3
$\begingroup$

Proposed: An approximation to n factorial in terms of $\pi$

$$\int_{0}^{\pi/4}x^n\sqrt{1+\sin(2x)}\mathrm dx=n!+F(n)\tag1$$ $n\ge1$

Where $F(n)$ is expressed in terms of $\pi$ for $n\ge2$

We notice that $$\lim_{n\to \infty }n!= F(n)$$

Examples:

$$\color{blue}{1\approx \sqrt{2}}$$

$$2\approx {\pi\over \sqrt{2}}$$

$$6\approx {\pi(32-\pi^2)\over 8\sqrt{2}}$$

$$24\approx {\pi(96-\pi^2)\over 8\sqrt{2}}$$

$$120\approx {5(6144-192\pi^2+\pi^4)\over 128\sqrt{2}}$$

$$720\approx {3\pi(30720-320\pi^2+\pi^4)\over 256\sqrt{2}}$$

My try:

Recall $$(1+x)^{1/2}=\sum_{k=0}^{\infty}{(-1/2)_k\over k!}(-x)^k$$ then $(1)$ becomes

$$\sum_{k=0}^{\infty}{(-1)^k\over k!}(-1/2)_k\int_{0}^{\pi/4}x^n\sin^k(2x)\mathrm dx\tag2$$

Apply IBP to

$$\int x^n\sin^{k}(2x)\mathrm dx={x^{n+1}\over n+1}\sin^k(2x)-{2k\over n+1}\int x^{k+1}\cos(2x)\sin^{k-1}(2x)\mathrm dx\tag3$$

How may one evaluate the closed form for $(1)?$

$\endgroup$
  • 1
    $\begingroup$ Observe $\sqrt{1+\sin(2x)}=\sin(x)+\cos(x)=\sqrt{2}\,\cos(\pi/4-x)$. It should be $n!-F(n)$, because the integral itself should be $O\left(\left(\frac{\pi}{4}\right)^n\right)$. $\endgroup$ – Professor Vector May 19 '17 at 8:53
  • $\begingroup$ Formulas for $(1)$ are written below. I've written a closed form. :-) $\endgroup$ – user90369 Aug 29 '17 at 8:35
1
$\begingroup$

$n\in\mathbb{N}_0$ :

$$\int\limits_0^{\pi/4} x^n\sqrt{1+\sin(2x)}dx=\int\limits_0^{\pi/4} x^n\cos x dx +\int\limits_0^{\pi/4} x^n\sin x dx$$

$$=n![t^n]\frac{1+t(\sqrt{2}e^{t\pi/4}-1)}{1+t^2}=n!\sum\limits_{k=0}^n\frac{a_k}{k!}\cos(\frac{\pi}{2}(n-k))$$

with $\enspace a_0=1\enspace$, $\enspace a_1=\sqrt{2}-1\enspace$, $\enspace \displaystyle a_{k\geq 2}=\sqrt{2}k(\frac{\pi}{4})^{k-1}$

$\endgroup$
0
$\begingroup$

Mathematica gives a closed(ish) form for the integral. \begin{equation} \int_0^{\pi/4}x^n\sqrt{1+\sin(2n)}\;dx = (\frac{1}{2}+\frac{i}{2})i^{-n}\left(((-1)^n-i)n!+(-1)^{n+1}\Gamma(n+1,-\frac{i\pi}{4})+i\Gamma(n+1,\frac{i\pi}{4}) \right) \end{equation} It contains imaginary numbers due to the way the incomplete gamma function is defined, but the results are all real for integer $n$, so your $F(n)$ function can be that minus $n!$. I would not know how to get there by hand.

$\endgroup$

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.