Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Statistical Thermophysics, Robertson

Download as pdf or txt
Download as pdf or txt
You are on page 1of 594

Canonical Systems 117

where the index n enumerates the states of the composite system and Ecn is the energy
when the system is in state n. Any state counted in Zc must be an identifiable combination
state of the composite system and not an alternative or incomplete representation of an
improperly defined composite state.
As an illustration, suppose Z 1 and Z2 in Eqs.(3.3.12) and (3.3.13) have just three states
each; let

Then the composite Z 12 of Eq. (3.3.15) becomes

or
Z12 = a1b 1 + a1b2 + a1b3 + a2 b1 + a2 b2 + a2b3 + a3b 1 + a3b2 + a3b3,
for nine composite states. But if the subsystems are identical and indistinguishable, then
ai = bi and the nine terms can be written as ai + 2a 1 a2 + a� + 2a1a3 + a� + 2a2 a3. A term
such as 2a 1 a2 arises as a1b2 + a2 b1, with ai = bi. In quantum theory, however, the state
of the system, is properly described as a two-particle state with states 1 and 2 occupied,
but by particles with no individuality, so it cannot be said that one particle is in state 1
and the other in state 2. The formalism says that each particle is in both single-particle
states, and that therefore only one such state exists. (This point is discussed further in the
commentary of this chapter, in Chapter 4, and in Appendix C.) Thus the proper partition
function should be
Zc = ai + a1a2 +a�+a 1 a3 +a;+a2 a3,
with only six (single) terms, rather than nine (counting the double ones).
If, instead of two such systems, there are N, then the function corresponding to zf is

which generates the number of permutations of N elements into three groups, where
I:
( N; m 1, m 2, m3) is the multinomial coefficient and mi = N. The proper partition function
should be simply

L a;:n 1 a;1 2 a-;i3 , again with L mi= N,


m1,m2,m3

but such a thing is difficult to write for a general partition function with an infinite number
of energy levels. However, when the number of levels becomes infinite, (or much larger than
N) the single-system partition function to the Nth power will still be of the form

(3.3.16)

with I,::,1 mi N. Evidently, then, most of the mi will be zero, and the largest terms
in Eq. (3.3.9) are those for which all the mi are 1 or 0. This is just the regime for which

You might also like