If we consider $s$-wave scattering for two scalar fields $\phi$ and $\chi$ with an interaction $\frac{g}{2}\phi^2\chi$, then the Lorentz-invariant scattering amplitude to second order is:

$\mathcal{M}_{fi} = \frac{-ig^2}{s - m_\chi^2}$, where $s$ is the Mandelstam-s ($E_{cm}^2$).

The cross section, which goes as $\left| \mathcal{M}_{fi}^2 \right |$, is clearly singular at the $\chi$ resonance. But I don't know how to reconcile that with the scattering formulat given in Sakurai 7.6.17:

The cross section by $\left| f(\theta)\right|^2$, where $f(\theta)$ is given by the partial wave decomposition:

$f(\theta) = \frac{1}{k}\sum_l (2l+1)e^{i\delta_l}\sin\delta_l P_l(\cos\theta)$

For $s$-channel scattering, only $l=0$ contributes, and I do not see how we can have a singularity in $f(\theta)$ anywhere, as I would expect for resonance scattering.

  • $\begingroup$ Anytime a physics calculation yields infinity, there is something wrong $\endgroup$ Feb 15, 2018 at 16:24
  • $\begingroup$ Usually, the fault lies in an approximation that may be accurate within some region but fails on other regions. Trust the finite resuly $\endgroup$ Feb 15, 2018 at 16:26

1 Answer 1


The issue is that the $\mathcal{M}_{fi}$ in the question is only calculated to second order. When you include loop corrections and "dress" the propagator, it takes on a different structure and the width of the peak becomes finite.


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.