Ensembles 63The idea that the macroscopic observables of a system are not sensitive to pre-
cise microscopic details is the basis of theensembleconcept originally introduced by
Gibbs. More formally, anensembleis a collection of systems described by the same
set of microscopic interactions and sharing a common set of macroscopic properties
(e.g. the same total energy, volume, and number of moles). Each system evolves under
the microscopic laws of motion from a different initial condition so thatat any point
in time, every system has a unique microscopic state. Once an ensemble is defined,
macroscopic observables are calculated by performing averages over the systems in
the ensemble. Ensembles can be defined for a wide variety of thermodynamic situ-
ations. The simplest example is a system isolated from its surroundings. However,
ensembles also describe systems in contact with heat baths, systems in contact with
particle reservoirs, systems coupled to pressure control mechanisms such as mechan-
ical pistons, and various combinations of these influences. Such ensembles are useful
for determining static properties such as temperature, pressure, free energy, average
structure, etc. Thus, the fact that the systems in the ensembleevolve in time does
not affect properties of this type, and we may freeze the ensembleat any instant and
perform the average over the ensemble at that instant. These ensembles are known
asequilibrium ensembles, and we will focus on them up to and including Chapter 12.
Finally, ensembles can also be defined for systems driven by external forces or fields
for the calculation of transport coefficients and other dynamical properties. These are
examples of non-equilibrium ensembles, which will be discussed in Chapters 13 and
14.
In classical ensemble theory, every macroscopic observable of a system is directly
connected to a microscopic function of the coordinates and momenta of the system.
A familiar example of this comes from the kinetic theory, where the temperature of
a system is connected to the average kinetic energy. In general, we will letAde-
note a macroscopic equilibrium observable anda(x) denote a microscopic phase space
function that can be used to calculateA. According to the ensemble concept, if the
ensemble hasZmembers, then the “connection” betweenAanda(x) is provided via
an averaging procedure, which we write heuristically as
A=
1
Z
∑Z
λ=1a(xλ)≡〈a〉. (2.3.1)This definition is not to be taken literally, since the sum may well be a continuous
“sum” or integral. However, eqn. (2.3.1) conveys the notion that the phase space
functiona(x) must be evaluated for each member of the ensemble at that point in
time when the ensemble is frozen. Finally,Ais obtained by performing an average
over the ensemble. (The notation〈a〉in eqn. (2.3.1) will be used throughout the book
to denote an ensemble average.)
Let us recall the question we posed earlier: If we could solve the equations of motion
for a very large number of particles, would the vast amount of detailed microscopic
information generated be necessary to describe macroscopic observables? Previously,
we answered this in the negative. However, the other side can also be argued if we
take a purist’s view. That is, all of the information needed to describe a physical
system is encoded in the microscopic equations of motion. Indeed, there are many