5

I am trying to evaluate the integral

$$\int_{0}^{+\infty}\frac{x-\sin(x)}{x^3(x^2+1)}\,\mathrm dx.$$

I am thinking of using contour integration, but I do not understand how to do it.

EDIT 1

To begin with, I thought it was necessary to find the poles. $x^3 (x^2 + 1)$.

  1. The first factor $x^3$ gives a pole at $x = 0$. Since the degree of $x^3$ is 3, this is a pole of order 3.

  2. The second factor $x^2 + 1$ equals zero when: $ x^2 + 1 = 0 \implies x = \pm i. $ These are simple poles because $x^2 + 1$ appears in the denominator to the first power.

Thus, the function $f(x)$ has the following poles:

$x = 0$— a pole of order 3, $x = i$— a simple pole, $x = -i$— a simple pole.

Next, we need to move on to the residue theorem, but since I have found 3 poles, I am stuck.

EDIT 2

Next, I used the residue theorem to evaluate the integral. Let (f(z)) represent the complex extension of the given function: $$f(z) = \frac{z - \sin(z)}{z^3(z^2 + 1)}.$$

To compute the integral, I considered a contour integral around a semicircular contour in the upper half-plane. This contour includes the poles (z = 0) and (z = i), while avoiding the pole at (z = -i).

The residue theorem states: $$\int_{\text{contour}} f(z) \, \mathrm dz = 2\pi i \sum \text{Res}(f, z_k),$$ where the sum is over all poles inside the contour.

Let’s compute the residues for the poles inside the contour.

Residue at (z = 0) (a pole of order 3):

To find the residue at (z = 0), we use the formula for residues at poles of order 3: $$\text{Res}(f, 0) = \frac{1}{(3-1)!} \lim_{z \to 0} \frac{d^2}{dz^2} \left( z^3 f(z) \right).$$

$$\text{Res}(f, 0) = \frac{i - 2}{2}.$$

Residue at (z = i) (a simple pole):

For a simple pole at (z = i), the residue is: $$\text{Res}(f, i) = \lim_{z \to i} (z - i) f(z).$$

At (z = i), this simplifies to: $$\text{Res}(f, i) = -\frac{e^{-1}}{i}.$$

Residue at (z = -i) (a simple pole):

Similarly, for the simple pole at (z = -i), the residue is: $$\text{Res}(f, -i) = \lim_{z \to -i} (z + i) f(z).$$

At (z = -i), this simplifies to: $$\text{Res}(f, -i) = -\frac{e}{i}.$$

Total Residue and Integral

Summing all residues: $$\text{Total Residue} = \frac{i - 2}{2} - \frac{e^{-1}}{i} - \frac{e}{i}.$$

Using the residue theorem: $$\int_{-\infty}^{+\infty} f(x) \, \mathrm dx = 2\pi i \cdot \text{Total Residue}.$$

Finally, the original integral (over $[0, \infty)$) corresponds to the imaginary part of the result: $$\int_0^{+\infty}\frac{\sin(x)}{x^3(x^2+1)}\,\mathrm dx = \pi \cdot \text{Im}(\text{Total Residue}).$$

$$Im\left(\frac{i-2}{2}+i(e^{-1}+e)\right) = \left(\frac{1}{2}+e^{-1}+e\right)\pi$$

Timur
  • 59
  • Do you know how to use the theorem for simpler integrals on $(-\infty, \infty)$? – Sean Roberson Dec 16 '24 at 15:43
  • 1
    Hint: the integrand is even – Lorago Dec 16 '24 at 15:44
  • @SeanRoberson Well, if I know, how will it help? – Timur Dec 16 '24 at 15:46
  • Knowing the simpler case will tell you how to 1) extend the integrand to a complex function, and 2) determine how to draw your contour. The non-simple pole at zero will require a bit of work. Edit your original post to include these details (don't bury them in comments, please!). – Sean Roberson Dec 16 '24 at 15:49
  • I was trying to calculate the integral along the contour $\Gamma_{rR}$, made up of circles $C_r, C_R$ the radii $r$ and $R$ , respectively , and the two banks of the section along the radius $[r, R]$ – Timur Dec 16 '24 at 15:49
  • And then, according to the theorem on the Calculation of residuals – Timur Dec 16 '24 at 15:51
  • 2
    @Timur inlude your attempt in the question please – Sine of the Time Dec 16 '24 at 15:53
  • well done for adding your attempt. Note that $x=0$ is a removable singularity – Sine of the Time Dec 16 '24 at 16:10
  • @SineoftheTime I don't understand how to fix it – Timur Dec 16 '24 at 16:25
  • @Timur You don't have to compute the residue since it's not a pole – Sine of the Time Dec 16 '24 at 16:30
  • It turns out that next you need to write a formula with deductions for $i$ and $-i$? – Timur Dec 16 '24 at 16:39
  • 1
    Notice that (because of the $\sin z$ term) the the integrand grows rapidly away from the real axis, so the integral of the integrand along $C_R$ will not vanish as $R \to \infty$. One option is to rewrite the integrand as the real part of $g(z) := \frac{z + i e^{i z}}{z^3 (1 + z^2)}$, but then you'll have to deal with its pole of order $3$ at $0$. You might be able to emulate the approach here: https://math.stackexchange.com/a/594681/155629 . – Travis Willse Dec 16 '24 at 17:18
  • Are you sure about the pole in 0 ? $\sin z - z\over z^3$ seems to be holomorphic. – Thomas Dec 16 '24 at 19:03
  • @Thomas I don't know – Timur Dec 17 '24 at 05:31
  • Indeed $\sin z= \sum (-1)^k{ z^{2k+1}\over (2k+1!)} $ – Thomas Dec 17 '24 at 06:03
  • @TravisWillse I think I understand you. Could you look at the solution? I added to the post – Timur Dec 17 '24 at 14:48
  • @Timur Like I mentioned in my previous comment, the integral of $f$ along the semicircle part of your contour is not $0$, so the contour integral is not equal to $\int_{-\infty}^\infty f ,dx$. As some of the other commenters have mentioned, the singularity of $\frac{\sin z - z}{z^3}$ at $0$ is removable, so its residue is $0$. (The singularity of $g(z) = \frac{z + i e^{i z}}{z^3 (1 + z^2)}$ at the origin is a pole of order $3$, however.) In any case, a c.a.s. gives the value $\frac\pi4 \left(1 - \frac2e\right)$ for the original integral. – Travis Willse Dec 17 '24 at 22:27
  • To be explicit, the singularity of $f$ at $0$ is removable because its numerator is $\frac16 z^3 + R(z)$ and its denominator is $z^3 + S(z)$ for some remainders $R(z), S(z) \in O(z^5)$, so in a series about $0$, $f \sim \frac16 + T(z)$, $T(z) \in O(z^2)$. – Travis Willse Dec 17 '24 at 22:30
  • This question is similar to: Solve the improper integral with techniques of complex analysis. If you believe it’s different, please [edit] the question, make it clear how it’s different and/or how the answers on that question are not helpful for your problem. – Travis Willse Dec 17 '24 at 22:34
  • The duplicate asks about an integral identical up to the even symmetry of the integrand. – Travis Willse Dec 17 '24 at 22:35

1 Answers1

1

Consider the function $f(z)=\frac{z+ie^{iz}}{z^3(1+z^2)}$. Note that this function has a pole of degree $3$ at $z=0$.

Now consider the Contour $C$ below.

enter image description here

We have:

$$2\pi i\{Res_{z=i} f(z)\}=\int_C f(z)dz\\=\int_r^R \frac{x-\sin x+i\cos x}{x^3(1+x^2)}dx+\int_{C_R} f(z)dz+\int_{-R}^{-r} \frac{x-\sin x+i\cos x}{x^3(1+x^2)}dx+\int_{C_r} f(z)dz.$$

When $R \to \infty $, it's easy to see that $\int_{C_R} f(z)dz \to 0.$ However, we need to be more careful about the Contour $C_r.$ In fact, we have:

$$\int_{C_r} f(z)dz=\int_{C_r} \frac{z+i-z+\frac{i(iz)^2}{2!}+\frac{i(iz)^3}{3!}+\frac{i(iz)^4}{4!}+...}{z^3(1+z^2)}dz.$$ However, note that for $|z| \leq 1$:

$$|\int_{C_r} \frac{\frac{i(iz)^3}{3!}+\frac{i(iz)^4}{4!}+...}{z^3(1+z^2)}dz| \leq (\frac{1}{3!}+\frac{1}{4!}+...) \pi r.$$

Therefore when $r \to 0$,

$$\int_{C_r} \frac{\frac{i(iz)^3}{3!}+\frac{i(iz)^4}{4!}+...}{z^3(1+z^2)}dz \to 0.$$

On the other hand,

$$\int_{C_r} \frac{z+i-z+\frac{i(iz)^2}{2!}}{z^3(1+z^2)}dz=\int_{C_r} \frac{i}{z^3(1+z^2)}dz+\int_{C_r} \frac{-iz^2}{2z^3(1+z^2)}dz.$$

Observe that:

$$\int_{C_r} \frac{i}{z^3(1+z^2)}dz=\int_{C_r} \frac{i}{z^3}dz+\int_{C_r} \frac{iz}{z^2+1}dz-\int_{C_r} \frac{i}{z}dz.$$

It's easy to see that the first integral above is zero. The second one is not dependent on the path when $r \to 0$ (because it would be analytic), so: $$\int_{C_r} \frac{iz}{z^2+1}dz=\int_{-r}^r \frac{ix}{x^2+1}dx=0.$$

And,

$$-\int_{C_r} \frac{i}{z}dz=-\int_{\pi}^0 \frac{ri \times ie^{i\theta}}{re^{i\theta}}d\theta=-\pi.$$

Also,

$$\int_{C_r} \frac{-iz^2}{2z^3(1+z^2)}dz=\int_{\pi}^0\frac{-i \times ire^{i\theta}r^2e^{2i\theta}}{2r^3e^{3i\theta}(1+r^2e^{2i\theta})}d\theta=\int_{\pi}^0\frac{1}{2(1+r^2e^{2i\theta})}d\theta.$$

Therefore, when $r \to 0$, we have: $\int_{C_r} \frac{-iz^2}{2z^3(1+z^2)} \to \frac{-\pi}{2}.$

So, when $r\to 0$,

$$\int_{C_r} f(z)dz \to \frac{-3\pi}{2}.$$

Finally,

$$\int_{-\infty}^{\infty} \frac{x-\sin x}{x^2(1+x^2)}dx=Re \{ 2\pi i \{Res_{z=i}f(z)\}+\frac{3\pi}{2}\}=Re \{2\pi i \frac{i(1+e)}{2e}+\frac{3\pi}{2}\}=\frac{\pi (e-2)}{2e}.$$


The answer above is the same as what Wolfram gives:

enter image description here

Reza Rajaei
  • 6,236