How does one evaluate the integral $\int\limits_{-\infty}^\infty {\exp(iax)\over1+ix}dx$? I tried Wolfram Alpha, but it just says "computation timed out"... I tried the indefinite integral and got an answer involving some weird function $E_1$. Is it possible to bypass the weird function? I presume the limits of my integral would eliminate that, but how?
-
Nothing special about $a$? – J. M. ain't a mathematician Nov 15 '11 at 00:06
-
On the other hand, residues look to be the best route for this... – J. M. ain't a mathematician Nov 15 '11 at 00:08
-
1J.M. is correct. Residue theory is the way to go. However, if you type "exp(iax)/(1+ix)" into Alpha it will give you an alternate form in terms of sines and cosines. If you integrate each term separately (clicking on the link for "give alpha more time" when it times out), it'll give you answers. Then piece them back together. – Bill Cook Nov 15 '11 at 00:16
-
Fourier maybe ? – Dhaivat Pandya Nov 15 '11 at 00:20
-
@J.M.: Thanks! :) Unfortunately I don't quite know what "residues" are... And yes, $a$ is just some constant. – owen Nov 15 '11 at 00:46
-
Okay... how good is your complex analysis? – J. M. ain't a mathematician Nov 15 '11 at 00:47
-
@J.M.: Erm, I'm a beginner...?:S but I'm happy to learn – owen Nov 15 '11 at 00:49
-
@BillCook: Ah, I've never noticed that "button"! – owen Nov 15 '11 at 00:50
-
I know $a$ is a constant; I was wondering if you were assuming, for instance, that $a$ is real. (If $a$ is purely imaginary, then we've no finite value to speak of.) – J. M. ain't a mathematician Nov 15 '11 at 00:52
-
@J.M.: Sorry about that, didn't mean to insult you! :S Anyway, yes, $a\in \mathbb R$ – owen Nov 15 '11 at 00:55
-
No worries, I never read it as an insult. :) I was actually trying to help you make your question more precise, complex analysis being tricky and all. What works nicely on the real line can get weird in the complex plane, you see... – J. M. ain't a mathematician Nov 15 '11 at 00:58
-
In any event, any good book on complex analysis should have a chapter/section on residues. – J. M. ain't a mathematician Nov 15 '11 at 00:59
-
@J.M.: Thanks! :) I'll try to get hold of one... – owen Nov 15 '11 at 01:00
-
Can you do the case $a=0$? – GEdgar Nov 15 '11 at 01:29
-
for $a = 1$ we get $\dfrac {2\pi} {e}$ – Comic Book Guy Apr 26 '12 at 10:45
-
This looks like the Fourier Transform. – night owl Jul 05 '12 at 04:15
-
Yes, this can be done using Fourier Transforms. $\frac{1}{1+ix}$ is the inverse Fourier Transform of a simple function of $\omega$ – Stefan Smith Oct 15 '12 at 01:02
2 Answers
Set $z=ix$ then $dz=idx$. The integral reads as $$\mathcal{I}=\frac{1}{i}\int^{i\infty}_{-i\infty}\frac{e^{az}}{1+z}\,dz=\frac{2\pi}{2\pi i}\int^{i\infty}_{-i\infty}\frac{e^{az}}{1+z}\,dz$$ The integral $$\frac{1}{2\pi i}\int^{i\infty}_{-i\infty}\frac{e^{az}}{1+z}\,dz=\mathcal{L}^{-1}(\frac{1}{1+z})(a)$$ is the Bromwich integral and the right side is the inverse Laplace transform. To evaluate this inverse Laplace transform consider the following integral $$\mathcal{J}=\oint_{\gamma}\frac{e^{az}}{1+z}\,dz$$ where $\gamma$ is a contour consisting of a vertical line on the imaginary axis and a semicircle on the left-half plane. We could partition the contour integral as follows: $$\oint_{\gamma}\frac{e^{az}}{1+z}\,dz=\int^{iT}_{-iT}\frac{e^{az}}{1+z}\,dz+\oint_{\Gamma_R}\frac{e^{az}}{1+z}\,dz$$ The last integral can be estimated as $$\Big|\oint_{\Gamma_R}\frac{e^{az}}{1+z}\,dz\Big|\leq\oint_{\Gamma_R}\Big|\frac{e^{az}}{1+z}\,dz\Big|\leq \frac{e^{-Ra}}{R-1}\to0$$ as $R\to\infty$. Therefore in the limit the only contribution comes from the first integral namely $$\lim_{T\to\infty}\int^{iT}_{-iT}\frac{e^{az}}{1+z}\,dz=\int^{i\infty}_{-i\infty}\frac{e^{az}}{1+z}\,dz$$ Within this contour there is only one simple pole of the integrand at $z=-1$ with residue $e^{-a}$. Appealing to Cauchy Theorem on residues then $$\oint_{\gamma}\frac{e^{az}}{1+z}\,dz=2\pi i\cdot e^{-a}\Rightarrow \int^{i\infty}_{-i\infty}\frac{e^{az}}{1+z}\,dz=2\pi i \cdot e^{-a} $$ But $$\mathcal{I}=\frac{1}{i}\int^{i\infty}_{-i\infty}\frac{e^{az}}{1+z}\,dz\Rightarrow \mathcal{I}=2\pi\cdot e^{-a}$$
- 6,277
-
You suffer from the same problem here as in your answer here http://math.stackexchange.com/questions/1046207/evaluating-an-integral-by-residue-theorem/1046390#1046390
I think you need to understand how inverse Laplace transforms work before publishing "solutions" like these. – Ron Gordon Dec 01 '14 at 12:55
$$\int_{-\infty}^{+\infty}\frac{e^{iax}}{1+ix}dx=\frac{2\pi}{e^a}\quad,\quad a\in\mathbb{R}_+^*$$
- 48,334
- 2
- 83
- 154
-
1This is not quite correct. Think about the integral for the inverse transform. – Ron Gordon Jun 12 '14 at 12:15
-
I think you should add the work you went through in your next answers. – lsoranco Feb 14 '19 at 16:56
-