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.

130 CHAPTER 6. DENSITY MATRIX FORMALISM.<br />

Now we can interchange factors and we find <strong>at</strong> the extremum th<strong>at</strong><br />

or<br />

∂<br />

∂x eτ(x) =<br />

∞<br />

n=1<br />

1<br />

(λ − 1)n−1<br />

(n − 1)!<br />

<br />

∂ρ ∂ log(ρ)<br />

= ρ<br />

∂x ∂x<br />

Using x = ρij we get <strong>at</strong> the extremum<br />

<br />

∂ρ ∂ log(ρ)<br />

= ρ<br />

∂ρij<br />

∂ρij<br />

<br />

∂τ<br />

= e<br />

∂x<br />

λ−1<br />

<br />

∂τ<br />

∂x<br />

and we can now take the element nm of this oper<strong>at</strong>or equ<strong>at</strong>ion to get<br />

<br />

∂ρmn ∂ log(ρmn)<br />

= ρ<br />

∂ρij<br />

∂ρij<br />

(6.65)<br />

(6.66)<br />

(6.67)<br />

(6.68)<br />

The left side is easy, and equal to δmiδnj. Hence <strong>at</strong> the extremum we have<br />

2 ∂ X<br />

<br />

= −kBe 1−λ δmiδnj<br />

(6.69)<br />

∂ρij∂ρnm<br />

Suppose we make an arbitrary small change in the density m<strong>at</strong>rix around<br />

the extremum. In second order we have<br />

∆X = <br />

2 ∂ X<br />

or<br />

ijmn<br />

∂ρij∂ρnm<br />

<br />

<br />

1−λ<br />

∆ρij∆ρnm = −kBe<br />

∆X = −kBe<br />

ijmn<br />

<br />

1−λ<br />

∆ρij∆ρji<br />

ij<br />

δmiδnj∆ρij∆ρnm (6.70)<br />

(6.71)<br />

Now we use the fact th<strong>at</strong> the density m<strong>at</strong>rix has to remain Hermitian, and hence<br />

∆ρ∗ ij = ∆ρji to get<br />

<br />

1−λ<br />

|∆ρij| 2<br />

(6.72)<br />

∆X = −kBe<br />

which shows th<strong>at</strong> the extremum is a maximum, indeed.<br />

Why bother?<br />

The whole approach sketched above seems very complic<strong>at</strong>ed, and for the<br />

simple microcanonical ensemble (st<strong>at</strong>es with given U, V, N) it is indeed. The<br />

advantage is th<strong>at</strong> it is easier to generalize for more complex systems. Th<strong>at</strong> is<br />

where we gain in this formalism. The only thing we have shown above is th<strong>at</strong><br />

ij

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

Saved successfully!

Ooh no, something went wrong!