1

Rotman wants me to prove the above in a very specific way:

Step 1: Prove that if p is a prime with $p\equiv 3$ mod 4, then either $a^2 \equiv 2$ mod p is solvable or $a^2 \equiv -2$ mod p is solvable. The hint says to note that the multiplicative group $F_p^\times$ is isomorphic to a group of order 2 times a group of odd order m, and the group of order m must contain 2 or -2 because $F_2 \times Z_m = (\{1\}\times H)\cup(\{-1\}\times H)$. Finally, note that every element in a group of odd order has a square root.

Pretty much all of that makes sense to me, except this: I believe we have to show the isomorphic copy of either 2 or -2 has to belong to the left part of the union, because in that case the isomorphic copy of the square root is just (1, square root of some element of H). But how do you do this? What prevents both 2 and -2 from mapping to the right part of the union?

Step 2: Prove that $x^4+1$ factors in $F_p[x]$, where p is an odd prime, if any of the following congruences are solvable:

  • $b^2\equiv -1$ mod p
  • $a^2 \equiv 2$ mod p
  • $a^2 \equiv -2$ mod p

This was easy. I don't need help on step 2.

Step 3: Show $x^4 + 1$ factors in $F_p[x]$ for all primes $p$

It's obviously true for p=2. I also see why it's true for $p\equiv 3$ mod 4. But I don't see why it's true for $p\equiv 1$ mod 4. Maybe it's true that $p\equiv 1$ mod 4 implies $b^2\equiv -1$ mod p is solvable, which would finish the problem, but I don't know how to show that.

Patrick M
  • 23
  • 4
  • 1
    Welcome to the site! This particular problem was discussed in this thread a decade ago. Hagen von Eitzen's solution is probably closer to what Step 2 is suggesting. My solutions use more facts from the finite field side and/or what the law of quadratic reciprocity says about the residue class of $2$. I don't have a copy of Rotman, so I cannot check what has been covered up to this point. – Jyrki Lahtonen Apr 29 '21 at 17:40
  • Sorry about not reading your question to the end. The existence of a modular square root of $-1$ is another common question. See for example 1, but probably there are better ones. If you want to do it yourself, one way is to show that when $p=2m+1$, $2\mid m$, then the residue class of the factorial $m!$ is a square root of $-1$. If you have seen Wilson's theorem, I'm sure you can do it. – Jyrki Lahtonen Apr 29 '21 at 17:50
  • You seem to know some TeX, so may also want to familiarize yourself with ApproachZero. Neither StackExchange search nor Google can grok TeX, but ApproachZero can handle it. It has only indexed this site and AOPS, but it is a useful tool! – Jyrki Lahtonen Apr 29 '21 at 17:52
  • 1
    Anyway, I upvoted your post for you gave a good description of your background. – Jyrki Lahtonen Apr 29 '21 at 17:52
  • An experienced user linked several solutions to this thread. – Jyrki Lahtonen Apr 29 '21 at 17:58

0 Answers0