2

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

How can I prove the above statement? Because Euler’s $φ$-function is bijection?

Cameron Buie
  • 105,149

1 Answers1

5

Let $N$ be a positive integer, $p$ be the least prime greater than $N+1$, and $n$ be any integer such that $\varphi(n) = N$. We now show that there are finitely many such $n$:

If $q \geq p$ is a prime divisor of $n$, then $n = q^{k}m$ for some $k, m \geq 1$ and $gcd(q^{k}, m) = 1$. Furthermore, $\varphi (n) = \varphi (q^{k}) \varphi (m) \geq q-1$ since $\varphi (m) \geq 1$ and $\varphi (q^{k}) = q^{k-1}(q - 1) \geq q - 1$. This leads us to a contradiction; for $q-1 \geq p-1 \geq N$. Thus, no prime divisor of $n$ is greater than $N+1$. In particular, the distinct prime divisors of $n$ belong to a finite set, say $\{p_1, p_2, .., p_m\}$.

Now, consider the prime decomposition of $n$, $p_1^{a_1}p_2^{a_2}...p_n^{a_n}$, then $\varphi(n) = \varphi(p_1^{a_1})\varphi(p_2^{a_2})...\varphi(p_m^{a_m})$. Thus, we have $\varphi(n) = \prod_{i = 1}^m p_i^{a_i-1}(p_i - 1)$. Note that, if $p^k$ is a factor of $n$, then $p^{k-1}$ must be a factor of $N$, which ultimately limits the possible values for $k$. Therefore, $n$ must be a product of a bounded set of primes, each raised to a bounded exponent; there are finitely many such $n$.

Finally, for each positive integer $N$, there is a largest integer $n$ with $\varphi(n) = N$. Thus, as $n$ approaches infinity, so does $\varphi(n)$.

Igor Rivin
  • 26,372
  • I hope you'll find time to enlarge upon this link-only answer with a fuller explanation. – hardmath Jan 14 '14 at 18:16
  • @hardmath why? The link contains a complete argument. – Igor Rivin Jan 14 '14 at 18:17
  • @vadim123 Why? It answers the question. – Igor Rivin Jan 14 '14 at 18:17
  • 4
    This link might be gone in a week but the question may last forever. Link-only answers are discouraged on Math.SE. If you don't want to summarize the link's solution, then you should leave the link as a comment. – vadim123 Jan 14 '14 at 18:18
  • For fillers, you can use {}{}{}...{}{} where the dots just indicate you place enough of them to fill the space needed. – Pedro Jan 14 '14 at 18:22
  • @PedroTamaroff Thanks! I personally think the whole character limitation stupid, so this was partly my way to make my opinion known. – Igor Rivin Jan 14 '14 at 18:23
  • @vadim123 Do you think posterity will thank us now? – Igor Rivin Jan 14 '14 at 18:24
  • @IgorRivin: The Help Center -> Answering sets out guidelines, in particular to provide context for links. In case the link goes bad, the most important idea from it should be quoted. – hardmath Jan 14 '14 at 18:30
  • Vadim's right: if the link disappears (as it's happened some 48285048345549386759321 times in the past), the answer becomes completely useless. It should bring the actual stuff, jsut as it does now. – DonAntonio Jan 14 '14 at 18:30
  • 1
    @IgorRivin, on behalf of posterity I wish to give you my warmest upvote. – vadim123 Jan 14 '14 at 18:31
  • @vadim123 I thank you and posterity (on second thought, what has posterity ever done for me? I am thanking just you). – Igor Rivin Jan 14 '14 at 18:32