3

I was skimming again through Dummit & Foote's Abstract Algebra and I came across this exercise:

Prove that for any given positive integer $N$ there exist only finitely many integers $n$ with $\varphi(n)=N$, where $\varphi$ denotes Euler $\varphi$-function. Conclude in particular that $\varphi(n)$ tends to infinity as $n$ tends to infinity.

I don't doubt that $\varphi(n) \to \infty$ as $n \to \infty$. However, if I recall Erdös proved that if there is an $x_0$ such that $\varphi(x_0)=N$, then there must be infinitely many integers such that $\varphi(x)=N$. See this article for a mention of this (it's in the very first $2$ lines).

Am I reading something wrong here? Has has Dummit and Foote actually asked the impossible? Or does the MathWorld site contain an error?

Martin Sleziak
  • 56,060
mathematics2x2life
  • 13,578
  • 2
    The link doesn't say quite what you think it does -- there are infinitely many integers with the same multiplicity, not infinitely many with the same solution to $\varphi (n) = N$. –  Nov 19 '13 at 01:48
  • @Marc & André: I knew it was something obvious but I stared at this for a good half hour not seeing it. I guess it's just one of those nights. Thank you both for pointing out what I should have seen! – mathematics2x2life Nov 19 '13 at 01:54
  • The phrasing could have been better, so don't feel bad. –  Nov 19 '13 at 01:58
  • Note that $$ \varphi(n) \geq \sqrt {\frac{n}{2}}, $$ for an easy example of a lower bound showing that $\varphi(n)$ tends to infinity. Also $$ \varphi(n) \geq ; 2 ; \left( \frac{n}{6} \right)^{2/3}, $$ – Will Jagy Nov 19 '13 at 02:41
  • FYI, I have given a proof here. – ccorn Dec 04 '13 at 14:21

1 Answers1

2

The result you refer to does not say what you think it does. What it says is that if there is an $m_0$ such that $\varphi(x)=m_0$ has exactly $k$ solutions, then there exist infinitely many $m$ such that $\varphi(x)=m$ has exactly $k$ solutions.

For example, let $k=3$. There is an $m_0$, namely $2$, such that $\varphi(x)=m_0$ has exactly $k$ solutions. These are $3$, $4$, and $6$. We can conclude that there are infinitely many $m$ such that $\varphi(x)=m$ has exactly $3$ solutions.

André Nicolas
  • 514,336