5

enter image description here

The diagram shows a square of area $An$ and an enclosed quarter-circle.
Line segments are drawn from the bottom-left vertex to points that are equally spaced along the quarter-circle.
The regions enclosed by the line segments and the quarter-circle have areas $a_1, a_2, a_3, ..., a_n$.

Find the value of $A$ such that $\lim\limits_{n\to\infty}\prod\limits_{k=1}^n a_k = 2$.

I have parced the problem to this:

$$\lim\limits_{n\to\infty}\prod\limits_{k=1}^n Anf(k)(g(k)-g(k-1))=2$$

where

$f(k)=\frac{3}{2}-\sin{\frac{k\pi}{2n}-\cos{\frac{k\pi}{2n}}}$

$g(k)=\arcsin{\left(\dfrac{\sin{\frac{k\pi}{2n}}-\cos{\frac{k\pi}{2n}}}{2\sqrt{f(k)}}\right)}$

Desmos suggests $A\approx 5.77987$. I am looking for a closed form for $A$.

I have tried to take the log of the product and relate the resulting sum to an integral, but I do not know how to deal with the $g(k-1)$.

(This question was inspired by a related question, where the product of the lengths of the line segments approaches $2$.)

EDIT

After some more exploration, it seems that $A$ also makes $\frac{1}{n}\prod\limits_{k=1}^{n}A\frac{\pi}{4}\left(\cos{\frac{k\pi}{2(n+1)}}+\sin{\frac{k\pi}{2(n+1)}}-1\right)$ converge to a positive number (approximately $0.8817$).

EDIT2

Based on this question, it seems that $A=\frac{4}{\pi}e^{4G/\pi}\sqrt2$ where $G$ is Catalan's constant.

Dan
  • 22,158
  • Can you provide a few more decimal places for the value of $A$? There's something I would like to try and confirm. – HackR Oct 24 '22 at 12:56
  • I can't really get a better approximation. I'm using desmos and adjusting the value of $A$ so that the product seems to converge (to $2$), because desmos only calculates the product up to $n=4370$. – Dan Oct 24 '22 at 13:00
  • Ah that's okay, no worries. – HackR Oct 24 '22 at 13:02
  • Dan, I am curious. I knew it was you just by reading the title… what motivates you to consider all these geometric products? +1 – FShrike Oct 24 '22 at 13:23
  • @FShrike I guess I like the combination of geometry and infinity. It's not obvious to me that these infinite products should exist, but they do. Circles in particular seem to yield a lot of these nice infinite products. – Dan Oct 24 '22 at 13:40
  • 1
    @FShrike It seems like infinite product of areas has not been studied very much. A search for the tag [infinite-product] yields 1401 results, but a search for the tags [infinite-product] and [area] yields 3 results, all of which are my questions. – Dan Oct 24 '22 at 14:42
  • 1
    Well, I have always enjoyed reading through your product questions – FShrike Oct 24 '22 at 14:43
  • 1
    What a fascinating question. Thanks for posting. – DecarbonatedOdes Oct 24 '22 at 23:17
  • 1
    With Mathematica, $5.7798725551614$ gives $1.99999999998$ with 10,000 iterations. – DecarbonatedOdes Oct 25 '22 at 00:47

0 Answers0