That looks ugly and beautiful simultaneously

Calculus Level 4

cos x x 2 + 1 d x \large \displaystyle \int_{-\infty}^{\infty} \dfrac {\cos x}{x^2 + 1} \, dx

Find the value of the closed form of the above integral.

Give your answer to 2 decimal places.


The answer is 1.16.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

3 solutions

Sharky Kesa
Jan 15, 2017

We will solve via a contour integration, using a semi-circular anti-clockwise contour called C C , with the straight part on the real axis, and the curved part called γ R \gamma_R . We will have radius R > 1 R>1 as there will be poles at ± i \pm i .

Let f ( z ) = e i z z 2 + 1 f(z) = \dfrac {e^{iz}}{z^2+1} . We have

C f ( z ) d z = R R f ( x ) d x + γ R f ( z ) d z \displaystyle \oint_{C} f(z) dz = \int_{-R}^R f(x) dx + \int_{\gamma_R} f(z) dz

We will now try to determine the poles of f ( z ) f(z) . Since e i z e^{iz} is always differentiable in the complex plane, it won't contain any poles. However, when z 2 + 1 = 0 z^2+1=0 , z = ± i z=\pm i , so f f has poles at ± i \pm i . Notice that only one of these poles lies inside C C , and the other one doesn't, so we only need to worry about one of them (WLOG i i is inside C C ).

We will now try to solve the LHS by using the Residue Theorem. First, we will find the residue of f ( z ) f(z) at z = i z=i .

lim z i ( z i ) e i z z 2 + 1 = lim z i e i z z + i = 1 2 e i \begin{aligned} \displaystyle \lim_{z\to i} (z-i) \dfrac {e^{iz}}{z^2+1} &= \lim_{z\to i} \dfrac {e^{iz}}{z+i}\\ &=\dfrac{1}{2ei}\\ \end{aligned}

Now, we can apply the Residue Theorem to get

C f ( z ) d z = 2 π i ( 1 2 e i ) = π e \begin{aligned} \displaystyle \oint_C f(z) dz &= 2\pi i \left ( \dfrac{1}{2ei} \right )\\ &= \dfrac {\pi}{e}\\ \end{aligned}

We will now evaluate the integral furthest to the right using the Estimation Lemma. By the Estimation Lemma, we have

γ R f ( z ) d z length ( γ R ) max z γ R f ( z ) \left | \displaystyle \int_{\gamma_R} f(z) dz \right |\leq \text{length }(\gamma_R) \cdot \max_{z \in \gamma_R} |f(z)|

Now, the length of γ R \gamma_R is just π R \pi R , because it is the perimeter of half a circle (a circle's circumference is 2 π R 2\pi R ). Now we just need to determine max z γ R f ( z ) \displaystyle \max_{z \in \gamma_R} |f(z)| .

Firstly, by the reverse Triangle Inequality z 2 + 1 z 2 1 |z^2 + 1| \geq |z|^2 - 1 . Now, z = R |z|=R whenever z z is on γ R \gamma_R because it is on the curved side of a semicircle, so z 2 + 1 R 2 1 |z^2 + 1| \geq R^2 - 1 . Thus, 1 z 2 + 1 1 R 2 1 \dfrac {1}{|z^2 + 1|} \leq \dfrac {1}{R^2 - 1} .

Now, we will try to bound the numerator e i z e^{iz}

e i z = e i ( ( z ) + ( z ) i ) = e ( z ) i e ( z ) = e ( z ) = e ( z ) 1 \begin{aligned} |e^{iz}| &= |e^{i(\Re(z) + \Im(z)i)}|\\ & = |e^{\Re(z)i}| |e^{- \Im(z)}| \\ &= |e^{- \Im(z)}|\\ &= e^{-\Im(z)}\\ &\leq 1\\ \end{aligned}

The third line comes from the fact that since ( z ) \Re(z) is a real number, it follows that e i ( z ) = cos ( ( z ) ) + i sin ( ( z ) ) e^{i\Re(z)} = \cos(\Re(z)) + i \sin(\Re(z)) , so e i ( z ) = 1 |e^{i\Re(z)}| = 1 . The fourth line is because the exponential term is always non-negative, so we don't need an absolute value sign. In the fifth line, we used the fact that γ R \gamma_{R} is defined to have ( z ) 0 \Im(z) \geq 0 .

Thus,

e i z z 2 + 1 1 R 2 1 \dfrac {|e^{iz}|}{|z^2+1|} \leq \dfrac {1}{R^2-1}

Therefore, we have

γ R f ( z ) d z π R R 2 1 \left | \displaystyle \int_{\gamma_R} f(z) dz \right | \leq \dfrac {\pi R}{R^2 - 1}

As R R tends to infinity, we find that the RHS tends to 0, so the LHS is just 0. We are now left with

R R f ( x ) d x + γ R f ( z ) d z = C f ( z ) d z f ( x ) d x = π e e i x x 2 + 1 d x = π e cos x + i sin x x 2 + 1 d x = π e cos x x 2 + 1 d x + i sin ( x ) x 2 + 1 d x = π e \begin{aligned} \displaystyle \int_{-R}^R f(x) dx + \int_{\gamma_R} f(z) dz &= \oint_{C} f(z) dz\\ \int_{-\infty}^{\infty} f(x) dx &= \dfrac {\pi}{e}\\ \int_{-\infty}^{\infty} \dfrac {e^{ix}}{x^2+1} dx &= \dfrac {\pi}{e}\\ \int_{-\infty}^{\infty} \dfrac{\cos x + i \sin x}{x^2 + 1} dx &= \dfrac {\pi}{e}\\ \int_{-\infty}^{\infty} \dfrac {\cos x}{x^2 +1} dx + i \int_{-\infty}^{\infty} \dfrac {\sin(x)}{x^2 + 1} dx &= \dfrac {\pi}{e}\\ \end{aligned}

Note that since sin ( x ) \sin(x) is an odd function, the integral of it must be 0. Thus, we are left with

cos x x 2 + 1 d x = π e \int_{-\infty}^{\infty} \dfrac {\cos x}{x^2 + 1} dx = \dfrac {\pi}{e}

You missed the factor of π R \pi R in the estimate of γ R f ( z ) d z \int_{\gamma_R} f(z) dz .

You could've used the Jordan's lemma as well.

Also, in the last part, you could've just used taken real part on both sides (rather than arguing that the imaginary part vanishes as sin is odd) :P

A Former Brilliant Member - 4 years, 5 months ago
Pi Han Goh
Jan 15, 2017

For a Fourier transform approach and a Laplace transform approach, see Mark Henning's solution here and my solution (in the same link), respectively.

Jessica Wang
Jan 19, 2017

There exist at least another two solutions, by Mark Hennings and me (and a generalisation of cos b x a 2 + x 2 d x \, \int_{-\infty }^{\infty }\frac{\cos{bx}}{a^{2}+x^{2}}\: dx is also given), in here .

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...