MHB Inf{frac{n*sqrt(3)} : n in integer}=0

  • Thread starter Thread starter caffeinemachine
  • Start date Start date
AI Thread Summary
The discussion centers on proving that the infimum of the fractional parts of multiples of the square root of 3, specifically $\inf \{ \text{frac}(n \sqrt{3}) : n \in \mathbb{Z}^+ \}$, is equal to zero. Participants note that the equidistribution theorem supports this claim, as the set of fractional parts is dense in the interval [0, 1]. A proof utilizing the Pigeonhole Principle is suggested, demonstrating that for sufficiently large n, there exist integers i and j such that their fractional parts are close enough to imply that the infimum approaches zero. Continued fractions are also mentioned as a constructive method to find approximations of $\sqrt{3}$ that yield small fractional parts. Overall, the consensus is that the infimum is indeed zero, with various approaches discussed to substantiate this conclusion.
caffeinemachine
Gold Member
MHB
Messages
799
Reaction score
15
Prove that $\inf \{ \text{ frac}(n \sqrt{3}) : n \in \mathbb{Z}^+ \}$ where $frac(x)$ is the fractional part of $x$
 
Mathematics news on Phys.org
caffeinemachine said:
Prove that $\inf \{ \text{ frac}(n \sqrt{3}) : n \in \mathbb{Z}^+ \}$ where $frac(x)$ is the fractional part of $x$
That is not a complete question.
But guessing at what it means, I think this theorem will help.
 
The OP misses "= 0," which is present in the thread title.

Equidistribution theorem is much stronger than the fact that $\{\mathop{\mbox{frac}}(n\sqrt{3})\mid n\in\mathbb{Z}^+\}$ is dense in [0, 1], which is sufficient here. This was http://www.mathhelpforum.com/math-help/f9/dynamics-circle-194819.html#post705803 in the old forum, but there are some differences, at least at first glance. Namely, \[\begin{aligned}0&=\inf\{m\cdot1+n\cdot \sqrt{3}\mid m\cdot1+n\cdot\sqrt{3}>0\mbox{ and }m,n\in\mathbb{Z}\}\\&\le\inf\{m\cdot1+n\cdot\sqrt{3}\mid m\cdot1+n\cdot \sqrt{3}>0,m\in\mathbb{Z}^-,n\in\mathbb{Z}^+\}\\& = \inf\{\mathop{\mbox{frac}} (n\sqrt{3}) \mid n\in\mathbb{Z}^+\}\end{aligned}\] but I am not sure about equality right away.
 
Plato said:
That is not a complete question.

I am so sorry. The infimum is to be proved to be equal to zero as Makarov has pointed out.

---------- Post added at 11:16 PM ---------- Previous post was at 11:05 PM ----------

Actually this can be done using Pigeon hole principle.

Denote $x_n=\text{frac}(n \sqrt{3})$

Let $n \in \mathbb{Z}^{+}$. Partition the interval $(0,1)$ into $n$ parts, viz, $(0,\frac{1}{n}),(\frac{1}{n},\frac{2}{n}), \ldots, (\frac{n-1}{n},1)$

Consider $n+1$ numbers, $x_1, x_2, \ldots, x_{n+1}$.

By PHP there exist $i,j, i \neq j$ such that $x_i,x_j \in (\frac{k}{n},\frac{k+1}{n})$.

This implies $\text{frac}(|i-j|\sqrt{3}) < \frac{1}{n}$. Thus $x_{|i-j|} < \frac{1}{n}$.

I can't think of any other proof.
 
caffeinemachine said:
By PHP there exist $i,j, i \neq j$ such that $x_i,x_j \in (\frac{k}{n},\frac{k+1}{n})$.

This implies $\text{frac}(|i-j|\sqrt{3}) < \frac{1}{n}$.
Not necessarily. Suppose $j > i$, but $\mathop{\mbox{frac}}(j\sqrt{3}) < \mathop{\mbox{frac}}(i\sqrt{3})$. Then \[\mathop{\mbox{frac}}(j\sqrt{3}-i\sqrt{3})=\mathop{\mbox{frac}}(j\sqrt{3})-\mathop{\mbox{frac}}(i\sqrt{3})+1\] so \[1-1/n<\mathop{\mbox{frac}}((j-i)\sqrt{3})<1\]

For example, $\sqrt{3}\approx1.73$ and $2\sqrt{3}\approx 3.46$, so $0 < \mathop{\mbox{frac}}(\sqrt{3})-\mathop{\mbox{frac}}(2\sqrt{3}) < 0.3$. However, $1-0.3<\mathop{\mbox{frac}}(2\sqrt{3}-\sqrt{3})<1$.
 
Evgeny.Makarov said:
Not necessarily. Suppose $j > i$, but $\mathop{\mbox{frac}}(j\sqrt{3}) < \mathop{\mbox{frac}}(i\sqrt{3})$. Then \[\mathop{\mbox{frac}}(j\sqrt{3}-i\sqrt{3})=\mathop{\mbox{frac}}(j\sqrt{3})-\mathop{\mbox{frac}}(i\sqrt{3})+1\] so \[1-1/n<\mathop{\mbox{frac}}((j-i)\sqrt{3})<1\]

For example, $\sqrt{3}\approx1.73$ and $2\sqrt{3}\approx 3.46$, so $0 < \mathop{\mbox{frac}}(\sqrt{3})-\mathop{\mbox{frac}}(2\sqrt{3}) < 0.3$. However, $1-0.3<\mathop{\mbox{frac}}(2\sqrt{3}-\sqrt{3})<1$.
Ah! yes. My mistake.
The proof changes a little bit.
We have $x_i, x_j \in (\frac{k}{n},\frac{k+1}{n})$
suppose $x_i>x_j$ then $frac((i-j) \sqrt{3}) < \frac{1}{n} \Rightarrow x_{i-j} < \frac{1}{n}$ EDIT: what if $i<j$ :O
Similarly for $x_j > x_i$
$x_i=x_j$ is not a possibility.
 
Well after more fiddling around I can only prove that $inf \{ n \sqrt{3} : n \in \mathbb{Z} \} =0$. I am not able to show that $inf \{ n \sqrt{3} : n \in \mathbb{Z}^+ \} =0$.
I don't even know whether the latter is true.
 
Just as a little $\LaTeX$ hint: use the backslash for infinum instead of just inf. You get '$\inf$' versus '$inf$'.
 
caffeinemachine said:
I can only prove that $\inf \{ n \sqrt{3} : n \in \mathbb{Z} \} =0$.
Well, this is not true, of course, because this set is not bounded from below... I'll return to this question when I have more time.
 
  • #10
Evgeny.Makarov said:
Well, this is not true, of course, because this set is not bounded from below... I'll return to this question when I have more time.
That was a typo again... sorry:
What I meant was $\inf \{ \text{frac}(n \sqrt{3}): n \in \mathbb{Z} \}=0$
 
  • #11
caffeinemachine said:
Prove that $\inf \{ \text{ frac}(n \sqrt{3}) : n \in \mathbb{Z}^+ \} = 0$ where $\text{frac}(x)$ is the fractional part of $x$
One very constructive way to approach this is to use the continued fraction approximants to $\sqrt3$. These are alternately slightly less than, and slightly greater than, $\sqrt3.$ Take one of the approximants that is less than $\sqrt3$, for example 265/153. You can then check that $153\sqrt3\approx 265.00377...$, whose fractional part is very small.
 
  • #12
Opalg said:
One very constructive way to approach this is to use the continued fraction approximants to $\sqrt3$. These are alternately slightly less than, and slightly greater than, $\sqrt3.$ Take one of the approximants that is less than $\sqrt3$, for example 265/153. You can then check that $153\sqrt3\approx 265.00377...$, whose fractional part is very small.
Thank you Opalg but I am ignorant about continued fractions. Anyways.. thanks for the help.
 
Back
Top