As I was studying the Cauchy's integral formula, I tried to do the integral:

\begin{equation} I = \int\limits_{-\infty}^{\infty} \frac{1}{x - a} e^{(i A x^2 + i B x)} dx \end{equation} with $A>0, B>0$ and $a > 0$.

Consider an integral on a complex plan: \begin{equation} J = \int\limits_{C + C_R} \frac{1}{z - a} e^{(i A z^2 + i B z)} dz \end{equation} where $C$ is along the real axis $-\infty \rightarrow +\infty$ and $C_R$ is the upper half circle $z = Re^{i\theta}$ with $R \rightarrow \infty$ and $\theta \in [0, \pi]$.

Naively, I would expect $C_R$ part of the integral gives zero and $C$ part of the integral gives $I$, then the $I$ can be derived by Cauchy's integral formula.

However, as I tried to check the $C_R$ part of the integral, I found that ($z = Re^{i\theta}$): $$ \begin{split} I_R &= \int\limits_0^{\pi} d\theta \frac{iRe^{i\theta}}{Re^{i\theta} - a} \exp\big(iAR^2e^{2i\theta}+iBRe^{i\theta}\big) \\ |I_R| &\leq \int\limits_0^{\pi} d\theta\left |\frac{iRe^{i\theta}}{Re^{i\theta} - a}\right| \Big|\exp\big(iAR^2e^{2i\theta}+iBRe^{i\theta}\big)\Big| \end{split} $$ where the first term

\begin{equation} \left|\frac{iRe^{i\theta}}{Re^{i\theta} - a}\right| \leq \frac{R}{R-a} \rightarrow 1 \ as\ R \rightarrow \infty \end{equation}

and the second term \begin{equation} \left|\exp(iAR^2e^{2i\theta}+iBRe^{i\theta})\right| \leq e^{-AR^2\sin(2\theta) - BR\sin(\theta)} \end{equation} will not approach to zero because of $e^{-AR^2\sin(2\theta)}$.

Is there anything wrong in my approach? And is there any other way I can perform this integral $I$?

Thanks a million for advises!

  • $\begingroup$ Have you tried a rectangle instead of circle? $\endgroup$ – DuFong Jun 1 at 6:32
  • $\begingroup$ @DuFong. Thanks for your suggestion. Yes, I tried a contour $C_1:-R \rightarrow R$, $C_2: R \rightarrow R+i\epsilon$, $C_3: R+i\epsilon \rightarrow -R+i\epsilon$ and $C_4: -R+i\epsilon \rightarrow -R$. The integrals of $C_2$ and $C_4$ approach to zero as $R\rightarrow \infty$. But the one of $C_3$ faces the same problem : $\int_{-\infty}^{\infty} dx e^{-2A\ x\ \epsilon}$ will not. Any other suggestions, please? $\endgroup$ – physics_rocks Jun 1 at 9:43
  • $\begingroup$ Before anything else, you should define what you mean by integrating over the point $x=a$. Is this meant to be a principal value? $\endgroup$ – Michael Engelhardt Jun 1 at 13:03
  • $\begingroup$ This integral is apparently related to the time-dependent linear Schrodinger equation with step function initial data, have you check the StackExchange: this. And to study this, inverse scattering is possible the most commonly used method in mathematical physics. By the way, in order to make sense, $\frac{1}{x-a}$ should be regarded as a distribution. As it is not easy to find the exact solution, one may consider finding asymptotics as $A\rightarrow \infty$ $\endgroup$ – DuFong Jun 1 at 13:57
  • $\begingroup$ @MichaelEngelhardt. Yes, it's a principal value. $\endgroup$ – physics_rocks Jun 2 at 7:27

Let me first remove the $Bx$ term by completing the square, $$I=\int\limits_{-\infty}^{\infty} \frac{e^{i A x^2+iBx}}{x - a}\,dx=e^{-iB^2/4A}\int\limits_{-\infty}^{\infty} \frac{e^{i A x^2}}{x - a-B/2A}\,dx.$$ Mathematica evaluates the Cauchy principal value of the integral in terms of Meijer G-functions, $$I=-\tfrac{1}{8} \pi ^{-5/2} e^{-iB^2/4A}\biggl\{G_{3,5}^{5,3}\left(\alpha\,\biggl| \begin{array}{c} 0,\frac{1}{4},\frac{3}{4} \\ 0,0,\frac{1}{4},\frac{1}{2},\frac{3}{4} \\ \end{array} \right)+8 \pi ^4 G_{7,9}^{5,3}\left(\alpha\,\biggl| \begin{array}{c} 0,\frac{1}{4},\frac{3}{4},-\frac{1}{8},\frac{1}{8},\frac{3}{8},\frac{5}{8} \\ 0,0,\frac{1}{4},\frac{1}{2},\frac{3}{4},-\frac{1}{8},\frac{1}{8},\frac{3}{8},\frac{5}{8} \\ \end{array} \right)+i G_{3,5}^{5,3}\left(\alpha\,\biggl| \begin{array}{c} \frac{1}{4},\frac{1}{2},\frac{3}{4} \\ 0,\frac{1}{4},\frac{1}{2},\frac{1}{2},\frac{3}{4} \\ \end{array} \right)+8 \pi ^4 i G_{7,9}^{5,3}\left(\alpha\,\biggl| \begin{array}{c} \frac{1}{4},\frac{1}{2},\frac{3}{4},-\frac{1}{8},\frac{1}{8},\frac{3}{8},\frac{5}{8} \\ 0,\frac{1}{4},\frac{1}{2},\frac{1}{2},\frac{3}{4},-\frac{1}{8},\frac{1}{8},\frac{3}{8},\frac{5}{8} \\ \end{array} \right)\biggr\},$$ with $$\alpha=\left(a+\frac{B}{2A}\right)^4\frac{A^2}{4}.$$

| 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.