I've been studying "Lifetime of a quasiparticle in an electron liquid", by Qian and Vignale. Much of it makes sense, but there is a detail in the calculation of the exchange term that doesn't make sense to me. Eqn. 23 gives

$$\frac{2\pi me^2 }{pqk_s\sqrt{k_s^2+4k_F^2-q^2}} $$

This follows from Eqn. 69 in the appendix of the article (omitting the Heaviside theta functions):

$$ \frac{2\pi me^2}{pq\sqrt{[p^2+k^2+k_s^2-{\bf k}\cdot {\bf q}]^2-[k^2-({\bf k}\cdot {\bf \hat{q}})^2][4p^2-q^2] }} $$

Equating the two, this tells me that

$$k_s\sqrt{k_s^2+4k_F^2-q^2}=\sqrt{[p^2+k^2+k_s^2-{\bf k}\cdot {\bf q}]^2-[k^2-({\bf k}\cdot {\bf \hat{q}})^2][4p^2-q^2] } $$

The authors obtain Eqn. 23 from Eqn. 69 by assuming that $p\sim k\sim k_F$ and ${\bf k}\cdot {\bf q}\sim -\frac{q^2}{2}$. The r.h.s. of the above then becomes

\begin{align} &\sqrt{[p^2+k^2+k_s^2-{\bf k}\cdot {\bf q}]^2-[k^2-({\bf k}\cdot {\bf \hat{q}})^2][4p^2-q^2] }\notag\\ \approx& \sqrt{[2k_F^2+k_s^2+q^2/2]^2-1/4[4k_F^2-q^2]^2 } \notag\\ &=\sqrt{4k_F^4+4k_F^2k_s^2+2k_F^2q^2+k_s^4+k_s^2q^2+q^4/4-4k_F^2+2k_F^2q^2-q^2/4 }\notag\\ &=\sqrt{4k_F^2k_s^2+k_s^4+k_s^2q^2+4k_F^2q^2 }\notag\\ &=k_s\sqrt{4k_F^2+k_s^2+q^2+4k_F^2q^2/k_s^2} \end{align}

which is clearly different from the authors' Eqn. 23. Is there an approximation they invoke that they do not mention? I checked my result numerous times and it appears to be mathematically sound. Specifically, Qian & Vignale's denomainator differs from mine in the sign of $q^2$.

EDIT: Fixed a minor typo pointed out by @vin92. The solution, however, still doesn't match that of Qian and Vignale.

  • $\begingroup$ Can you explain the difference between $\hat{\boldsymbol{q}}$ and $\boldsymbol{q}$? I just took a quick look but can it be that you messed up in the second line of your developement? Since you say that $k^2-(\boldsymbol{k}\cdot \hat{\boldsymbol{q}})^2\approx \dfrac{1}{4}[4k_F^2-q^2]$ but it should be $\dfrac{1}{4}[4k_F^2-q^4]$. $\endgroup$ – vin92 May 22 at 9:25
  • $\begingroup$ @vin92 $\hat{\bf q}$ is the unit vector. When you derive Eqn. 69 in the paper, this value comes up (I believe that's correct because I followed that portion of the calculation assuming this, and I got the correct answer). Using the information supplied, we then find the term in the second line of my derivation. Also, assuming that you are correct and it is a $q^4$ term, I do not believe this would solve the discrepancy. $\endgroup$ – Joshuah Heath May 22 at 12:55
  • $\begingroup$ Ok thanks for the notation. Are you sure you expanded $-1/4[4k_F^2-q^2]^2$ correctly? I get $-1/4[4k_F^2-q^2]^2=-4k_F^2+2k_F^2q^2-q^4/4$. $\endgroup$ – vin92 May 22 at 13:25
  • $\begingroup$ @vin92 Thank you for pointing that out. I have corrected that issue in the main part of the text; it seems as if it modifies it slightly, but the main issue is still there; namely, the sign of the -q^2 term (assuming the last term in the square root of my expression can be approximated to zero). $\endgroup$ – Joshuah Heath May 22 at 15:03
  • $\begingroup$ Yes I thought the same, I'm sorry I don't have the time to look deep in the article but could it be that there is a relation between $k_s,k_F$ and $q$, then it may be posssible to approximate the term $q^2[1+2k_F^2/k_s^2]=-q^2$. $\endgroup$ – vin92 May 22 at 15:52

I have contacted one of the authors of this work, and I found out that there is a small typo in the appendix. Eqn. 68 in the Appendix of the arXiv version (Eqn. A2 in the PRB version), which reads

$$ A_3=2\pi e^2 \int_{-1}^1 dx \delta\left( \omega +\frac{pqx}{m}+\frac{q^2}{2m} \right)\frac{1}{\sqrt{ (p^2+k^2+k_s^2+2pk\cos\theta x)^2-4(pk\sin\theta)^2(1-x^2) }} $$

should be $$ A_3=2\pi e^2 \int_{-1}^1 dx \delta\left( \omega +\frac{pqx}{m}+\frac{q^2}{2m} \right)\frac{1}{\sqrt{ (p^2+k^2+k_s^2-2pk\cos\theta x)^2-4(pk\sin\theta)^2(1-x^2) }} $$

Redoing my previous calculation, we find that

\begin{align} &\sqrt{[p^2+k^2+k_s^2+{\bf k}\cdot {\bf q}]^2-[k^2-({\bf k}\cdot {\bf \hat{q}})^2][4p^2-q^2] }\notag\\ \approx& \sqrt{[2k_F^2+k_s^2-q^2/2]^2-1/4[4k_F^2-q^2]^2 } \notag\\ &=\sqrt{4k_F^4+4k_F^2k_s^2-2k_F^2q^2+k_s^4-k_s^2q^2+q^4/4-4k_F^2+2k_F^2q^2-q^4/4 }\notag\\ &=\sqrt{4k_F^2k_s^2+k_s^4-k_s^2q^2}\notag\\ &=k_s\sqrt{4k_F^2+k_s^2-q^2} \end{align}

This leads to Eqn. 24 in the text, a major result in this paper which, from the above calculation, is shown to be correct.

| cite | improve this answer | |

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.