08.04.2013 Views

Quantum Information Theory with Gaussian Systems

Quantum Information Theory with Gaussian Systems

Quantum Information Theory with Gaussian Systems

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.2 Reversible <strong>Gaussian</strong> qca<br />

case, which corresponds to sin α(k) = 0. The imaginary part is readily obtained<br />

from (4.23) as ImPk = cosα(k) −Γ(k) / 2 sinα(k) , which proves (4.22). For the<br />

remaining projector, we get Re Pk = RePk and ImPk = − Im Pk from the beginning<br />

of the proof. Hence Pk is indeed the complex conjugate of Pk. <br />

Convergence<br />

The decomposition (4.21) of ˆ Γ(k) is particularly useful for a compact description of<br />

the iterated transformation ˆ Γt(k). By (4.15),<br />

ˆΓt(k) = ˆ Γ(k) t = e itα(k) Pk + e −itα(k) Pk , (4.24)<br />

since as projectors on disjoint eigenspaces Pk and Pk obey P 2 k = Pk , P 2<br />

k = Pk and<br />

Pk · Pk = 0. With this relation, the time-dependent correlation function γt(x) is<br />

obtained by inverse Fourier transform from (4.17) as<br />

γt(x) = 1<br />

2π<br />

= 1<br />

2π<br />

<br />

π<br />

−π<br />

π<br />

<br />

−π<br />

+ 1<br />

2π<br />

dk e ikx ˆ Γ T<br />

t (k) · γ0(k) · ˆ Γ t(k)<br />

<br />

ikx<br />

dk e e 2itα(k) P T<br />

k · γ0(k) · Pk + e −2itα(k) <br />

T<br />

Pk · γ0(k) · Pk <br />

π<br />

−π<br />

<br />

ikx<br />

dk e P T<br />

<br />

T<br />

k · γ0(k) · Pk + Pk · γ0(k) · Pk .<br />

(4.25)<br />

In (4.25), the transformation is separated into a time-dependent, oscillating part<br />

in the first term and a stationary part in the second. In the limit of large time<br />

t, the rapidly oscillating term vanishes and the correlation function converges by<br />

an argument similar to the method of stationary phase: Starting from a product<br />

state (or any clustering state), γ0(k) is continuous; since we excluded the degenerate<br />

case, ˆΓ(k), Pk and Pk are continuous, too, and the whole integrand is well-behaved.<br />

Note that α(k) is differentiable and has only finitely many extrema (cf. Fig. 4.3<br />

and caption). The main contribution to the integral stems from intervals where<br />

α ′ (k) ≈ 0, i.e. from around the extrema of α(k) at kn, ordered such that kn ≤ kn+1<br />

for n = 1, 2, . . .N < ∞. We explain the argument for the first term in the first<br />

integrand of (4.25). Splitting the integral at the extrema of α(k) at kn and writing<br />

ĉ(k) = P T<br />

k · γ0(k) · P k , we obtain:<br />

π<br />

dk ĉ(k)exp ikx + 2 itα(k) =<br />

−π<br />

N<br />

n=1<br />

kn−ǫ<br />

kn+ǫ <br />

dk ĉ(k)exp ikx + 2 itα(k) <br />

N<br />

+<br />

<br />

≡A<br />

n=1<br />

<br />

dk ĉ(k)exp ikx + 2 itα(k) <br />

+R<br />

kn+1−ǫ<br />

kn+ǫ<br />

<br />

≡B<br />

85

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

Saved successfully!

Ooh no, something went wrong!