The edge of a fractional quantum Hall state is an example of a chiral Luttinger liquid. Take, for the sake of simplicity, the edge of the Laughlin state. The Hamiltonian is:

$$H = \frac{2\pi}{\nu}\frac{v_c}{2} \int_{\textrm{edge}} dx \rho(x)^2 $$

Here $\nu$ is the filling fraction, which is constant, $v_c$ is the velocity of the edge mode and $\rho$ is the charge density operator. You can think of this Hamiltonian as a delta-function interaction $V(x,x') = \delta(x-x')$.

Together with this Hamiltonian there is also the commutation relation of the field $\rho$:

$$ [\rho(x),\rho(x')] = i\frac{\nu}{2\pi} \partial_x \delta(x-x')$$

I haven't gone through the exercise myself, but I presume this is derived by going to momentum space, obtaining the canonical momenta through Hamilton's equation of motion and performing a canonical quantization. These equal-time commutation relations together with the Heisenberg equations of motion lead to:

$$\partial_t \rho(x,t) = i[H,\rho(x,t)] = v_c \partial_x\rho(x,t) $$

This demonstrates that the edge is chiral, since $(\partial_t - v_c\partial_x)\rho(x,t) = 0$ and therefore the correlator involving $\rho(x,t)$ (or any other correlator) is a function of $t+x/v_c$ alone (hence the name "chiral" and "left-moving").

There are also particle excitations (e.g. the electron) which are generated through vertex operator $\Psi(x,t)$ (these can be motivated through bosonization and/or conformal field theory, but I won't go into that). In any case these field operators have the following equal-time commutation relations with the current operator:

$$[\rho(x),\Psi(x')] = Q \Psi(x')\delta(x-x')$$

Here $Q$ is the charge of the operator $\Psi$ with respect to the charge density operator $\rho$. This charge is then of course the electric charge.

My question is now: how do you generalize the commutation relations to non-equal time? What is:

$$[\rho(x,t),\rho(x',t')] = \ldots $$ $$[\rho(x,t),\Psi(x',t')] = \ldots $$



You may simply replace the arguments $ x \to x+vt $ in all of those commutators. The way to prove that is to go to momentum space, where $\rho (k, t) = \rho (k,0) e^{i k v t} $.

Proving that about the commutator $[\rho (x,t),\psi (x',t')]$ is a little trickier, because we don't know the explicit form of $\psi (t)$, but the way I did it was to use $$ [\rho (x,t),\psi (x',t')] = U^\dagger (t') [\rho (x,t-t'),\psi (x',0)] U (t') $$ Where U is the time evolution operator. This way only the time evolution of the density is needed.

  • $\begingroup$ Fantastic. It's so elegant I wonder why I couldnt come up with it. I went through the whole thing and it works out likes you said ($x\rightarrow x+vt$). For what it's worth: the field operators take on the form $\Psi(x) \sim exp(i \alpha \int_{-\infty}^x \rho(x))$, so the relation is carried over automatically. $\endgroup$ – Olaf Aug 26 '11 at 14:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.