I am looking to derive the relation $$\frac{1}{X + i\delta} = \text{P.V} \frac{1}{X} - i \pi \delta(X)$$ In particular, I don't see where the factor of $\pi$ comes from in the derivation. I proceed by computing the imaginary part of the l.h.s as a discontinuity: Take $X = 1-zt$, then, in complex $z$ plane, $$\text{Disc}_z \frac{1}{(1-zt)} = \text{lim}_{\epsilon \rightarrow 0 } \left( \frac{1}{(1-(z+i\epsilon)t)} - \frac{1}{(1-(z-i\epsilon)t)}\right)$$ $$= \text{lim}_{\epsilon \rightarrow 0 } \left( \frac{1}{(1-zt -i\epsilon)} - \frac{1}{(1-zt+i\epsilon)}\right) = \text{lim}_{\epsilon \rightarrow 0 } \frac{2i \epsilon}{(1-zt)^2 + \epsilon^2} = 2i \delta(1-zt) $$ The imaginary part is therefore $i \delta(1-zt)$ so I am off from the actual result by a minus and a factor of $\pi$. Can anyone see where I went wrong?


  • $\begingroup$ ur last equality is not true, there should be a pi due to normailzation constraints $\endgroup$ – tired Mar 14 '16 at 9:20
  • $\begingroup$ @tired: Many thanks! Could you elaborate on what these normalisation constraints are? $\endgroup$ – CAF Mar 14 '16 at 9:22

The best strategy is to rationalize $$ \frac{1}{X + i\epsilon} = \frac{X-i \epsilon}{(X+i\epsilon)(X-i\epsilon)}=\frac{X}{X^2+\epsilon^2}+i\frac{-\epsilon}{X^2+\epsilon^2}\ . $$ The imaginary part is a nascent delta function, as the limit of a Lorentzian with an 'infinitely sharp' peak (see http://mathworld.wolfram.com/DeltaFunction.html) $$ \lim_{\epsilon\to 0^+}\frac{\epsilon}{\pi (X^2+\epsilon^2)}=\delta(X)\ . $$ The factor of $\pi$ is due to normalization: $$ \int_{-\infty}^\infty \delta(x-X)dX=1 $$ by definition for all $x$, and the integral of the Lorentzian is normalized to $1$ only with that factor of $\pi$. See also https://en.wikipedia.org/wiki/Sokhotski%E2%80%93Plemelj_theorem


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.