Annu. Rev. Astron. Astrophys. 1984. 22: 471-506 Copyright © 1984 by Annual Reviews. All rights reserved

9. SOME CONCLUDING COMMENTS

The foregoing sections have discussed some physical processes and some idealized models that are relevant to the general phenomenon of active galactic nuclei. It is, however, depressingly evident how tenuous are the links between these models and the actual observations. Partly, this is because the subject is just beginning; but it is also partly because the observations relate only very indirectly to the primary energy source - they may instead tell us about secondary reprocessing that has occurred on much larger scales.

Exhortations and hopes for the future can be summarized in three categories:

1. On the purely theoretical level, even the simple "toy" models discussed here need further investigation - they involve effects in Kerr geometry, collective processes and radiative transfer in pair-dominated plasma, and acceleration of high gamma particles, none of which are yet well explored or understood. We need to clarify the stability of the various axisymmetric configurations: this should narrow down the embarrassing freedom we now have in specifying the angular momentum and the enthalpy distribution in tori. Large-scale computer simulations could be crucial here.

Computer simulations should permit us also to relax the assumption of stationarity (59), which has been implicit in most work on accretion flows. It may be more realistic to envisage that the feeding process and the subsequent viscous redistribution of angular momentum and drainage into the hole are sporadic. There is, after all, observational evidence for variability on all time scales. Three-dimensional gas-dynamical codes could also check whether the Lense-Thirring effect does indeed align the flow in the way simple arguments suggest. A further valuable computational development will be the advent of MHD codes able to treat electromagnetic processes around black holes, as well as the initiation (and possible magnetic confinement) of relativistic jets.

Detailed computations would also be worthwhile on other classes of relativistic systems relevant to the evolutionary tracks in Figure 1 (20). In particular, supermassive stars with realistic differential rotation should be investigated. For a suitable angular momentum distribution, these could acquire a high gravitational binding energy [cf. the massive disks (15, 75) that have been treated analytically]. Redistribution of angular momentum within such objects would be likely to cause their inner regions to collapse, leaving a massive torus around the resultant black hole. If too massive, this would be subject to gravitational instability and could fragment. Otherwise, it would evolve on a Kelvin or viscous time scale (whichever was shorter). Such models also remind us that evolution need not be restricted to the slow time scales of order t_E (Equation 4), but that rare "hypernovae" may Occur.

2. The "peripheral fuzz" at r >> r_g in the emission-line region and the radio structures involves physics that is less extreme and more familiar than that in the central relativistic domain. However, it is here that one is perhaps more pessimistic about theoretical progress. This is because in the central region, even though the physics may be exotic, we have a relatively "clean" problem: axisymmetric flow in a calculable gravitational field. On the other hand, in the large-scale sources, environmental effects are plainly crucial: progress will be slow, for the same reasons that weather prediction is difficult.

The subject has proceeded in a highly compartmentalized way: the central engine, the emission-line region, the radio jets, etc., are modeled somewhat disjointly. To a certain extent this is inevitable - after all, the relevant scales may differ by many powers of ten. As data proliferate on source morphology, it no longer seems premature to develop more comprehensive models, nor to understand the relation of AGNs to their parent galaxies: If we compare spirals and ellipticals, are the central masses different? Is the fueling different? What other environmental influences determine the kind of AGN that is observed? And do stellar-mass compact objects within our own Galaxy offer many clues to the mechanisms of AGNs?

3. It is perhaps salutary (especially for relativists) to remain aware that Einstein's theory is empirically validated only in the weak-field limit. An extra motive for studying the central region is therefore to seek a diagnostic (by refining our models for galactic nuclei) that could test strong-field general relativity and check whether the space-time around a rotating black hole is indeed described by the Kerr metric.

Ginzburg (53) has recently remarked on how surprisingly slowly most sciences develop. Concentrated activity over a short time-base may give the illusion that progress is fast, but the advance of science - particularly where data are sparse - displays a slowly rising trend, with large-amplitude "sawtooth" fluctuations superposed on it as fashions come and go. There has been progress toward a consensus, in that some bizarre ideas that could be seriously discussed a decade ago have been generally discarded. But if we compare present ideas with the most insightful proposals advanced when quasars were first discovered 20 years ago (such proposals being selected, of course, with benefit of hindsight), progress indeed seems meager. It is especially instructive to read Zeldovich & Novikov's (1964) paper entitled "The Mass of Quasi-Stellar Objects" (133). In this paper, on the basis of early data on 3C 273, they conjectured the following: (a) Radiation pressure perhaps balances gravity, so the central mass is ~ 10⁸ M. (b) For a likely efficiency of 10%, the accretion rate would be 3M yr^-1. (c) The radiation would come from an effective "photosphere" at a radius ~ 2 × 10¹⁵ cm (i.e. >> r_g), outside of which line opacity would cause radiation to drive a wind. (d) The accretion may be self-regulatory, with a characteristic time scale of ~ 3 yr. These suggestions accord with the ideas that remain popular today, and we cannot yet make many firmly based statements that are more specific.

ACKNOWLEDGMENTS

I am grateful for discussions and collaboration with many colleagues, especially M.C. Begelman, R.D. Blandford, A.C. Fabian, and E.S. Phinney.