0
$\begingroup$

How to find a number $n$ such that $$\frac{n}{\phi(n)} > 10,$$ where $\phi(n)$ denotes the Euler's phi function?

I was trying to find the smallest one, so was keeping each prime once. I tried with the number $$n = 2\times3\times5\times7\times11\times13\times17\times19$$ and some more numbers, but not working. Need some help!

Note that $$\phi(n) = n \times \prod_{p} \left(1-\frac1p \right) = n \times \prod_{p} \left(\frac{p-1}{p} \right)$$ hence $$\frac{n}{\phi(n)} = \prod_{p} \left(\frac{p}{p-1} \right).$$

$\endgroup$
13
  • 6
    $\begingroup$ Just keep on multiplying with the next prime. It will take some time until you beat $10$, but eventually you will beat every given limit this way, and then you have also found the smallest number breaking the limit. $\endgroup$
    – Peter
    Commented Oct 2, 2020 at 14:32
  • 1
    $\begingroup$ This is indeed the right approach, but it will take some work. If my computation is right, you have to take the product of primes all the way up to 257 for the quotient to exceed 10. $\endgroup$
    – Wojowu
    Commented Oct 2, 2020 at 14:33
  • 1
    $\begingroup$ @user8795 I think this can only be determined (quickly) with a computer. Use one of the standard number theory tools like pari/gp or python. $\endgroup$
    – Peter
    Commented Oct 2, 2020 at 14:37
  • 2
    $\begingroup$ The $55^{th}$ prime is indeed $257$. $\endgroup$
    – player3236
    Commented Oct 2, 2020 at 14:39
  • 1
    $\begingroup$ Look at A091439 for more info (not that there is much). This confirms what others have said about using the $55$th primorial. $\endgroup$ Commented Oct 2, 2020 at 14:39

3 Answers 3

2
$\begingroup$

The solution in the comments, done by several people, is saying that, with $f(n)=\frac{n}{\phi(n)}$ we have $$ f(p_{55}\#)=10.003719732091010383325131639818913973 $$ where $$ p_{55}\#=16516447045902521732188973253623425320896207954043566485360902980990824644545340710198976591011245999110 $$ and that this is the smallest such integer with $f(n)>10$.

More information is found in the following post, by considering $1/f(n)$:

$\frac{\phi(m)}{m}$ is dense in $[0,1]$

$\endgroup$
1
  • $\begingroup$ Note that it is nontrivial to prove that the "champion numbers" for $n/\phi(n).$ I think so; in any case, the champions for $\phi(n)$ are the primorials. math.univ-lyon1.fr/~nicolas/petitsphiDPP.pdf $\endgroup$
    – Will Jagy
    Commented Oct 3, 2020 at 0:29
0
$\begingroup$

A more "manual" approach, considering $$\log{(1+x)}\geq \frac{x}{1+x}, \forall x>-1$$ We can build an estimation as per the following: $$\frac{n}{\varphi(n)}= \prod\limits_{p}\frac{p}{p-1}= \prod\limits_{p}\left(1+\frac{1}{p-1}\right)\geq \prod\limits_{p} e^{\frac{\frac{1}{p-1}}{1+\frac{1}{p-1}}}=\\ \prod\limits_{p} e^{\frac{1}{p}}=e^{\sum\limits_{p}\frac{1}{p}}$$ or, we are forcing for $$\sum\limits_{p}\frac{1}{p} \geq \log{10}$$ But $$\sum\limits_{p\leq x}\frac{1}{p}\geq \log{\log{(x+1)}}-\log{\frac{\pi^2}{6}}$$ then forcing further for $$\log{\log{(x+1)}} \geq\log{10}+ \log{\frac{\pi^2}{6}} \iff\\ \log{(x+1)}\geq 10\cdot\frac{\pi^2}{6} \iff \\ x \geq e^{10\cdot\frac{\pi^2}{6}} -1=13927008.866545...$$ reveals we have to consider all the primes between $1$ and $13927009$.


The crude estimates above can be fine tuned further, given $$\lim _{x\to \infty }\left(\sum _{p\leq x}{\frac {1}{p}}-\log \log x\right)=M\approx 0.261497...$$ for sufficiently large $x$ $$\sum _{p\leq x}{\frac {1}{p}} > \log \log x + \frac{M}{2}$$ then $$x\geq e^{e^{\log{10} -\frac{M}{2}}}\approx 6466.46...$$

$\endgroup$
0
$\begingroup$

The other answer covers the $x=10$ case, but I'll show that if you want $\frac n{\varphi(n)}>x$ for any real number $x$ that is always possible.

I'm going to use the notation

$$f(x)\sim g(x)\iff \lim_{x\to\infty}\frac{f(x)}{g(x)}=1$$

It is unfortunately not standard. I lifted it from "An Introduction to the Theory of Numbers" by Hardy and Wright.

Mertens' third theorem tells us that

$$\prod_{p\leq a}1-\frac 1 p \sim \frac{e^{-\gamma}}{\ln(a)}$$

For a primorial $n$ we then have

$$\frac n{\varphi(n)}\sim\ln(a)e^\gamma$$

Since we want the quantity on the LHS to be greater than $x$ we need to $a$ to be approximately

$$a \sim\exp\left(e^{-\gamma}x\right)$$

This increases exponentially in $x$ which explains why the answer to your question is so big. For $x=10$ that gets you $a\approx 274$, close to the true value of 257.

$\endgroup$

You must log in to answer this question.

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