2

$\int_{0}^{\infty} \frac{(\sin{x})^2}{x} \ dx $ converges?


The question has been answered in:


I understood every solution that were presented in the links above, but I didn't understand the intuition behind it. How do you know whether to prove or disprove that $\int_{0}^{\infty} \frac{(\sin{x})^2}{x} \ dx $ converges in the first place? It is important for me to understand the thought process behind it.

Thank you

Lior
  • 623
  • I'm not at all a mathematician, so this may be useless, but if you see the integral as a Riemann sum, of small terms sin(0+dx)/0+dx * dx adding many times, you get sin(0 + n dx)/ n dx and this goes to 0 as you add up infinitely small terms. Or in wolfram alpha https://www.wolframalpha.com/input?i=%28sin%5E2%28x%29%29%2Fx&assumption=%22ClashPrefs%22+-%3E+%7B%22Math%22%7D –  Jul 11 '23 at 13:10
  • 1
    When I have to decide wether an integral of this kind converges or not I try to look at a generalized power series expansion of the integrand function. In this case, for instance, you can check that you end up with something of order $\mathcal{O}(x)$ without oscillations (since the intgrand is non-negative) therefore you could easily guess that it is divergent. However this trick only works if you work with good enough functions... – Matteo Menghini Jul 11 '23 at 13:12
  • @MatteoMenghini If it's not too much to ask I'd like to see how you solve it. I learnt so much from you yesterday (and I want to be sure I understand you well). – Lior Jul 11 '23 at 13:16
  • @Minsky I'm not sure I understand your what sin(0+dx)/0+d stands for. But thank you anyway :) – Lior Jul 11 '23 at 13:19
  • I don't think the question is so much about “whether to prove or disprove” that the integral converges, it's whether the integral converges or diverges. Whichever is the case, that's what you should prove. – Matthew Leingang Jul 11 '23 at 15:35

3 Answers3

3

This proof was in one of the posts you linked, but hopefully I can present it in a way that explains the visual intuition behind it a bit more, and why this is actually more general than just $\sin(x)^2$.

In a sense, the fact that it is a sine wave isn't really the important part. You can probably sense that taking a slightly different shape than $\sin(x)^2$ won't affect whether or not this integral converges. In my mind, the convergence has more to do with the fact that $\sin(x)^2$ is akin to a periodic step function. We can indeed bound it with one as you can see below.

So instead of $\sin(x)^2$, let's consider the periodic function: $$ f(x) := \begin{cases} 0 &\text{if } \lfloor x\rfloor \text{ is even}, \\ 1 &\text{if } \lfloor x\rfloor \text{ is odd}, \end{cases} $$ which just alternates between $0$ and $1$ every interval of length $1$. Now we can consider the simpler integral: $$ \int_0^\infty \frac{f(x)}{x}\,dx. $$ Here's a graph of what $\frac{f(x)}{x}$ looks like

Notice that our integral becomes a sum of integrals of $\frac{1}{x}$ over every other interval of length $1$: $$ \int_0^\infty \frac{f(x)}{x}\,dx = \sum_{n=1}^{\infty} \int_{2n-1}^{2n} \frac{1}{x}\,dx, $$ but since $x\mapsto\frac{1}{x}$ is a decreasing function, we can bound it by the upper limit of these integrals, and so: $$ \int_0^\infty \frac{f(x)}{x}\,dx \geq \sum_{n=1}^{\infty} \int_{2n-1}^{2n} \frac{1}{2n}\,dx = \sum_{n=1}^{\infty} \frac{1}{2n}. $$ But the harmonic series $\sum_{n=1}^{\infty} \frac{1}{n}$ is famously divergent, and so our integral diverges too! In fact, one of the ways to show the harmonic series diverges is to compare it to the whole integral of $\frac{1}{x}$ over $(1,+\infty)$.

Those arguments would be true for any positive periodic function, since you can always bound them by periodic step functions. A possible way to think about it is this: periodically cutting out parts of $\frac{1}{x}$ is never enough to make the integral converge, since you can always compare it to a rescaled version of $\frac{1}{x}$. That itself is true because rescaling the $x$-axis by some constant $\alpha$ just amounts to rescaling $\frac{1}{x}$ by $\frac{1}{\alpha}$.

That's not to say that this an obvious characteristic you should remember by heart. I think the better lesson is to play around with the parameters of an integral, try to see what affects it the most drastically, and really focus on that. Everything else is just flavor, which tends to get in the way of simpler formulas.

CiaPan
  • 13,049
Willow Wisp
  • 1,153
  • 2
    Nice. Even more crudely, you can think about $\sin^2 x$ as “hovering” around $1/2$. So the convergence of $\int_1^\infty \frac{\sin^2x}{x},dx$ should be (again, crudely) comparable to $\int_1^\infty \frac{1/2}{x},dx$ – Matthew Leingang Jul 11 '23 at 15:40
  • 1
    Wow, what a wonderful answer. You made me get the idea for real. Thank you!! – Lior Jul 17 '23 at 18:24
2

Before we begin, recall

$$\int_{0}^{\infty} {\frac{1}{x}}\hspace{1pt}dx=\infty.$$

In fact for any $t\in{\mathbb{R}\setminus\{0\}}$ we have

$$\int_{0}^{\infty} {\frac{1}{tx}}\hspace{1pt}dx=\infty.$$

Having that in mind we consider the behavior of $\sin^2(x)$ in order to determine if it has some average value. We have to realize now that $\sin^2(x)\geq{0}$ and moreover that the function has an average value of $1/2$ over intervals of the form $[(n\pi,(n+1)\pi]$, which we can use to partition the whole interval for our integral, hence we have

$$\int_{0}^{\infty} {\frac{\sin^2(x)}{x}}dx=\sum_{n=0}^{\infty} {\int_{n\pi}^{(n+1)\pi} {\frac{\sin^2(x)}{x}dx}}\approx\sum_{n=0}^{\infty} {\int_{n\pi}^{(n+1)\pi} {\frac{1}{2x}dx}}=\int_{0}^{\infty} {\frac{1}{2x}}dx=\infty.$$

This isn't a very rigorous argument by any means but we build a sufficient base to confidently proceed forward assuming our integral diverges. When assessing whether more complicated-looking integrals diverge I recommend going about it in this way, essentially stripping away everything until you have a function whose behavior you are comfortable with. At that point we can try to reinsert all the other parts, making sense of what each one is contributing to our integral.

If you would like to continue building your intuition consider trying to determine if these converge or not, they shouldn't be too difficult.

$$\int_{0}^{\infty} {\frac{e^{-x}\sin(x)}{x}}dx,\hspace{4pt}\int_{0}^{\infty} {\frac{e^{-x}\sin(x)}{x^2}}dx,\hspace{4pt}\int_{0}^{\infty} {\frac{e^{-x}\sin^2(x)}{x}}dx,\hspace{4pt}\int_{0}^{\infty} {\frac{\sin^{2n-1}(x)}{x}}dx$$

1

More precisely, since $\sin^2(x) \ge \frac12$ for $(n+\frac14)\pi \le x \le (n+\frac34)\pi$,

$\begin{array}\\ \int_{n\pi}^{(n+1)\pi}\dfrac{\sin^2(x)dx}{x} &\ge \int_{(n+\frac14)\pi}^{(n+\frac34)\pi}\dfrac{\sin^2(x)dx}{x}\\ &\ge \int_{(n+\frac14)\pi}^{(n+\frac34)\pi}\dfrac{dx}{2x}\\ &\ge \dfrac{\pi/2}{2(n+\frac34)\pi} \qquad \text{(width }\pi/2) \\ &\ge \dfrac1{4(n+1)}\\ \end{array} $

and the sum of these diverges.

marty cohen
  • 107,799