While solving a QFT exercise, I'm trying to calculate the Feynman propagator $$ \underbrace{\partial_\mu \phi(x) \phi(y)} = \langle 0 \vert T { \partial_\mu \phi(x)\phi(y) } \vert 0\rangle $$ where the derivative acts over the $x$ variable.

In this task I encountered the problem of a relative minus sign between the two terms of the $T$-ordered product: \begin{equation}\tag{1} \underbrace{\partial_\mu \phi(x)\phi(y)} = \int \frac{d^3\mathbf p}{(2\pi)^3 2E_\mathbf{p}} \left(-ip_\mu \right)\left( \theta (x^0 - y^0) e^{-ip(x-y)} - \theta (y^0 - x^0)e^{ip(x-y)}\right) \end{equation} whereas my first guess would've been \begin{equation}\tag{2} \langle 0|T{\partial_{\mu}\phi(x)\phi (y)}|0\rangle =\int \frac{d^4p}{(2\pi)^4} (-ip_{\mu}) \cdot \frac{ i }{p^2-m^2+i\epsilon} e^{-ip(x-y)} \end{equation} leading to a Feynman contribution $$ \frac{ p_\mu }{p^2-m^2+i\epsilon} $$ to the scattering amplitude.

My first question therefore reduces to: are $(1)$ and $(2)$ somehow equivalent, and if so, how?

Furthermore, accepting the validity of $(2)$, the scalar propagator does not recognize any orientation, thus $p_\mu$ seems to be ill-defined, again, due to a sign. Actually, another arbitrariness rises from the choice of deriving with respect to $x$ instead of $y$, when using Wick Theorem.

The same problem arises in computing the vector (e.g. photon) propagator $$ \underbrace{ \partial_\mu A_\nu(x) A_\rho(y)} $$


1 Answer 1


Yes, (1) and (2) are equivalent. The issue of a relative minus sign is due to the derivative, assuming $\mu\neq 0$.

Eq. (1) is rather the intuitive expression for the time-ordered product of field operators $\partial_\mu\phi(x)$ and $\phi(y)$. (2) is a fancy way of writing the same thing, making use of contour integrals.

First, note that $p^0=\pm E_p$ is pole of the expression $1/(p^2-m^2)=1/((p^{0})^2-\mathbf p^2-m^2)$, so consider the integral

\begin{equation} \int_{-\infty}^{\infty} \frac{dp^0}{2\pi} \frac{i}{p^2-m^2+i\epsilon}e^{-ip\cdot (x-y)} \equiv \int_{-\infty}^{\infty} \frac{dp^0}{2\pi} \frac{i}{(p^0 -E_p+i\epsilon)(p^0+E_p-i\epsilon)} e^{-ip\cdot(x-y)} \end{equation}

This can be converted into a contour integral with the semicircular arc at infinity in the lower half complex plane for $x^0 - y^0> 0$ and in the upper half complex plane for $x^0 -y^0 < 0$. In the first case, the pole lying inside the contour is $p^0=E_p - i\epsilon$, so, by the residue theorem, the integral equates to

\begin{equation} \frac{e^{-iE_p(x^0-y^0)+i\mathbf p\cdot (\mathbf{x-y})}}{2E_p} \end{equation}

Note that the contour runs clockwise, so we incur an additional minus sign.

For the case $x^0 - y^0<0$, by the same procedure (contour anti-clockwise), we end up with

\begin{equation} \frac{e^{iE_p(x^0-y^0) +i\mathbf p\cdot (\mathbf{x-y})}}{2E_p} =\frac{e^{-iE_p(y^0-x^0)-i\mathbf p\cdot (\mathbf{y-x})}}{2E_p} \end{equation}

To put this in the standard form, replace $\mathbf p \rightarrow -\mathbf p$. The spatial integral $d^3p$ remains unchanged, however, the derivative $-ip_\mu$ picks up a minus sign.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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