10

Prove that two distinct numbers of the form $a^{2^{n}} + 1$ and $a^{2^{m}} + 1$ are relatively prime if $a$ is even and have $\gcd=2$ if $a$ is odd.

My attempt:
If $a$ is even, let $a = 2^{s}k$ for some integers $k, s$
Then, $$a^{2^{n}} + 1 = 2^{2^{n}s}\cdot k^{2^n} + 1$$ and $$a^{2^{m}} + 1 = 2^{2^{m}s}\cdot k^{2^m} + 1$$ To prove that they're relatively prime, we need to show that their gcd = 1. And I was stuck here, how could I prove that gcd of two numbers is $1$?

A hint would be sufficient. Thanks.

user26857
  • 52,094
roxrook
  • 12,081

3 Answers3

10

If $x = -1 \mod p$, then $x^{2^n} = 1 \mod p$.

Assume $n \gt m$. So if $p$ divides $x + 1 = a^{2^m} + 1$, then, $a^{2^n} + 1 = x^{2^{n-m}} + 1 = 1 + 1 = 2 \mod p$.

Aryabhata
  • 82,206
  • @Moron: What's $p$ in your hint? I tried understand your hint, but I felt lost :(. Thanks. – roxrook Feb 28 '11 at 23:40
  • @Chan: $p$ is any prime which divides $x+1$. Basically we are showing that if $p$ divides $a^{2^m} + 1$, then $p$ cannot possibly also divide $a^{2^n} + 1$, unless $p=2$. – Aryabhata Mar 01 '11 at 00:09
  • 1
    @Moron: Thanks again. But I lost, I knew that $p$ is prime, but I can't find a way to link it to my problem. Furthermore, how does it relates to the parity of $a$? – roxrook Mar 01 '11 at 04:55
  • @Chan: If two number are not relatively prime, there is some prime $p$ which divides both. We showed that that is not possible for $p \gt 2$. If a is odd, then all numbers are divisible by 2. – Aryabhata Mar 01 '11 at 05:40
  • @Moron: Thank you. I will think more about it. – roxrook Mar 01 '11 at 05:51
9

Hint:

Consider the following proof when $a=2$: https://planetmath.org/fermatnumbersarecoprime

Try adapting it to work for all $a$.

Hint 2: Factor $a^{2^n}-1$.

user26857
  • 52,094
Eric Naslund
  • 72,099
  • @Eric Naslund: I tried to adapt the proof you mentioned, but I could not prove that the gcd must be 2, since $a$ has another variables :(. Could you help? – roxrook Feb 28 '11 at 23:50
  • 1
    @Chan: Ok, so we need to do a little more, you are correct. It suffices to consider $$a^{2^n}+1$$ modulo $4$. Since $$a^{2^n}$$ is a square, it is congruent to 0 or 1, and hence is congruent to 1 when $a$ is odd. Then we see $$a^{2^n}+1\equiv 1\pmod{4}$$ so that each term is divisible by 2 but not 4. That is where the $2$ in the $\gcd$ comes from. Use the previous method to show that other primes cannot divide it. – Eric Naslund Mar 01 '11 at 06:17
  • 2
    @EricNaslund,@Chan: What about the case of a is even? – Alexy Vincenzo Jun 29 '13 at 19:11
2

Proof $\ (A\!+\!1, A^{2K}\!\!+1)\,=\, (A\!+\!1,\ (-1)^{2K}\!+1) = (A\!+\!1,2),\, $ [OP is $ A = a^{\large 2^{\Large N}}\!\!,\,$ wlog $N\! <\! M$]

using $\,\ \ (A\!-\!n,\ \ F(A)\ )\, = \, (A\!-\!n,\ \ F(n) )\ $ for all polynomials $\ F(X)\in \mathbb Z[X],\ A,n\in \mathbb Z$

by $\!\bmod A\!-\!n\!:\ A\equiv n\,\Rightarrow\, F(A)\equiv F(n)\ $ by the Polynomial Congruence Rule, and

also by $\ (B,\ C)\ =\ (B,\ C \bmod B) = $ modular property of GCD (used in Euclidean algorithm)

Remark $\, $ For the general case $\ (a^m+1,a^n+1)\ $ see this answer.

Bill Dubuque
  • 272,048
  • 1
    Someone is downvoting all the answers on this thread. +1. This is a more generally applicable answer. – Aryabhata Feb 28 '11 at 05:18
  • 3
    At this point, I think you under evaluate a very important part of mathematics. That is the ability to communicate it to other people. Every answer you down voted was one where I was much clearer to someone wanting to understand the basics. Each answer you write isn't even written in English nicely, is unclear, and not at the level the OP wanted. Having said that, do I think they are bad answers that deserve to be down voted? No, that I reserve for actual things that do not contribute at all mathematically. – Eric Naslund Feb 28 '11 at 16:20
  • 5
    At the very least, if you do not like my answers, and think they need to be pushed in a certain direction, please comment and tell me. Whenever you down vote a message box comes up asking you to do so. It is impossible for me understand your reasoning when you do not tell me. I do not see at all what was wrong with most of these answers, and I honestly have no problem with downvotes that are explained, because then whatever problem there is can be improved. As is it resembles that you just do not like the fact that mine are accepted over yours, which is childish. – Eric Naslund Feb 28 '11 at 16:28
  • @readers Eric's answer now consists of a dead link. Speculation about downvotes is usually wrong, almost always nonconstructive, and always off-topic. Worth emphasis: this answer is the only answer with a complete, rigorous proof (and even a generalization) - something we should strive for in answers. Yet someone downvoted it (presumably explained by the above strange comments). Alas, gamification is the root of all evil. – Bill Dubuque Apr 28 '20 at 15:55