04.08.2013 Views

Statistical Mechanics - Physics at Oregon State University

Statistical Mechanics - Physics at Oregon State University

Statistical Mechanics - Physics at Oregon State University

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

6.1. DENSITY OPERATORS. 121<br />

The partition function is<br />

Z = <br />

n<br />

pn = 1 En<br />

k e− B T (6.8)<br />

Z<br />

En<br />

−<br />

e<br />

k B T = <br />

n<br />

En<br />

− k 〈n| e B T |n〉 (6.9)<br />

Now we use the fact th<strong>at</strong> the st<strong>at</strong>es in the sum are eigenst<strong>at</strong>es of the Hamiltonian,<br />

and we obtain<br />

Z = <br />

n<br />

H − k 〈n| e B T −<br />

|n〉 = Tr e H<br />

kB T (6.10)<br />

At this point we will switch to the more common not<strong>at</strong>ion for the inverse<br />

. The density oper<strong>at</strong>or is given by<br />

temper<strong>at</strong>ure, β = 1<br />

kBT<br />

ρ = 1<br />

Z<br />

<br />

e −βEn |n〉 〈n| (6.11)<br />

n<br />

and using the fact again th<strong>at</strong> we expand in eigenst<strong>at</strong>es of the Hamiltonian:<br />

ρ = 1<br />

Z e−βH<br />

(6.12)<br />

These definitions are basis-dependent. The only question is how to define the<br />

exponent of an oper<strong>at</strong>or, but th<strong>at</strong> turns out to be easy since the power series<br />

for exponents always converge uniformly on any finite interval. Hence the ex-<br />

ponent of an oper<strong>at</strong>or O is defined by e O = ∞<br />

n=0 1<br />

n! On . The m<strong>at</strong>hem<strong>at</strong>ical<br />

filed of functional analysis, very useful for physicists, gives many more details.<br />

Products of oper<strong>at</strong>ors are well defined. Therefore, the density oper<strong>at</strong>or can be<br />

specified without knowing the eigenst<strong>at</strong>es of the Hamiltonian. This is definitely<br />

an advantage. Note, however, th<strong>at</strong> in many cases the most practical manner<br />

to do calcul<strong>at</strong>ions is by using eigenvalues. But there are certainly other cases<br />

where it is better not to do so. Finally, the expect<strong>at</strong>ion value of an oper<strong>at</strong>or A<br />

in this formalism is<br />

< A >= Tr (ρA) = Tr (e−βH A)<br />

Tr (e −βH )<br />

(6.13)<br />

This is an important form for expect<strong>at</strong>ion values. It can be used in diagramm<strong>at</strong>ic<br />

techniques in the same manner as Feynman diagrams. The only substitution<br />

we need to make is β → ıt. We replace inverse temper<strong>at</strong>ure by imaginary time.<br />

Now wh<strong>at</strong> does th<strong>at</strong> mean???<br />

Formulas (6.10) and (6.12) contain functions of oper<strong>at</strong>ors. Elementary functions<br />

of an oper<strong>at</strong>or are positive integral powers of an oper<strong>at</strong>or. For example,<br />

O 2 |x〉 = O |y〉 with |y〉 = O |x〉. Sums and differences of oper<strong>at</strong>ors are also<br />

easily defined. More complic<strong>at</strong>ed functions of oper<strong>at</strong>ors can be defined through<br />

power series, just like functions of a complex variable in complex analysis. These<br />

power series often have a certain radius of convergence, and expressed in terms

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!