3

Potentially more math than physics here, but I'm looking for the scattering amplitude of the potential:

$$U(r) = \frac{U_o}{(r^2+d^2)^2}.$$

Where $d$ is a constant.

Using the first Born approximation for a spherically symmetric potential leads to:

$$f(q) = \frac{U_0}{q}\int_d^{\infty}\frac{\sin(qr)}{(r^2+d^2)^2}rdr$$

Where $q=|\mathbf{k_i}-\mathbf{k_f}|$ is the momentum transfer. I have yet to find a way to solve this integral analytically (as is suggested is possible) and integral tables have proven only useful in transforming the integral into others that are equally complicated. Any advice or insight is greatly appreciated!

$$\textbf{SOLUTION}$$

In my set up of the Born approximation integral I incorrectly assumed that the polarization potential goes to zero for $r<d$. Given that this is not the case, the integral from above becomes

$$f(q) = \frac{U_0}{q}\int_0^{\infty}\frac{sin(qr)}{(r^2+d^2)^2}rdr$$

This integral can be looked up in an integral table by Dwight, but that's not what I'm here for. To go about solving this, we first set $d^2=z$ notice that

$$\frac{\partial}{\partial z}\Big(\frac{1}{r^2+z}\Big) = -\frac{1}{(r^2+z)^2}$$

so that

$$f(q) = -\frac{U_0}{2iq}\frac{\partial}{\partial z}\Bigg[\int_0^{\infty}\frac{e^{iqr}rdr}{r^2+z} - \int_0^{\infty}\frac{e^{-iqr}rdr}{r^2+z}\Bigg]$$

where I have also expanded the $sin(qr)$ term into it's exponential representation. Letting $r \rightarrow -r$ in the second integral yields

$$f(q) = -\frac{U_0}{2iq}\frac{\partial}{\partial z}\int_{\infty}^{\infty}\frac{e^{iqr}rdr}{r^2+z} = -\frac{U_0}{2iq}\frac{\partial}{\partial z}\int_{\infty}^{\infty}\frac{e^{iqr}rdr}{(r+i\sqrt{z})(r-i\sqrt{z})}$$

revealing two simple poles at $\pm i\sqrt{z}$. Closing the integral in the upper-half plane and using Cauchy's integral formula allows us to evaluate the integral

$$\frac{U_0}{2iq}\int_{\infty}^{\infty}\frac{e^{iqr}rdr}{(r+i\sqrt{z})(r-i\sqrt{z})} = \frac{U_0 \pi}{2q}e^{-q\sqrt{z}}$$

Carrying out the partial derivative and back substituting $z=d^2$ gives us the final result:

$$f(q) = \frac{U_0 \pi}{4d}e^{-qd}$$.

NOTE: This solution was typed up quickly. There may be a factor or a negative, etc. that is missing but the overall technique should hold. Let me know of any errors or improvements you might run across. Thanks!

  • Any specific reason for the restriction of the integral to $r>d$? The integral over $(0,\infty)$ is perfectly doable. – E.P. Nov 14 '16 at 22:11
  • 1
    In units of $d=1$, your integral is $U_0/q$ times $\frac{1}{4} e^{-q} \left(\pi q+e^q (i q (\text{Ci}((-1+i) q)-\text{Ci}((1+i) q)) \cosh (q)+q (\text{Si}((1+i) q)-\text{Si}((-1+i) q)) \sinh (q)+\sin (q))\right)$ where $\mathrm{Ci}$ and $\mathrm{Si}$ are the cos/sin integrals. As Emilio points out, the lower limit should be $0$, in which case the integral reads $\frac{1}{4} q\pi e^{-q}\ $ times $U_0/q$. – AccidentalFourierTransform Nov 14 '16 at 22:29
  • @E.P. was correct. The lower bound should be zero. I incorrectly assumed a zero potential inside $r<d$. Given the entire positive real line the integral can be carried out with a trick with a partial derivative to transform the integrand such that it has two simple poles (not necessary but convenient) and extends to the entire real line. Using Cauchy's integral theorem we can carry out the integral over the closed loop. Thanks for the responses! – FORTRAN90 Nov 15 '16 at 23:15
  • 1
    @Jack For the sake of future visitors who might find this via google search, the ideal course of action now is for you to write a short summary of how you resolved this - what integral you ended up doing and why, how you integrated it, and what the result is. – E.P. Nov 15 '16 at 23:24
  • @E.P I edited my original post. Thanks for the suggestion. New to SE. – FORTRAN90 Nov 16 '16 at 00:18
  • @Jack We were all new at some point ;-). Here the best thing is to post the solution as an answer and mark it as accepted; that makes it clear that the question is resolved. – E.P. Nov 16 '16 at 00:57
  • Also, if it's headers you want, try #'s at the start of the line, or an underline of hyphens or of equal signs. – E.P. Nov 16 '16 at 00:58

0 Answers0