs known as of the early 1990s was of order 104 (Véron-Cetty and Véron 1991, Hewitt and Burbidge 1993). It is therefore possible to explore the distribution of QSOs as a function of z, to compute their comoving space density, and to determine how the AGN population evolves with time.

Simple Tests for Evolution

Determination of the space density and luminosity function of QSOs is a difficult undertaking, and as seen in Chapter 10, the results are sensitive to many different types of selection effects. Despite these many difficulties, it was clear even during the first decade of QSO research that the comoving density of QSOs varies strongly with redshift, with an especially large number of QSOs at z ∼ 2. However, it is also at about this redshift that UV-excess techniques are most sensitive, since this is where the Lya emission line falls in the U band, so originally it was not entirely clear to what extent the inferred high densities of QSOs at z ∼ 2 were due to selection effects rather than a real peak in the comoving space density. For this reason we begin this chapter with a discussion of simple tests for changes in space density as a function of z, or equivalently, lookback time.

The Log N – Log S Test in a Non-Euclidean Universe

The classic test for a constant space density of sources is the log N - log S test that was introduced in Chapter 10.

We have seen in the previous chapters that QSOs are valuable probes of the early Universe because they can be detected at large cosmological distances. The concomitant large lookback times provide a means of studying the Universe during the era when galaxy formation is expected to have occurred. QSOs can also be used in another way as a cosmological probe, namely as background sources against which we see intervening objects. Since the ‘sight lines’ to individual QSOs are of order Gpc, the chances of finding objects such as galaxies between us and any given QSO are non-negligible. Gas along the line of sight will produce absorption lines in the spectra of QSOs, and the redshift of these absorption lines zabs will reflect the cosmological distance of the absorbing cloud rather than that of the QSO (which will have emission-line redshift zem), so we can expect that QSO spectra will show absorption lines characterized by zabs < zem. An ‘absorption-line system’ consists of a number of absorption lines in a QSO spectrum that are all at very nearly the same redshift zabs and presumably arise in the same absorber. Thus, the objects probed are not the QSOs themselves, but the intervening material that produces the absorption spectrum. We include discussion of absorption-line characteristics (a) because some absorption lines actually appear to arise in material associated with the QSOs themselves, (b) because absorption by intervening material modifies the QSO spectrum we observe, and (c) because the study of QSO absorption lines has historically been closely associated with the study of the QSOs themselves.

Because the highest-luminosity AGNs can be detected at very large distances (and hence ‘lookback’ times) they provide an important probe of the history of the Universe. Luminous quasars are detected at redshifts up to z ≈ 5; we see these objects as they were when the Universe was ∼ 10% its current age. Quasars are the most distant discrete objects that we have observed, and they are therefore of tremendous importance in understanding the formation of discrete structures from the primordial gas and as measures of the first appearance of metals. Furthermore, as w7e shall see in subsequent chapters, the luminosity function of quasars has varied over the observable history of the Universe, and this has something to tell us about the formation and evolution of galaxies and photoionization equilibrium of the intergalactic medium as a function of redshift, among other things. Finally, as we will see in Chapter 12, quasars can be used as background sources against which we can detect the absorption signatures of less luminous objects, thus providing us with a probe of otherwise unobservable gas at high redshifts.

Determination of the space density and luminosity function of any type of astronomical object is difficult, primarily because of the different volumes over which the most luminous and least luminous objects of a given class can be detected. In the case of quasars, the situation is further complicated by the fact that quasars are observed at sufficiently large distances that the curvature and expansion of the Universe must be taken into account.

In the last four chapters, we have examined the individual components that constitute AGNs. At this point, it is worth briefly summarizing these as we proceed to develop a more global picture of the AGN phenomenon. Although direct proof is lacking, the evidence points towards gravitational accretion of matter by supermassive black holes as being the primary energy source in AGNs. Gravitational potential energy is converted into radiation via viscous dissipation in an accretion disk surrounding the black hole. For a luminous Seyfert galaxy, the black-hole mass is inferred to be something like ∼ 107M⊙ (i.e., RS ≲ 1013 cm), and the UV/optical continuum-emitting region of the purported accretion disk is smaller than ∼ 1015 cm. The corresponding X-ray-emitting region appears to be smaller still, perhaps only several times RS. Surrounding this is the BLR, which has a typical size of 1016cm or so, but whose specific geometry and kinematics are poorly known. It appears that most of the IR continuum emission arises on spatial scales larger than the dust sublimation radius (≳ 1017 cm). This is also where we find the lower-density NLR, where the gas motions are dominated by gravity, but where we also see evidence for interaction with jets in the form of shock-heating and outflowing gas.

In general the term ‘active galactic nucleus’, or AGN, refers to the existence of energetic phenomena in the nuclei, or central regions, of galaxies which cannot be attributed clearly and directly to stars. The two largest subclasses of AGNs are Seyfert galaxies and quasars, and the distinction between them is to some degree a matter of semantics. The fundamental difference between these two subclasses is in the amount of radiation emitted by the compact central source; in the case of a typical Seyfert galaxy, the total energy emitted by the nuclear source at visible wavelengths is comparable to the energy emitted by all of the stars in the galaxy (i.e., ∼ 1011L⊙), but in a typical quasar the nuclear source is brighter than the stars by a factor of 100 or more. Historically, the early failure to realize that Seyferts and quasars are probably related has to do with the different methods by which these two types of objects were first isolated, which left a large gap in luminosity between them. The appearance of quasars did not initially suggest identification with galaxies, which is a consequence of the basic fact that high-luminosity objects, like bright quasars, are rare. One is likely to find rare objects only at great distances, which is of course what happens with quasars. At very large distances, only the star-like nuclear source is seen in a quasar, and the light from the surrounding galaxy, because of its small angular size and relative faintness, is lost in the glare of the nucleus. Hence, the source looks ‘quasi-stellar’.

The narrow-line region (NLR) in AGNs is of interest for at least three interrelated reasons. First, the NLR is the largest spatial scale where the ionizing radiation from the central source dominates over other sources. Second, the NLR is the only AGN component which is spatially resolved in the optical – this is of particular importance as the NLR is clearly illuminated in a non-isotropic manner by the central source. Finally, the NLR dynamics might tell us something about how AGNs are fueled.

Narrow-Line Spectra

As in the case of the BLR, the relative strengths of the emission lines we observe in NLR spectra allow us to discern some of the properties of the ionizing spectrum. Unlike the BLR, the electron densities in the NLR are low enough that many forbidden transitions are not collisionally suppressed. This allows us to use the intensity ratios of certain pairs of forbidden lines to measure the electron densities and temperatures in the NLR gas (§6.2). In comparison to the BLR, the analysis is simplified by the low densities. On the other hand, however, in the case of the NLR an additional complication is introduced into the spectroscopic analysis by significant amounts of dust since the NLR arises outside the dust sublimation radius (eq. 4.15); indeed, it may well be that the radius where dust sublimates provides the fundamental demarcation between the BLR and the NLR.

In contrast to what was believed for the first twenty years of AGN studies, the continuum spectra of AGNs are quite complex. As mentioned in Chapter 1, at least to a low-order approximation the SED of AGNs can be described as a power law of the form Fv ∝ v-α, where α is generally between zero and unity. This led to the initial suspicions that this continuum is non-thermal in origin. It is certainly tempting to attribute the bulk of an AGN spectrum to synchrotron emission, primarily because of the broadband energy characteristics of the emission and because of the similarity of the spectra to known synchrotron sources such as supernova remnants and extended radio sources. By the end of the 1970s, the best working model to produce the broad-band continuum was the synchrotron self-Compton (SSC) mechanism. Given a power-law distribution of energies, relativistic electrons in a magnetic field can produce a synchrotron power law spectrum over many decades of frequency. Moreover, it is possible in principle to produce the higher-energy emission, all the way up to X-rays, via the SSC process; the SSC process becomes important when the synchrotron radiation density becomes sufficiently high that the emitted photons are inverse-Compton scattered off the very electrons which are responsible for the synchrotron radiation. The major difficulty in understanding whether a particular source of radiation is pure synchrotron emission or SSC is that SSC-boosted photons have the same relative energy distribution as the original photons, thus providing no unique indication of the process.

One of the main goals of QSO research is to use these objects as a probe of the history of the Universe. Two specific aims are first, to determine the characteristics of the QSO population as a function of redshift, and second, to find the lookback time at which QSOs first appeared, as this provides some measure of the time scale for galaxy formation in the early Universe. Both of these important aims require large and preferably unbiased samples of QSOs. In this chapter, we consider how large samples of QSOs might be obtained through various survey techniques, and how the samples we obtain might be affected by various biases.

The measurable quantity that will result from surveys is the QSO ‘surface density’ dN(F,z)/dΩ i.e., number of QSOs per unit solid angle (square degree) as a function of flux F and redshift z. From this, we can compute the ‘luminosity function’, which is the relative number of AGNs at a given luminosity, and the ‘space density’, which is the total number of sources per unit comoving volume! over some specified luminosity range – when the luminosity function is correctly normalized, the total space density is simply the integral of the luminosity function over its entire range.

The primary goal of QSO surveys then is to determine dN(F,z)/dΩ in an accurate and unbiased fashion. This is a difficult and complicated undertaking because QSOs are faint and their surface density is low; the total surface density of QSOs brighter than B = 21 mag is only ∼40deg-2.

As we pointed out in Chapter 3, the main problem with sustaining an active nucleus by gravitational accretion over its lifetime of at least 108 years is funneling enough mass into the nucleus. Removing a sufficient amount of angular momentum from the gas flowing into the nucleus requires breaking the azimuthal symmetry of the galaxy's gravitational potential. A clear way to do that is by gravitational interactions with other systems, as was originally suggested by Toomre and Toomre (1972) and by Gunn (1979). This provides motivation for examining the nearby environment of AGNs to see if indeed there is evidence for interactions with nearby galaxies. The two specific questions that we want to consider are:

1. What kinds of galaxies harbor AGNs? Are there any discernible differences between galaxies with active nuclei and those without them?

2. Does the presence or absence of companion galaxies have anything to do with whether or not a galaxy harbors an AGN?

We will consider these issues separately, although they are clearly related.

Host Galaxies

The study of the ‘host galaxies’, those galaxies that contain active nuclei, is a very difficult undertaking. The major problems were alluded to at the beginning of this book: the light from the AGN itself often dominates the total light from the galaxy, particularly in the case of the highest-luminosity AGNs, which are spatially rare and thus typically found only at great distances. Consequently the work on the lowerluminosity end of the AGN distribution, i.e., Seyfert galaxies, has tended to yield less ambiguous results.

Like many textbooks, this one arose out of the author's frustration. While I believe that there are many excellent journal articles, scholarly reviews, conference proceedings, and even a few advanced monographs on active galactic nuclei (AGNs), there is no single place where a beginning student can get the very basic background necessary to get the most out of the more research-oriented material. The aims of this book are thus actually twofold: first, I wanted to summarize our basic, if marginal, understanding of AGNs at what I believe is a level of familiarity that should be expected of doctoral-level students in astronomy, and second, I wanted to provide a fairly comprehensive introduction to AGNs that would serve as a gateway to the more specialized review articles and research literature for students who have research ambitions in the field. The intended audience is thus advanced undergraduate and beginning graduate students in astronomy and astrophysics. Fairly complete undergraduate preparation in physics is assumed, as is some basic understanding of extragalactic astronomy.

I have tried to focus on basic issues and avoid minutiae and arcane issues, even though some of these undoubtedly will turn out to be tremendously important in the future. I have attempted to compile the basic background material that is by and- large familiar to researchers in AGNs, although I caution that it is by no means complete: research-level competence in the field of AGNs will require a good deal more background than is given here.

This book had its origins in a workshop held in Cape Town from June 27 to 2 July 1994, with participants from South Africa, USA, Canada, UK, Sweden, Germany, and India. The meeting considered in depth recent progress in analyzing the evolution and structure of cosmological models from a dynamical systems viewpoint, and the relation of this work to various other approaches (particularly Hamiltonian methods). This book is however not a conference report. It was written by some of the conference participants, based on what they presented at the workshop but altered and extended after reflection on what was learned there, and then extensively edited so as to form a coherent whole. This process has been very useful: a considerable increase in understanding has resulted, particularly through the emphasis on relating the results of the qualitative analysis to possible observational tests. Apart from describing the development of the subject and what is presently known, the book serves to delineate many areas where the answers are still unknown. The intended readers are graduate students or research workers from either discipline (cosmological modeling or dynamical systems theory) who wish to engage in research in the area, tackling some of these unsolved problems.

The role of the two editors has been somewhat different.

In Section 5.1 we give an overview of the use of qualitative methods in analyzing Bianchi cosmologies, expanding on the brief remarks in the Introduction to the book. Section 5.2 provides an introduction to the use of expansion–normalized variables in conjunction with the orthonormal frame formalism, thereby laying the foundation for the detailed analysis of the Bianchi models with non–tilted perfect fluid source in Chapters 6 and 7. In Section 5.3 we discuss, from a general perspective, the use of dynamical systems methods in analyzing the evolution of Bianchi cosmologies, referring to the background material in Chapter 4.

Overview

As explained in Section 1.4.2 there are two main approaches to formulating the field equations for Bianchi cosmologies:

1. the metric approach,

2. the orthonormal frame approach.

In the metric approach the basic variables are the metric components gαβ(t) relative to a group–invariant, time–independent frame (see (1.89)). This approach was initiated by Taub (1951) in a major paper. After a number of years researchers became aware that the Bianchi models admitted additional structure, namely the automorphism group, which plays an important role in identifying the physically significant variables (also referred to as gauge–invariant variables, or the true degrees of freedom). This group is defined to be the set of time–dependent linear transformations (1.87) of the spatial frame vectors that preserve the structure equations (1.88).