0

I have to check the convergence of the improper integral $$ \int_{0}^{\infty} \cos^2x\,dx .$$

I have tried to solve it in the following manner: $$\begin{align}\int_{0}^{\infty}\cos^2x\,dx &= \lim_{B \to \infty} \int_{0}^{B}\cos^2x\,dx \\&= \lim_{B \to \infty} \int_{0}^{B}(1-\sin^2(x))\,dx \\ &= \lim_{B \to \infty}[x]_{0}^{B} -\lim_{B \to \infty} \int_{0}^{B}\sin^2(x)\,dx .\end{align}$$

Now the first term tends to infinity as $B \to \infty,$ so the integral is not convergent.

But unfortunately the answer is given to be convergent. I don't understand how ?

Also I have proven in the same way that $$ \int_{0}^{\infty} (\sin x^2)^2\,dx $$ is divergent. So basically I think that if my process is wrong then I am wrong in both cases.

Am I wrong ? If so, looking for guidance then. Thank you.

Decaf-Math
  • 4,522
hiren_garai
  • 1,193

1 Answers1

3

You cannot proceed this way, as you end up with something of the form $$ \int_0^B dx\, \cos^2 x = u_B - v_B $$ and $\lim_{B\to\infty} u_B = \infty$. But if $\lim_{B\to\infty} v_B = \infty$ as well, you have an indeterminate form and cannot conclude.

However, note that $\cos^2$ is perdiodic with period $\pi$, so that for any integer $k$ $$ \int_0^{k\pi} dx\, \cos^2 x = k\int_0^\pi dx\, \cos^2 x = k\cdot \frac{\pi}{2} $$ and therefore $$ \int_0^{B} dx\, \cos^2 x = \int_0^{\lfloor B/\pi\rfloor \pi} dx\, \cos^2 x + \underbrace{\int_{\lfloor B/\pi\rfloor \pi}^B dx\, \cos^2 x}_{\geq 0} \geq \lfloor B/\pi\rfloor\cdot \frac{\pi}{2} \xrightarrow[B\to\infty]{} \infty $$

Clement C.
  • 67,323
  • Thank you for your answer. In a similar way I can prove the divergence of $$ \int_{0}^{\infty} \text{(sin x²)² dx} $$, right ? – hiren_garai Oct 01 '18 at 14:01
  • @hiren_garai yes, indeed. – Clement C. Oct 01 '18 at 14:02
  • 1
    I have one more query, can I visualise the problem in the way that it is the integration of a positive term from $0$ to $\infty$ and so the area of the curve will be unbounded and hence the integral is divergent ? It's just to find a rough idea about the problem . @ClementC. – hiren_garai Oct 01 '18 at 14:09
  • 1
    Of course, one may write $$\int_0^L \cos^2(x),dx=\frac12 L+\frac14 \sin(2L) $$and conclude. – Mark Viola Oct 01 '18 at 14:37
  • I got it.@MarkViola. thanks for your time. – hiren_garai Oct 01 '18 at 14:48
  • @MarkViola Thanks for the comment (and indeed, once can just write the explicit close-form integral, as you point out). The periodic "trick" is helpful in general, but not necessary here. – Clement C. Oct 01 '18 at 15:09
  • Hi Clement! It's always nice to see you on MSE. And indeed, the periodic "trick" has broad application. (+1) – Mark Viola Oct 01 '18 at 15:25