I am trying to prove this below. $$ I:=\int_0^\infty \frac{\cos x}{x}\left(\int_0^x \frac{\sin t}{t}dt\right)^2dx=-\frac{7}{6}\zeta_3 $$ where $$ \zeta_3=\sum_{n=1}^\infty \frac{1}{n^3}. $$ I am not sure how to work with the integral over $t$ because it is from $0$ to $x$. If we can somehow write $$ \int_0^\infty \frac{\cos x}{x} \left(\int_0^\infty \frac{\sin t}{t}dt-\int_x^\infty \frac{\sin t}{t}dt \right)^2dx=\int_0^\infty \frac{\cos x}{x}\left(\frac{\pi}{2}-\int_x^\infty \frac{\sin t}{t}dt\right)^2dx. $$ I do not want to use an asymptotic expansion on the integral over $t$ from $x$ to $\infty$, I am looking for exact results. Note we can use $\int_0^\infty \frac{\sin t}{t}dt=\int_0^\infty \mathcal{L}[\sin t(s)]ds=\frac{\pi}{2}.$ Other than this approach I am not really sure how to go about this. Note by definition $$ \int_0^x \frac{\sin t}{t}dt\equiv Si(x), $$ but I'm not too sure what this definition can be used for in terms of a proof. Also note $$ \int_0^\infty \frac{\cos x}{x}dx \to \infty. $$

  • $\begingroup$ As quick hint, exand the integrant of the inernal integral into series, seems to me that will do the job. $\endgroup$ – TMS May 24 '14 at 23:44
  • $\begingroup$ I'd have written $\zeta(3)$ rather than $\zeta_3$. ${}\qquad{}$ $\endgroup$ – Michael Hardy May 24 '14 at 23:45
  • $\begingroup$ @MichaelHardy Okay that is great to know thanks a lot. $\endgroup$ – Jeff Faraci May 24 '14 at 23:47
  • 4
    $\begingroup$ See here for a nice solution: integralsandseries.prophpbb.com/… scroll to the bottom thread $\endgroup$ – Cody May 25 '14 at 0:00
  • 1
    $\begingroup$ @Integrals : Okay.. I wish Shobhit himself to post it here since he's the one who answered there. $\endgroup$ – gar May 25 '14 at 16:47

It's strange that you're unable to access that link. I do not understand that.

I accredit this to Shobit.

Anyway, what Shobit done was to write it as a triple integral.



Now, use the known result: $$\int_{0}^{\infty}\frac{\cos(bx)-\cos(ax)}{x}dx=\log(a/b)$$

Using this, it can now be written in terms of a log:


Now, integrating this w.r.t y is where the dilogs come into play:


Maybe you can finish it now on your own. It is now a matter of known dilog integrals. This is the bulk of it. But, if you need more I can write the rest later.

  • 1
    $\begingroup$ Thanks. This is fine to finish from here. I didn't see the Frullani like integral at all. And obviously writing it as a triple integral is very nice insight, I thought of neither. +1 $\endgroup$ – Jeff Faraci May 26 '14 at 22:09
  • $\begingroup$ Thanks Integral. If Shobit ever shows back up, please go ahead and give him a greenie because, afterall, it is his answer. I was just helping out because you could not access the link. $\endgroup$ – Cody May 27 '14 at 8:52

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.