0
$\begingroup$

This is a question on computation. If we set the quantum field canonical commutation relations as

$$ [\hat{\pi}(x),\hat{\phi}(x')] = -i\hbar \delta(x-x'), $$

the goal is to find the Fourier transform of this commutation relation:

$$ [\hat{\pi}_k,\hat{\phi}_{k'}] = -i\hbar \delta_{k,k'}. $$

My Fourier transforms are defined as

$$ \hat{\phi}_k = \frac{1}{L^{1/2}}\int_0^L dx e^{-ikx} \hat{\phi}(x), $$

$$ \hat{\pi}_k = \frac{1}{L^{1/2}}\int_0^L dx e^{ikx} \hat{\pi}(x). $$

Direct substitution yields

$$ [\hat{\pi}_k,\hat{\phi}_k'] = \frac{1}{L}\int_0^Ldx\int_0^L dx' e^{ikx-ik'x'}[\hat{\pi}(x),\hat{\phi}(x')] = \frac{-i\hbar}{L}\int_0^L dx e^{ix(k-k')}, $$

where this last integral is a delta function. My question is, is it true that

$$ \int_0^L\! dx ~e^{ix(k-k')} ~=~ L~\delta_{k,k'}? $$

(I get always confused with the normalization constants.)

$\endgroup$
  • 1
    $\begingroup$ The commutator you start with are usually called the "equal time commutation relations", because most field theorists use the Heisenberg or interaction pictures. $\endgroup$ – Sean E. Lake Dec 6 '17 at 0:03
1
$\begingroup$

Yes, $\int_0^L dxe^{ix(k-k')} = L\delta_{k,k'}$ which is the Kroenecker delta for discrete Fourier series. For true Fourier transform (where $L\rightarrow\infty$), the relation holds true but the Kroenecker $\delta_{k,k'}$ delta gets replaced by a Dirac delta $\delta(k-k')$.

Proof for discrete Fourier series (since you defined your transform with definite integrals):

$$\int_0^L dxe^{i(k-k')x} = \frac{e^{i(k-k')L} -1}{i(k-k')}$$ Now, for Fourier series, this transform is defined only for $k=\frac{2\pi n}{L}$ for any $n\in\mathbb{N}$. Thus, if $k=k'$, the integral just gives $L=L\delta_{k,k'=k}$ ($\delta_{k,k'}=1$ if $k=k'$ and $0$ if not). For $k\neq k'$, we get:

$$ \frac{e^{i(k-k')L} -1}{i(k-k')} = L\frac{e^{2\pi i(n-n')} -1}{2\pi i(n-n')}=0=L\delta_{k,k'\neq k} $$

since $e^{2\pi in}=1$ for any $n\in\mathbb{N}$. Thus you have your result:

$$\int_0^L dxe^{i(k-k')x} = L\delta_{k,k'}$$ for any $k$.

For the continuous case the identiy holds but you replace the Kroenecker delta by a Dirac delta (proof not necessarily obvious). Since it is beyond the scope of this question I won't draw it here :).

$\endgroup$

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.