7
$\begingroup$

(A related MSE question by P. Singh.) First define,

$$F_k = 1-5\left(\frac{1}{2}\right)^k+9\left(\frac{1\cdot 3}{2\cdot 4}\right)^k-13\left(\frac{1\cdot 3\cdot 5}{2\cdot 4\cdot 6}\right)^k+17\left(\frac{1\cdot 3\cdot 5\cdot 7}{2\cdot 4\cdot 6\cdot 8}\right)^k-\cdots$$

Or more concisely,

$$F_k = \sum_{n=0}^\infty\, (-1)^n\,(4n+1) \left(\frac{\Gamma\big(n+\tfrac{1}{2}\big)}{\Gamma\big(n+1\big)\;\Gamma\big(\tfrac{1}{2}\big)}\right)^k$$

The gamma quotient is also,

$$ \frac{\Gamma\big(n+\tfrac{1}{2}\big)}{\Gamma\big(n+1\big)\;\Gamma\big(\tfrac{1}{2}\big)} = \frac1{2^{2n}}\frac{(2n)!}{\,n!^2}$$

Then we have the nice closed-forms,

\begin{align} F_1 &= 0\\[4pt] F_2 &= \dfrac{2\sqrt2\,\Gamma\big(\tfrac{1}{2}\big)}{\Gamma^2\big(\tfrac{1}{4}\big)}\\[4pt] F_3 &= \dfrac{2}{\Gamma^2\big(\tfrac{1}{2}\big)}=\dfrac{2}{\pi} \\[4pt] F_4 &= \; \color{red}{??}\\[4pt] F_5 &= \dfrac{2}{\Gamma^4\big(\tfrac{3}{4}\big)} \end{align}

I found $F_2$ empirically. The closed-form of $F_3=\large\frac{2}{\pi}$ was included by Ramanujan in his letter to Hardy, and $F_5$ is also by him. (I presume a version of $F_2$ may be in his Notebooks.)


Alternatively, while the $F_k$ are clearly a sum of two generalized hypergeometric functions $_pF_q$, we learn from this post that they can also be expressed as just one $_pF_q$,

\begin{align} F_1 &= \,_2F_1\big(\tfrac12,\tfrac54;\tfrac14;-1\big)\\[4pt] F_2 &= \,_3F_2\big(\tfrac12,\tfrac12,\tfrac54;\tfrac14,1;-1\big)\\[4pt] F_3 &= \,_4F_3\big(\tfrac12,\tfrac12,\tfrac12,\tfrac54;\tfrac14,1,1;-1\big)\\[4pt] F_4 &= \,_5F_4\big(\tfrac12,\tfrac12,\tfrac12,\tfrac12,\tfrac54;\tfrac14,1,1,1;-1\big)= \; \color{red}{??}\\[4pt] F_5 &= \,_6F_5\big(\tfrac12,\tfrac12,\tfrac12,\tfrac12,\tfrac12,\tfrac54;\tfrac14,1,1,1,1;-1\big) \end{align}

and so on.


Question: What is the closed-form of $F_4,\,F_6$ and others, if any?

$\endgroup$
10
  • $\begingroup$ Maybe you can reduce the hypergeometic closed form here? $\endgroup$ Commented Oct 27, 2023 at 13:18
  • $\begingroup$ @ТymaGaidash Wolfram can actually reduce $F_5$, but I don't know why it has trouble with $F_4$. $\endgroup$ Commented Oct 27, 2023 at 13:23
  • 1
    $\begingroup$ @TitoPiezasIII Once again, an interesting question! Do you have a numerical expansion of $F_4$? Might be worth checking out such a representation with an inverse symbolic calculator $\endgroup$
    – Max Muller
    Commented Oct 27, 2023 at 15:44
  • 1
    $\begingroup$ @MaxMuller Thanks. This WA command should give it to 100 digits, It starts as 0.79443356000157023489670877173541913133327484204150... I assumed it was a product of gamma functions and some radicals, but no hits on Mathematica's integer relations so far. $\endgroup$ Commented Oct 27, 2023 at 16:58
  • 1
    $\begingroup$ Although not quite the same, we have $$\int_0^1K^2dk=\frac{\pi^4}{32}\sum_{n=0}^{\infty}\binom{2n}{n}^6\left(\frac{1+4n}{2^{12n}}\right)$$ $$\int_0^1kK^4dk=\frac{3\pi^6}{256}\sum_{n=0}^{\infty}\binom{2n}{n}^8\left(\frac{1+4n}{2^{16n}}\right)$$ $\endgroup$ Commented Feb 23 at 5:17

0

You must log in to answer this question.

Browse other questions tagged .