Consider a Klein-Gordon field $\phi$, which satisfies $$(\Box+ \omega_0^2)\phi=0$$ on points $x \equiv \{x_0,\vec{x}\},y\equiv \{y_0,\vec{y} \}$ of 4D Minkowski-spacetime. The field commutator is $$ [\phi(x),\phi(y)]=c \int_{\mathbb{R}^3} \frac{1}{(2\pi)^3 2 \sqrt{\lvert \vec{k} \rvert^2 + \omega_0^2}}\left ( e^{-i\sqrt{\lvert \vec{k} \rvert^2 + \omega_0^2} \ (x_0-y_0)} - e^{i\sqrt{\lvert \vec{k} \rvert^2 + \omega_0^2} \ (x_0-y_0)} \right )e^{i \vec{k}\cdot (\vec{x}-\vec{y})} \ dk. $$ I want to verify that this integral is equal to $$ [\phi(x),\phi(y)]=c \ \text{sgn}(x_0-y_0) \left ( i \omega_0 \theta (\tau^2) \frac{J_1(\omega_0 \tau)}{4\pi \tau} - \frac i {2\pi} \delta (\tau^2) \right ) $$ where $\tau\equiv\sqrt{(x_0-y_0)^2 - \lvert \vec{x}-\vec{y} \rvert^2}$, $\theta$ is the Heavyside function, and $J_1$ is the Bessel function. This identity looks intractable to me.

Writing the integral in polar coordinates and doing the integral over the angles simplifies the expression somewhat but I still cannot derive the result using either theorems regarding the Fourier transform or substitution techniques.

  • $\begingroup$ I guess you want to be integrating $k$ over $\mathbb{R}^2$? $\endgroup$ – Fabian Jan 16 '15 at 8:47
  • $\begingroup$ Also there is some further mistake: as it stands the term in the brackets simply vanishes. $\endgroup$ – Fabian Jan 16 '15 at 8:51
  • 1
    $\begingroup$ As you have posted it additionally on physics.stackexchange, I vote to close this question as it better fits to the other forum. $\endgroup$ – Fabian Jan 16 '15 at 9:39
  • 4
    $\begingroup$ Should be migrated but it is already there. $\endgroup$ – Fabian Jan 16 '15 at 9:40
  • 2
    $\begingroup$ This question was ruled off-topic at Physics.SE because it is Homework-like. The only difficulty is in computing the integral. There is no conceptual physics question to be asked here. I have changed some of the notation so it is less physics-centric and hopefully it is understandable and can be reopened. $\endgroup$ – Kevin Driscoll Jan 20 '15 at 22:59

($c\equiv 1$) The commutator can be written as the sum of the positive- and negative-frequency part of the Pauli-Jordan function: $$ i[\phi(x),\phi(y)]=D^+(x-y)+D^-(x-y), $$ where \begin{align} D^\pm(x) &= \pm\frac{1}{i(2\pi)^3}\int\frac{d^3k}{2\sqrt{\mathbf k^2+m^2}} e^{\pm i\sqrt{\mathbf k^2+m^2}\,x^0-i\mathbf k\cdot\mathbf x}. \end{align} Performing the $3$-momentum integration and letting $r\equiv|\mathbf x|$, one has $$ D^+(x) = -\frac{1}{8\pi^3r}\int_{-\infty}^{+\infty}\frac{e^{i\sqrt{\mathbf \rho^2+m^2}x^0+i\rho r}}{\sqrt{\rho^2+m^2}}\rho d\rho=\frac{1}{4\pi r}\frac{\partial}{\partial r}\,f(r), $$ where $$ f(r) = \frac{i}{2\pi}\int_{-\infty}^{+\infty}\frac{e^{i\sqrt{\mathbf \rho^2+m^2}x^0+i\rho r}}{\sqrt{\rho^2+m^2}}d\rho =\frac{i}{2\pi}\int_{-\infty}^{+\infty}e^{im(x^0\cosh\varphi+r\sinh\varphi)}d\varphi, $$ using $\rho = m \sinh\varphi$. Now: let $\lambda\equiv(x^0)^2-r^2$,

1) if $x^0,\lambda>0$, let $x^0=\sqrt{\lambda}\cosh\varphi_0$ and $r=\sqrt{\lambda}\sinh\varphi_0$, so that \begin{align} -f(r)=\frac{1}{i2\pi}\int_{-\infty}^{+\infty}e^{im\sqrt{\lambda}\cosh(\varphi+\varphi_0)}d\varphi=\frac{1}{2}H_0^1(m\sqrt{\lambda})=\frac{1}{2}J_0(m\sqrt{\lambda})+\frac{i}{2}Y_0(m\sqrt{\lambda}) \end{align} (see here for reference on integral representations of Bessel functions);

2) if $x^0>0,\lambda<0$, let $x^0=\sqrt{-\lambda}\sinh\varphi_0$ and $r=\sqrt{-\lambda}\cosh\varphi_0$, so that \begin{align} f(r)=\frac{i}{2\pi}\int_{-\infty}^{+\infty}e^{im\sqrt{-\lambda}\sinh(\varphi+\varphi_0)}d\varphi=\frac{i}{\pi}K_0(m\sqrt{-\lambda}); \end{align}

3) if $x^0<0,\lambda>0$, let $x^0=-\sqrt{\lambda}\cosh\varphi_0$ and $r=\sqrt{\lambda}\sinh\varphi_0$, so that \begin{align} f(r)=\frac{i}{2\pi}\int_{-\infty}^{+\infty}e^{-im\sqrt{\lambda}\cosh(\varphi-\varphi_0)}d\varphi=\frac{1}{2}H_0^2(m\sqrt{\lambda})=\frac{1}{2}J_0(m\sqrt{\lambda})-\frac{i}{2}Y_0(m\sqrt{\lambda}); \end{align}

4) if $x^0,\lambda<0$, let $x^0=-\sqrt{-\lambda}\sinh\varphi_0$ and $r=\sqrt{-\lambda}\cosh\varphi_0$, so that \begin{align} f(r)=\frac{i}{2\pi}\int_{-\infty}^{+\infty}e^{-im\sqrt{\lambda}\sinh(\varphi-\varphi_0)}d\varphi=\frac{i}{\pi}K_0(m\sqrt{-\lambda}). \end{align} From all this: let $\text{sign}(x^0)\equiv \varepsilon(x^0)$ $$ f(r) = \begin{cases} \frac{1}{2i}Y_0(m\sqrt{\lambda})-\frac{\varepsilon(x^0)}{2}J_0(m\sqrt{\lambda})\text{ for }\lambda>0\\ \frac{i}{\pi}K_0(m\sqrt{-\lambda})\text{ for }\lambda<0. \end{cases} $$ Now, to take the derivative with respect to $r$, use $$ \frac{1}{4\pi r}\frac{\partial}{\partial r}=-\frac{1}{2\pi}\frac{\partial}{\partial \lambda} $$ together with the fact that the function $f$ is discontinuous at $\lambda=0$ due to of $J_0(0^+)=1$ showing up for for positive $\lambda$, then: $$ D^+(x)=\frac{1}{4\pi}\varepsilon(x^0)\delta(\lambda)-\frac{im}{8\pi\sqrt{\lambda}}\theta(\lambda)\left[Y_1(m\sqrt{\lambda})-i\varepsilon(x^0) J_1(m\sqrt\lambda) \right]-\frac{im}{4\pi^2\sqrt{-\lambda}}\theta(-\lambda)K_1(m\sqrt{-\lambda}). $$ Working in a similar fashion, $$ D^-(x)=\frac{1}{4\pi}\varepsilon(x^0)\delta(\lambda)-\frac{im}{8\pi\sqrt{\lambda}}\theta(\lambda)\left[-Y_1(m\sqrt{\lambda})-i\varepsilon(x^0) J_1(m\sqrt\lambda) \right]+\frac{im}{4\pi^2\sqrt{-\lambda}}\theta(-\lambda)K_1(m\sqrt{-\lambda}). $$ Finally $$\boxed{ i[\phi(x), \phi(y)]=\frac{1}{2\pi}\varepsilon(x^0)\delta(\lambda)-\frac{m}{8\pi\sqrt{\lambda}}\theta(\lambda)\varepsilon(x^0) J_1(m\sqrt\lambda).} $$

(Ref: This closely follows the derivation in Bogoliubov's book.)


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.