4
$\begingroup$

$n$ is a positive integer, then

$$\prod_{k=1}^n\frac{2k-1}{2k}\lt\frac1{\sqrt{3n}}$$


with mathematical induction, we can prove this.

But I would love to find a wonderful method without mathematical induction

Thank you!

$\endgroup$
  • 2
    $\begingroup$ Can one refer to $\binom{2n}{n}$? $\endgroup$ – André Nicolas Jun 7 '14 at 7:28
  • $\begingroup$ @AndréNicolas Yes, sure $\endgroup$ – 2016 Jun 7 '14 at 7:28
  • 1
    $\begingroup$ math.stackexchange.com/questions/58560/… $\endgroup$ – user940 Jun 7 '14 at 15:37
  • 1
    $\begingroup$ This question gives the better upper bound $\frac1{\sqrt{\pi n}}$. $\endgroup$ – user21467 Jun 7 '14 at 15:37
4
$\begingroup$

This is a beautiful inequality and, if you know or are allowed to use the $\Gamma$ function, the lhs simplifies to $$\prod_{k=1}^n\frac{2k-1}{2k}=\frac{\Gamma \left(n+\frac{1}{2}\right)}{\sqrt{\pi } \Gamma (n+1)}$$ For large values of $n$, an asymptotic development is given by $$\frac{\Gamma \left(n+\frac{1}{2}\right)}{\sqrt{\pi } \Gamma (n+1)}=\frac{\sqrt{\frac{1}{n}}}{\sqrt{\pi }}-\frac{\left(\frac{1}{n}\right)^{3/2}}{8 \sqrt{\pi }}+\frac{\left(\frac{1}{n}\right)^{5/2}}{128 \sqrt{\pi }}+O\left(\left(\frac{1}{n}\right)^{7/2}\right)$$ and the difference between lhs and rhs is found to always be negative and to increase to $0^-$. For the difference, we have $$rhs-lhs=\left(\frac{1}{\sqrt{\pi }}-\frac{1}{\sqrt{3}}\right) \sqrt{\frac{1}{n}}-\frac{\left(\frac{1}{n}\right)^{3/2}}{8 \sqrt{\pi }}+\frac{\left(\frac{1}{n}\right)^{5/2}}{128 \sqrt{\pi }}+O\left(\left(\frac{1}{n}\right)^{7/2}\right)$$

$\endgroup$
3
$\begingroup$

Building on André Nicolas' idea, you may rewrite $\prod_{k=1}^n\frac{2k-1}{2k}$ as

$$\frac{1}{2}\cdot\frac{3}{4}\cdot\frac{5}{6}\cdot\frac{7}{8}\cdots\frac{2n-1}{2n}=\frac{1\cdot 2\cdot 3\cdot 4\cdot 5\cdots (2n-1)2n}{(2\cdot 4 \cdot 6\cdot 8\cdots 2n)^2}=\frac{(2n)!}{(2^n\cdot 1\cdot 2\cdot 3\cdots n)^2}=\frac{1}{4^n}\binom{2n}{n}.$$

Now, $\binom{2n}{n}$ is the central binomial coefficient. Its asymptotic is both well-known and easily determined as

$$\binom{2n}{n}\sim \frac{4^n}{\sqrt{\pi n}},$$

so that asymptotically,

$$\prod_{k=1}^n \frac{2k-1}{2k}\sim \frac{1}{\sqrt{\pi n}}<\frac{1}{\sqrt{3n}}\quad(\pi=3.1416...).$$

Moreover, wikipedia also lists a property of central binomial coefficients which directly proves your result, namely,

$$\binom{2n}{n}\le \frac{4^n}{\sqrt{3n+1}}<\frac{4^n}{\sqrt{3n}}.$$

$\endgroup$

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.