Processing math: 100%

Saturday, October 01, 2011

$#*! I unlearned #1: scalar field theory

The Hamiltonian:

H=12d3x[π2+(ϕ)2+m2ϕ2]
where π(x) is the conjugate momentum density of field ϕ(x) is quadratic. In particular by going to momentum space with:
ϕ(x)=d3p(2π)312ωp(ap+ap)eipxπ(x)=d3p(2π)3(i)ωp2(apap)eipx
H become just sum of uncouple oscillators:
H=d3p(2π)3ωp(apap+12[ap,ap])
with well-known spectrum and stuff. More important are these commutation relations:
[H,ap]=ωpap,[H,ap]=ωpap

The time dependent of ϕ and π are obtained in the usual way:

ϕ(x)=ϕ(x,t)=eiHtϕ(x)eiHt.
The usual commutation relations hold at any time t:
[ϕ(x,t),π(y,t)]=δ3(xy),[ϕ(x,t),ϕ(y,t)]=[π(x,t),π(y,t)]=0.
Using the Heisenberg EQO iQt=[Q,H], we get iϕt=iπ and iπt=i(2m2)ϕ, which combine to give:
2t2ϕ=(2m2)ϕ,
the Klein-Gordon equation. Again writing in term of creation and annihilation operators:
ϕ(x)=d3p(2π)312p0(apeipx+apeipx)π(x)=ϕ(x)t
Note that the field operator ϕ is Hermitian, so this field is its own anti-particle.

The probability amplitude of creating a field at y and observing it in x is the propagator:

D(xy)=0|ϕ(x)ϕ(y)|0=d3p(2π)312p0eip(xy)
Consider a time-like interval: x0y0=t and xy=0:
D(xy)=4π(2π)30dpp22p2+m2eip2+m2t=14π2mdEE2m2eiEteimt
So we get an oscillating propagator. Nothing surprising here

Now consider a purely spacial separation: x0y0=0 and xy=r:

D(xy)=d3p(2π)3eipx2p0=i2(2π)2rdppeiprp2+m2
The integrand has two branch point at ±m, so to do this integral, we push the integral up and wrap around the upper branch cut. Let z=ipr, we obtain:
D(xy)=14π2r2mrdzzezz2(mr)2emr
at large r. The propagator seems to violate causality since it is finite for a space-like interval.

No comments: