I am aware of the Sokhotski–Plemelj theorem (I have also heard people referring to it as the "Dirac identity") which states that in the limit $\eta\rightarrow 0^+$

$$\frac{1}{x\pm i\eta}=\mathcal P\frac{1}{x}\mp i\pi\delta(x) \, .$$

Now, I am reading the book "Solid State Physics" by G. Grosso and G. Pastori Parravicini which states on page 430 that using the above formula it can "easily be proved" that

$$\frac{\hbar\omega}{E_j-E_i-\hbar\omega-i\eta}= \frac{E_j-E_i}{E_j-E_i-\hbar\omega-i\eta}-1\, .$$

However, I fail to see how the latter formula follows from the former. Is there a trick that I am missing here?


Noting the simple relation $$ \frac{\Delta E_{i,j}}{\Delta E_{i,j}-\hbar\omega-i\eta}-\frac{\Delta E_{i,j}-\hbar\omega-i\eta}{\Delta E_{i,j}-\hbar\omega-i\eta}=\frac{\hbar\omega+i\eta}{\Delta E_{i,j}-\hbar\omega-i\eta} $$ and by Sokhotski-Plemelj theorem $$ \lim_{\eta\rightarrow 0^+} \frac{i\eta}{\Delta E_{i,j}-\hbar\omega-i\eta}=0 $$ because, if the limit $$ \lim_{\eta\rightarrow0^+}\int_{a}^{b}\frac{f(x)}{x\pm i\eta}dx $$ exist, that's what the theorem says, then $$ \lim_{\eta\rightarrow0^+}\int_{a}^{b}\frac{i\eta f(x)}{x\pm i\eta}dx= \left(\lim_{\eta\rightarrow0^+} i\eta\right)\left( \lim_{\eta\rightarrow0^+}\int_{a}^{b}\frac{f(x)}{x\pm i\eta}dx\right)=0 $$

for any target function $f(x)$.

  • $\begingroup$ Why does the last part follow from the Sokhotski-Plemelj theorem? Also, there's a sign mistake in the second equation. Also, I don't think you need the first equation at all, as the second one is trivially true. $\endgroup$ – DanielSank Feb 2 '16 at 3:34
  • $\begingroup$ @DanielSank I updated my answer correcting the sign mistake and explaining in detail the application of Sokhotski-Plemelj theorem. $\endgroup$ – Nogueira Feb 2 '16 at 4:34

I don't think we need Sokhotski-Plemelj for this. Think of $E_j - E_i$ as a fixed value $E$. Then the formula is re-written as

$$\frac{\hbar \omega}{E - \hbar \omega - i \eta}\, .$$

Now let $x \equiv \hbar \omega$ and you get

$$\frac{x}{E -x - i \eta} \, .$$

This integral is dominated by the part where $x \approx E$ so let's try shifting the variables $y \equiv E - x$,

$$\frac{E - y}{y - i \eta}$$

and then expand the numerator and put the original variables back in:

$$\frac{E_j - E_i}{E_j - E_i - \hbar \omega - i \eta} - \frac{E_j - E_i - \hbar \omega}{E_j - E_i - \hbar \omega - i \eta} \, .$$

The first term already matches the first term in the target expression, so we only need to worry about the second term. The second term is 1 because, well, in the limit $\eta \rightarrow 0$ it's identically 1.


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.