0
$\begingroup$

I'm considering here the fact that $$\lim\limits_{R\to\infty} \int_{\Gamma_R} \frac {e^{iz}}{z^2+1} dz=0$$ , where $\Gamma$ is a contour defined as a semicircle centred about the origin, of radius $R>1$, in the upper half-plane of $\mathbb{C}$, positively oriented.

Now, I think, one could express $\int\limits_{-\infty}^{\infty} \frac {sin(t)}{t^2+1} dt$ as $$\lim\limits_{R\to\infty} \int_{\Gamma_R} \frac {sin(t)}{t^2+1} dt$$ But here's where I don't know what to do next. Would appreciate a hint.

On a side note, $$\int\limits_{-\infty}^\infty \frac {e^{iz}}{z^2+1} dz\ne 0$$ , so I'm not sure what I'm not seeing here because $$\lim\limits_{R\to\infty} \int_{\Gamma_R} \frac {e^{iz}}{z^2+1} dz=0$$, so where's the analogy?

$\endgroup$
  • 1
    $\begingroup$ Are you aware of the residue theorem? $\endgroup$ – joriki Jun 16 '16 at 23:35
  • $\begingroup$ Haven't studied it yet. $\endgroup$ – sequence Jun 17 '16 at 0:03
4
$\begingroup$

The integrand is and odd function of $t$ and the integral is absolutely convergent, so $$\int_{-\infty}^{\infty}\frac{\sin t}{t^2+1}dt=0$$ EDIT: Writing out the theorem longhand, $$\begin{align}\int_{-\infty}^{\infty}\frac{\sin t}{t^2+1}dt&=\lim_{M\rightarrow-\infty}\int_M^0\frac{\sin t}{t^2+1}dt+\lim_{N\rightarrow\infty}\int_0^N\frac{\sin t}{t^2+1}dt\\ &=\lim_{M\rightarrow-\infty}\int_{-M}^0\frac{\sin(-u)}{u^2+1}(-du)+\lim_{N\rightarrow\infty}\int_0^N\frac{\sin u}{u^2+1}du\\ &=\lim_{M\rightarrow-\infty}-\int_0^{-M}\frac{\sin u}{u^2+1}du+\lim_{N\rightarrow\infty}\int_0^N\frac{\sin u}{u^2+1}du\\ &=-\int_0^{\infty}\frac{\sin u}{u^2+1}du+\int_0^{\infty}\frac{\sin u}{u^2+1}du=0\end{align}$$ Where we have used the substitution $u=-t$ in the left integral and $u=t$ in the right integral. Notice that all that was required for this proof to work was that the integrand was an odd function, the interval of integration was $(-\infty,\infty)$, although an interval of $(-a,a)$ would have worked and that the function was Riemann integrable over the interval. This theorem can reduce some perfectly horrible-looking integrals to zero with relatively little effort.

$\endgroup$
  • $\begingroup$ Can you please be more specific? $\endgroup$ – sequence Jun 17 '16 at 0:04
  • $\begingroup$ @sequence: if $f\in L^1(\mathbb{R})$ and $f$ is an odd function, $$\int_{-\infty}^{+\infty}f(x)\,dx = \color{red}{0}.$$ $\endgroup$ – Jack D'Aurizio Jun 17 '16 at 0:14
  • $\begingroup$ What is $L^1(\mathbb{R})$? $\endgroup$ – sequence Jun 17 '16 at 0:32
1
$\begingroup$

First Solution:

From Residue theorem: $$\int\limits_{-\infty}^\infty \frac {e^{iz}}{z^2+1} dz = \frac{\pi}{e}$$

By Euler's formula: $$\int\limits_{-\infty}^\infty \frac {e^{iz}}{z^2+1} dz =\int\limits_{-\infty}^\infty \frac{cos z}{z^2+1} dz + i\int\limits_{-\infty}^\infty \frac{sin z}{z^2+1} dz$$

In this post, it shows that: $$\int\limits_{-\infty}^\infty \frac{\cos z}{z^2+1} dz = \frac{\pi}{e}$$

In conclude: $$\int\limits_{-\infty}^\infty \frac{\sin z}{z^2+1} dz = 0$$

Alternative Solution(inspired by the same post):

Because $sin(z)$ behaves not like $e^{iz}$ over the complex plane, the straightforward way to use residue theorem is invalid.

By Euler's formula: $$\int\limits_{-\infty}^\infty \frac {\sin z}{z^2+1} dz = \frac{1}{2} i (\int\limits_{-\infty}^\infty \frac {e^{-ix}}{z^2+1} dz-\int\limits_{-\infty}^\infty \frac {e^{ix}}{z^2+1} dz) $$

From Residue theorem: $$\int\limits_{-\infty}^\infty \frac {e^{ix}}{z^2+1} dz = \frac{\pi}{e} $$ $$\int\limits_{-\infty}^\infty \frac {e^{-ix}}{z^2+1} dz = \frac{\pi}{e} $$

The proof is done.

$\endgroup$
  • $\begingroup$ The Residue Theorem has not been studied yet, so a solution should be possible without it. $\endgroup$ – sequence Jun 17 '16 at 0:32
  • 1
    $\begingroup$ That doesnt matter. The key issue about your analogy is that $\lim\limits_{R\to\infty} \int_{\Gamma_R} \frac {sin(t)}{t^2+1} dt \neq 0$ even though $\lim\limits_{R\to\infty} \int_{\Gamma_R} \frac {e^{iz}}{z^2+1} dz=0$ $\endgroup$ – Zack Ni Jun 17 '16 at 0:36
  • 1
    $\begingroup$ The Residue Theorem is a very simple theorem about the relation between poles and closed contour integration. Hopefully, you can learn it in several secs. $\endgroup$ – Zack Ni Jun 17 '16 at 0:38

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.