2
$\begingroup$

In this question I asked to prove that

$$\int_{0}^{1}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx=\frac{\pi e}{24}.$$

If we take a look at the plot of the integrand, then we could see some symmetry-property.

enter image description here

Because the integrand is symmectic to the $y=1/2$ line, it is also true, that

$$\int_{0}^{1/2}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx=\int_{1/2}^{1}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx=\frac{\pi e}{48}.$$

Question. Is there a close-form of $I(a,b)=\int_{a}^{b}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx$ for some else $a<b$? If it is too hard, then is there a closed-form for some $a<b$, just for the real or imaginary part? For example

$$\Re \left[ \int_{1}^{3/2}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx \right],$$ $$\Re \left[ \int_{3/2}^{2}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx \right],$$ or $$\Im \left[ \int_{1}^{2}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx \right]?$$

$\endgroup$
  • $\begingroup$ I think we'd be pushing our luck asking for a closed form for general limits $a,b$. Closed forms might exist when the limits coincide with zeros of the integrand, perhaps. $\endgroup$ – David H Oct 11 '14 at 0:57
  • $\begingroup$ @DavidH This is why I asked these three specific integrand limits at the end of the question. Of course anything else would be nice. $\endgroup$ – user153012 Oct 11 '14 at 1:05
0
$\begingroup$

Is there a close-form of $I(a,b)=\displaystyle\int_{a}^{b}\sin{(\pi x)}x^x(1-x)^{1-x}\,dx$ for some other $a<b$?

Yes, of course there is. An infinity of them, in fact. But, in that case, a and b will lack a closed form. :-) For instance, all rational and algebraic numbers possess a closed form expression. Now, limiting ourselves to those in between $0$ and $\dfrac{e\pi}{24}$, which themselves are infinite in number, for each of them there are infinitely many solutions a and b within the interval $[0,1]$, so that $I(a,b)$ will be equal to that particular rational or algebraic value, since $f(t)=I(a,t)$ is a monotonous function for any t in $[a,1]$. So I assume that you are interested in triplets $\Big(a,b,I(a,b)\Big)$, other than the ones already mentioned, and where all three values have a closed form, in which case the answer would be that no others are known to exist $($key word here being “known”, since proving this is, to my knowledge, still an open problem$)$.

$\endgroup$
  • $\begingroup$ By closed form of an algebraic number I mean its minimal polynomial. I am not implying solvability in terms of radicals. $\endgroup$ – Lucian Oct 11 '14 at 2:18

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.