9
$\begingroup$

Can any natural number $n$ be expressed as $n=\lceil \frac{3^a}{2^b} \rceil$ or $n=\lfloor \frac{3^a}{2^b} \rfloor$ where $ a,b \in \mathbb{N}$ ?

I have found no approach to this problem. All suggestions are appeciated.

Examples:

$0=\lfloor \frac{3^0}{2^1} \rfloor$

$1=\lceil \frac{3^0}{2^1} \rceil$

$2=\lceil \frac{3^1}{2^1} \rceil$

$3=\lceil \frac{3^1}{2^0} \rceil$

$4=\lfloor \frac{3^2}{2^1} \rfloor$

$5=\lceil \frac{3^2}{2^1} \rceil$

$\endgroup$
12
$\begingroup$

Yes, all natural numbers $n$ can be represented as $n=\left\lceil \frac{3^a}{2^b} \right\rceil$ or as $n=\left\lfloor \frac{3^a}{2^b} \right\rfloor$ (whichever one you want). Let's do floors, for concreteness.

For $n=\left\lfloor \frac{3^a}{2^b} \right\rfloor$ you want positive integers $a$ and $b$ such that $n \le \frac{3^a}{2^b} < n + 1$, or equivalently $\log n \le a \log 3 - b \log 2 < \log(n + 1)$, or in other words $$\frac{\log n}{\log 2} \le \left(a \frac{\log 3}{\log 2} - b\right) < \frac{\log(n+1)}{\log 2}.$$

That is, if you call the ratio $\frac{\log 3}{\log 2}$ as $\theta$, then you want to find an integer $a$ such that the fractional part of $a\theta$ lies between the fractional part of $\frac{\log n}{\log 2}$ and that of $\frac{\log(n+1)}{\log 2}$.

This problem is known as “inhomogeneous diophantine approximation”. As $\theta$ is irrational, its fractional parts are dense and equidistributed in $[0, 1)$ (this is a theorem), so you can make the fractional part of $a\theta$ arbitrary close to any number you choose. There are many questions on this site about it; two I've answered are at Fractional part of $b \log a$ and at Prefix of Fibonacci number but you can also click on the “Linked” and “Related” on those questions for more.

$\endgroup$
  • $\begingroup$ (Well, to be pedantic, $0$ can't be represented as the ceiling of such a ratio… but otherwise it holds for all other nonnegative integers.) $\endgroup$ – ShreevatsaR May 21 '18 at 8:43
  • $\begingroup$ To be even more pedantic, not everyone considers $0$ a natural number. There exists a very popular distinction that only positive integers are natural, and nonnegative integers are whole. $\endgroup$ – gen-z ready to perish May 21 '18 at 8:51
  • $\begingroup$ @ChaseRyanTaylor Yes I know, but the question gave $0$ as an example — so considering that the OP may care about that case too, I chose to ignore the ambiguous terminology of “natural number” (which I repeated in the answer) and instead use “nonnegative integer” in the “pedantic” comment. $\endgroup$ – ShreevatsaR May 21 '18 at 8:53
  • $\begingroup$ Touché ${}{}{}{}{}{}{}{}$ $\endgroup$ – gen-z ready to perish May 21 '18 at 9:06
  • $\begingroup$ @ChaseRyanTaylor: Arguably, "whole" numbers could equally well be $\Bbb Z$, $\Bbb Z^+$, or $\Bbb N$. I prefer "counting numbers" for $\Bbb Z^+$, and think that the natural numbers $\Bbb N$ include $0$. $\endgroup$ – John Bentin Dec 19 '18 at 13:38

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.