MODERN COSMOLOGY

Explaining homogeneity and structure 147

equation of state: at early enough times, the equation of state was such as to
cause smoothing on all scales; but at later times, it was such as to cause structure
growth on particular scales. The inflationary scenario, and the models that build
on it, are remarkably successful in this regard, particularly through predicting the
CBR anisotropy patterns (the ‘Doppler peaks’) which seem to have been found
now (but significant problems remain, particularly as regards compatibility with
the well-established nucleosynthesis arguments).
Given these astrophysical and physical processes, explanation of the large-
scale isotropy and homogeneity of the universe together with the creation
of smaller-scale structures means determining the dynamical evolutionary
trajectories relating initial to final conditions, and then essentially either (a)
explaining initial conditions or (b) showing they are irrelevant.

3.9.1 Showing initial conditions are irrelevant

This can be attempted in a number of different ways.

3.9.1.1 Initial conditions are irrelevant because they are forgotten

Demonstrating minimal dependence of the large-scale final state on the initial
conditions has been the aim of

• thechaotic cosmologyprogramme of Misner, where physical processes such
  a viscosity wipe out memories of previous conditions [97]; and


• theinflationary family of theories, where the rapid exponential expansion
  driven by a scalar field smooths out the universe and so results in similar
  memory loss [79].

The (effective) scalar field is slow-rolling, so the energy condition (3.36) is
violated and a period of accelerating expansion can take place through many e-
foldings, until the scalar field decays into radiation at the end of inflation. This
drives the universe model towards flatness, and is commonly believed to predict
that the universe must be very close indeed to flatness today, even though this is
an unstable situation, see the phase planes of
againstS[94]. It can also damp
out both anisotropy, as previously explained and inhomogeneity, if the initial
situation is close enough to a FL model of that inflation can in fact start. In
a chaotic inflationary scenario, with random initial conditions occurring at some
initial time, inflation will not succeed in starting in most places, but those domains
where it does start will expand so much that they will soon be the dominant feature
of the universe: there will be many vast FL-like domains, each with different
parameter values and perhaps even different physics, separated from each other
by highly inhomogeneous transition regions (where physics may be very strange).
In the almost-FL domains, quantum fluctuations are expanded to a very large scale
in the inflationary era, and form the seeds for structure formation at later times.
Inflation then goes on to provide a causal theory of initial structure formation