I'm reading Strocchi's book on The Non-Perturbative Foundations of Quantum Field Theory.

In the chapter concerning point-splitting regularization, where the free Dirac current is defined as follows $$ j_\mu(x)\equiv \lim_{\epsilon\to0}\left[\bar\psi(x+\epsilon)\gamma_\mu\psi(x)-\langle\bar\psi(x+\epsilon)\gamma_\mu\psi(x)\rangle\right] $$ to give a precise meaning to the product of two distributions evaluated at the same point, there is the following claim about the spectral properties of the current commutator function: $$ \langle[j_\mu(x),j_\nu(y)]\rangle = (\Box g_{\mu\nu}-\partial_\mu\partial_\nu)\int d\rho(\mu^2) i\Delta(x-y;\mu^2) $$ where the $i\Delta(x-y;\mu^2)$ is the commutator function of a free scalar field o mass $\mu$, and $$ \rho(\mu^2)= \frac{1}{3(2\pi)^2}\sqrt{1-\frac{4m^2}{\mu^2}}\left(1+\frac{2m^2}{\mu^2}\right). $$ The hint is to calculate this spectrum by inserting a complete set of states of electron-positron pairs. How can I do this?

I tried insertig them in the 2-point function obtaining: (I keep the spin label implicit in the sum over the complete set) $$ \langle j_\mu(x) j_\nu(y)\rangle= \int d\mu^2 \int \frac{d^3k}{(2\pi)^3}\frac{1}{2\omega_\mu(\mathbf{k})} \langle j_\mu(x)|\mathbf{k};\mu^2\rangle\langle\mathbf{k};\mu^2| j_\nu(y)\rangle=\\ =\int d\mu^2 \int \frac{d^3k}{(2\pi)^3}\frac{1}{2\omega_\mu(\mathbf{k})}e^{-ik\cdot(x-y)} \langle j_\mu(0)|\mathbf{k};\mu^2\rangle\langle\mathbf{k};\mu^2| j_\nu(0)\rangle $$ now, I can consider a Lorentz boost $\Lambda_\mathbf{k}$ such that it brings us in the rest frame of the single particle state $|\mathbf{k};\mu^2\rangle$: $U(\Lambda_\mathbf{k})|\mathbf{k};\mu^2\rangle=|\mathbf{0};\mu^2\rangle$, and by covariance of the (vector) current $j_\mu$ we have $$ \langle j_\mu(x) j_\nu(y)\rangle= \int d\mu^2 \langle j_\rho(0)|\mathbf{0};\mu^2\rangle\langle\mathbf{0};\mu^2| j_\lambda(0)\rangle \int \frac{d^3k}{(2\pi)^3}\frac{1}{2\omega_\mu(\mathbf{k})}e^{-ik\cdot(x-y)}(\Lambda_\mathbf{k})^\rho_\mu(\Lambda_\mathbf{k})^\lambda_\nu. $$ Assuming that the above structure is correct, since the differential operator $\Box g_{\mu\nu}-\partial_\mu\partial_\nu$ is dictated by Lorentz covariance and the current conservation, we have indeed: $$ \langle j_\mu(x) j_\nu(y)\rangle=(\Box g_{\mu\nu}-\partial_\mu\partial_\nu)\int d\rho(\mu^2) i\Delta^+(x-y;\mu^2)=\\ =(\Box g_{\mu\nu}-\partial_\mu\partial_\nu)\int d\rho(\mu^2) \int \frac{d^3k}{(2\pi)^3}\frac{1}{2\omega_\mu(\mathbf{k})}e^{-ik\cdot(x-y)}=\\ =\int d\rho(\mu^2) \int \frac{d^3k}{(2\pi)^3}\frac{-\mu^2 g_{\mu\nu}+k_\mu k_\nu}{2\omega_\mu(\mathbf{k})}e^{-ik\cdot(x-y)} $$ and tracing and equating both expressions we get, since $\Lambda_\mu^\rho g^{\mu\nu}\Lambda_\nu^\lambda=g^{\rho\lambda}$: $$ \rho(\mu^2)=-\frac{g^{\rho\lambda}}{3\mu^2}\langle j_\rho(0)|\mathbf{0};\mu^2\rangle\langle\mathbf{0};\mu^2| j_\lambda(0)\rangle. $$ My problem is here: the matrix element $$ \langle j_\rho(0)|\mathbf{0};\mu^2\rangle $$ seems to vanish identically since there are three annihilation/creation operator in a vacuum expectation value! What did I miss?


Ok, I guess I must sum over particle-antiparticle intermediate states and not just single particle states, which lead the v.e.v. to vanish identically.

Indicating the invariant phase space element as $$ \int d\Pi(\mathbf{k})\equiv\int \frac{d^3k}{(2\pi)^3}\frac{1}{2\omega_m(\mathbf{k})} $$ we get, letting $\mathbf{k}$ and $\mathbf{p}$ be the momentum respectively of electron and positron and $q^2$ the square of the center of mass energy $$ \langle j_\mu(x) j_\nu(y)\rangle= \int dq^2 \int d\Pi(\mathbf{k})\int d\Pi(\mathbf{p}) \langle j_\mu(x)|\mathbf{k},\mathbf{p};q^2\rangle\langle\mathbf{k},\mathbf{p};q^2| j_\nu(y)\rangle $$ now, I can consider a Lorentz boost $\Lambda$ that brings us in the center of mass frame of the particle-antiparticle state and by covariance of the (vector) current $j_\mu$ we have $$ \langle j_\mu(x) j_\nu(y)\rangle= \int dq^2 \langle j_\rho(0)|\mathbf{p},-\mathbf{p};q^2\rangle\langle\mathbf{p},-\mathbf{p};q^2| j_\lambda(0)\rangle \frac{|\mathbf{p}|}{(2\pi)^24E_{CM}} \int \frac{d^3P}{(2\pi)^3}\frac{1}{2\omega_m(\mathbf{P})}e^{-iP\cdot(x-y)}(\Lambda)^\rho_\mu(\Lambda)^\lambda_\nu, $$ where we have performed a simplification in the 2-body phase space integral (Peskin, page 107), even though I'm not sure on how to get rid of the integration over solid angle $d\Omega$, which usually enters in the definition of cross section.

By tracing with $g^{\rho\lambda}$ we get, comparing to the general expression above for the 2-point fuction: $$ \rho(q^2)=-\frac{1}{3q^2}\frac{|\mathbf{p}|}{(2\pi)^24E_{CM}}\langle j_\rho(0)|\mathbf{p},-\mathbf{p};q^2\rangle\langle\mathbf{p},-\mathbf{p};q^2| j^\rho(0)\rangle; $$ with a bit of Dirac algebra: letting $p_1=(\omega(\mathbf{p}),\mathbf{p})$, $p_2=(\omega(\mathbf{p}),-\mathbf{p})$ $$ \langle j_\rho(0)|\mathbf{p},-\mathbf{p};\mu^2\rangle\langle\mathbf{p},-\mathbf{p};\mu^2| j^\rho(0)\rangle=-\text{Tr}\left[(\hat p_1+m)\gamma_\rho(\hat p_2-m)\gamma^\rho\right]=-8(2m^2+p_1\cdot p_2) $$ and since $$ |\mathbf{p}|=\sqrt{\omega(\mathbf{p})^2-m^2}=\frac{E_{CM}}{2}\sqrt{1-\frac{4m^2}{q^2}} $$ we get $$ \rho(q^2)=-\frac{1}{3q^2}\frac{\frac{E_{CM}}{2}\sqrt{1-\frac{4m^2}{q^2}}}{(2\pi)^24E_{CM}}[-8(2m^2+p_1\cdot p_2)]=\frac{1}{6(2\pi)^2}\sqrt{1-\frac{4m^2}{q^2}}\left(\frac{2m^2}{q^2}+1\right) $$ which looks good a part from the extra factor 2 in the denominator.


It does vanish and it should, because in a non-broken symmetry phase the current operator cannot create a single particle state.

You should have summed over all intermediate states, including the vacuum and all multi particle states, whereas you seem to have only inserted a single particle state.

| cite | improve this answer | |
  • $\begingroup$ Mmh, you're right, I should have understood it since the hint is to sum over intermediate particle-antiparticle states. I'll try $\endgroup$ – Brightsun Apr 11 '15 at 20:22
  • $\begingroup$ Tried applying your suggestion, if you have a moment could you please tell me your opinion now? thanks $\endgroup$ – Brightsun Apr 12 '15 at 10:05

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.