I am trying to prove that the improper integral $\int_{0}^{\infty} \cos(x^3)\mathrm{d}x$ is convergent (which seems counterintuitive since the integrand doesn't tend to $0$ as $x$ tends to $\infty$). I've tried integration by parts or looking for an appropriate trigonometric expression, but I don't get successful results. Any ideas?

  • 1
    $\begingroup$ Let $u = x^3$, with $dx = \frac{du}{3u^{2/3}}$. Then integrate by parts. $\endgroup$ – Joey Zou Sep 25 '16 at 8:07
  • $\begingroup$ ... you should obtain $\dfrac{\Gamma(\frac{1}{3})}{2\sqrt{3}}$. $\endgroup$ – Jean Marie Sep 25 '16 at 8:10
  • $\begingroup$ Thanks! Just realised the power of Euler's Gamma Function! However, the problem was posed in a part of the problem set where Euler's Gamma Function wasn't still introduced (note that it is only asking for convergence, not a concrete value), so I would appreciate any other solution regarding standard convergence criteria for improper integrals (direct comparison, Dirichlet, etc). $\endgroup$ – Mr Peanutbutter Sep 25 '16 at 9:16
  • $\begingroup$ Now, I have realized that in fact, I have fullfilled the mission (proving the convergence of the integral) by bringing back the issue to a known (converging) integral. But I understand that you would prefer a proof by majorations, bracketting between convergent series, etc... $\endgroup$ – Jean Marie Sep 25 '16 at 10:10

I address the initial question by proving that the looked for integral is a recognized integral as a certain cosine transform. Working this way, we obtain the exact value of this integral.

Let us start from the expression given by the change of variables $X^3=x \Longleftrightarrow X=x^{1/3}$ suggested by @Joey Zou:

$$\tag{1}I:=\int_0^{\infty}cos(X^3)dX=\frac{1}{3}J \ \ \text{with} \ \ J:=\int_0^{\infty}cos(x)x^{-2/3}dx$$

Let us establish that:

$$\ \tag{2} I=\dfrac{\Gamma(\frac{1}{3})}{2\sqrt{3}}$$

by using the following result, deduced from a table of cosine transforms (classical tables of Gradshteyn and Ryzhik), valid for all $\nu$ such that $0 \leq \nu \leq 1$.

$$\ \sqrt{\frac{2}{\pi}}\int_0^{\infty}\cos(\xi x)x^{-\nu}dx=\sqrt{\frac{\pi}{2}}\dfrac{1}{\Gamma(\nu)}\dfrac{1}{\cos(\frac{\pi \nu}{2})}\xi^{\nu-1}$$

With $\xi=1$ and $\nu=\frac{2}{3}$, we get

$$\ \tag{3} J=\int_0^{\infty}\cos(x)x^{-\frac{2}{3}}dx=\dfrac{\pi}{2}\dfrac{1}{\Gamma(2/3)}\dfrac{1}{\cos(\frac{\pi}{3})}=\pi\dfrac{1}{\Gamma(2/3)}$$

Setting $x=1/3$ in the complement's formula

$$\Gamma(x)\Gamma(1-x)=\dfrac{\pi}{\sin{\pi x}}$$

we have:

$$\tag{4}\Gamma(2/3)=\dfrac{2\pi}{\sqrt{3} \Gamma({1/3})}$$

Taking account of (3) and (4) in the definition of $J$ (relationship (1)) gives (2).

As an additional information, here is a graphical representation of $f$ defined by $f(x)=\cos(x^3)$. The convergence of its integral over $(0,\infty)$ can be understood by a mutual (approximate) annihilation of the positive and negative areas of the increasingly smaller "spikes" of the curve.

For example $\ \int_0^{3}cos(x^3)dx\approx 0.808977$ is rather close to the exact value $\ \int_0^{\infty}cos(x^3)dx=\dfrac{\Gamma(\frac{1}{3})}{2\sqrt{3}}\approx 0.773343$.

enter image description here

| cite | improve this answer | |

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.