2
$\begingroup$

show that:

$$\prod_{k=1}^{n-1}\sin{\dfrac{k\pi}{tn}}=\dfrac{\sqrt[t]{n}}{2^{n-1}}?(not, true),t\in N^{+}$$

maybe for $t$is real numbers also true?

I can show when $t=1$ case. because I use $$z^{n-1}-1=(z-x_{1})(z-x^2_{1})\cdots (z-x^{n-1}_{1}),x_{1}=\cos{\dfrac{2\pi}{n}}+i\sin{\dfrac{2\pi}{n}}$$ also see Prove that $\prod_{k=1}^{n-1}\sin\frac{k \pi}{n} = \frac{n}{2^{n-1}}$

But for $t\neq 1$,I can't use this identity

$\endgroup$
5
  • $\begingroup$ Is $t$ an integer? $\endgroup$ Commented Dec 30, 2014 at 5:56
  • $\begingroup$ But maybe for $t\in R$ is also true? $\endgroup$
    – math110
    Commented Dec 30, 2014 at 6:21
  • 1
    $\begingroup$ I do not think the statement is true for $ t = 3, n = 2$. It states that $ \sin \frac{ \pi} { 3 \times 2} = \frac{ \sqrt[3]{2} } { 2} $. It is clear the RHS is not equal to $ \frac{1}{2} $. Did I make a mistake? $\endgroup$
    – Calvin Lin
    Commented Dec 30, 2014 at 6:36
  • 1
    $\begingroup$ @CalvinLin you're right, and I too was skeptical of the formula. $\endgroup$
    – kobe
    Commented Dec 30, 2014 at 6:46
  • $\begingroup$ @CalvinLin,if my is not true,then How find it? $\endgroup$
    – math110
    Commented Dec 30, 2014 at 7:05

1 Answer 1

2
$\begingroup$

Another reason for which the formula cannot be true is that $\log(\sin x)$ is an integrable function over $(0,1)$, hence for $t\in\mathbb{N}$ it is expected that: $$\sum_{k=1}^{n-1}\log\sin\frac{\pi k}{t n}\approx n\int_{0}^{\pi/t}\log\sin x\,dx \approx -n\frac{1+\log\frac{t}{\pi}}{t}$$ by Riemann sums, while: $$\log\frac{\sqrt[t]{n}}{2^{n-1}}\approx -n\log 2+\frac{\log n}{t}$$ has the wrong magnitude.

To compute the product, a possibility is given by the Fourier series identity: $$-\log\sin x = -\log 2-\sum_{k\geq 1}\frac{\cos(2kx)}{k}. $$

$\endgroup$
1
  • $\begingroup$ Nice,But we can find this closed form? $\endgroup$
    – math110
    Commented Dec 30, 2014 at 10:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .