Starting from the free lagrangian

$$\mathscr L = \bar\Psi(i\displaystyle{\not}\partial - m)\Psi$$

I compute the canonical momenta

$$\Pi =\frac{\partial \mathscr L}{\partial\dot{\Psi}}=i\Psi^\dagger \qquad \bar\Pi =\frac{\partial \mathscr L}{\partial\dot{\bar\Psi}}= 0$$

and then I can perform canonincal quantization imposing the canonical anticommutation rules (equal time anticommutator)

$$\left\{\Psi_\alpha(x^0,\vec{x}),\Pi_\beta(x^0,\vec{y})\right\} = i\left\{\Psi_\alpha(x^0,\vec{x}),\Psi_\beta^\dagger(x^0,\vec{y})\right\} \equiv i\delta_{\alpha\beta}\delta^3\left(\vec{x}-\vec{y}\right) \tag1$$ $\alpha,\beta = 1,2,3,4$ are spinor indices.

I know the covariant anticommutator ($x$ and $y$ are 4-vectors)

$$\left\{\Psi_\alpha(x),\bar\Psi_\beta(y)\right\} = iS_{\alpha\beta}(x-y)=i(\displaystyle{\not}\partial+m)_{\alpha\beta}\Delta(x-y) \tag2$$

A possible definition of the $\Delta$ function is

$$\Delta(x) = -\frac{i}{2}\int\frac{d^3\vec{k}}{(2\pi)^3}\frac{e^{-ikx}-e^{ikx}}{\omega_k}= -\int\frac{d^3\vec{k}}{(2\pi)^3}\frac{\sin(kx)}{\omega_k}\qquad \text{where} \quad \omega_k = \sqrt{\left|\vec{k}\right|^2+m^2}$$

Now I can rewrite (2) as

$$\left\{\Psi_\alpha(x),\Psi^\dagger_\sigma(y)(\gamma_0)^\sigma_{\ \ \beta}\right\} = \left\{\Psi_\alpha(x),\Psi^\dagger_\sigma(y)\right\}(\gamma_0)^\sigma_{\ \ \beta} \tag3$$

If I take equal time anticommutator in (2), I get

$$\left\{\Psi_\alpha(x^0,\vec{x}),\bar\Psi_\beta(x^0,\vec{y})\right\} = i(\displaystyle{\not}\partial+m)_{\alpha\beta}\Delta(0,\vec{x}-\vec{y}) = 0$$


$$\Delta(0,\vec{x}) = -\int\frac{d^3\vec{k}}{(2\pi)^3}\frac{\sin(-\vec{k}\cdot\vec{x})}{\omega_k} = 0 \quad \text{integral of an odd function}$$

But looking at (3) one might conclude

$$0 = \left\{\Psi_\alpha(x^0,\vec{x}),\bar\Psi_\beta(x^0,\vec{y})\right\}=\left\{\Psi_\alpha(x^0,\vec{x}),\Psi^\dagger_\sigma(x^0,\vec{y})\right\}(\gamma_0)^\sigma_{\ \ \beta} \tag4$$

Which contradict the canonical anticommutation rules (1). For sure I made a mistake, but I can find it!

  • $\begingroup$ Note that it's not: $\mathscr L = i\bar\Psi(\displaystyle{\not}\partial - m)\Psi$, but rather $\mathscr L = \bar\Psi(i\displaystyle{\not}\partial - m)\Psi$ $\endgroup$
    – Omry
    Commented Mar 14, 2017 at 16:25
  • $\begingroup$ yes you're right, stupid mistake I'm sorry! I will immidiately edit it. $\endgroup$
    – M. M. R.
    Commented Mar 14, 2017 at 16:33

1 Answer 1


Even though $\Delta(0,\boldsymbol x)$ is given by an integral of an odd function, the integral diverges for $\boldsymbol x=\boldsymbol 0$. The correct statement is $$ \Delta(0,\boldsymbol x)=0 $$ and $$ \dot\Delta(0,\boldsymbol x)=\delta(\boldsymbol x) $$ (up to a phase, depending of conventions)

This is consistent with the canonical algebra.


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.