3
$\begingroup$

We know for field $\mathbb R$, ${\mathbb R}^{*p}={\mathbb R}^{*}$, for any prime $p\neq 2$ and ${\mathbb R}^{*2}\neq {\mathbb R}^{*}$. (${\mathbb R}^{*}$ is the multiplicative group of $\mathbb R$.) Can we find a subfield $F$ of $\mathbb C$ such that ${ F}^{*p}={ F}^{*}$ for any prime $p\neq 5$ and ${F}^{*5}\neq F^{*}$?

1 Answers 1

1

Here's a possible construction.

If $F$ is a countable subfield of $\mathbb{C}$, we can form the field $F'$ by splitting all polynomials of the form $X^p-a$ for $p\neq 5$ a prime and $a\in F$. There are countably many such polynomials, so order them in some way and split them successively.

Now, iterate this procedure. Starting with $F_0=\mathbb{Q}$, define $F_{n+1}=F_n'$ and $F_\infty = \cup F_n$. Then $F_\infty$ is a subfield of $\mathbb{C}$, and any of its elements contains all of its $p$th roots for $p\neq 5$ a prime. (This is because any such element belongs to some $F_n$, so all its roots are already in $F_{n+1}$.)

I claim that no fifth root of $2$ is in $F_\infty$. (Thanks to Yuval Dor for this argument). In fact, let's even extend $F_\infty$ a little bit: in the first step we constructed $F_1$ by adding, among other things, all roots of unity except those of order 5. Let's throw in the roots of $X^5-1$ as well, so now $F_1$ actually contains all roots of unity of prime order.

No fifth root of $2$ can belong in this extended $F_1$. If any root did, then all of them would, since $F_1$ contains the 5th roots of unity. But the splitting field of $X^5-2$ is a non-abelian extension of $\mathbb{Q}$, so it cannot be contained in a cyclotomic field.

Finally, by Kummer theory, splitting polynomials of the form $X^p-a$ over a field that has the necessary roots of unity can only result in a trivial extension of an extension of order $p$. So the order $5$ can never show up, meaning $\sqrt[5]{2}$ can't appear.

  • 1
    If you start with the field generated by all roots of unity of order coprime with 5, then it is easy to see what the prime divisors of all finite subextensions of your field are, I guess.2017-01-08
  • 0
    I can't prove that $\sqrt[5]{2}\notin F_\infty$.2017-01-08
  • 0
    Notice in zero characteristic, a divisible field contains any root of unity2017-01-08
  • 0
    @Reza I added a proof to my answer.2017-01-10
  • 0
    Dear prof. Alon Amit. I friendly thanks for your response. In final paragraph you claimed that for any two field $K,F$ such that $F_1\subset F \subset K\subset F_\infty$, $[K:F]\neq 5$, why? it is not clear for me.2017-01-10
  • 2
    Hmm. You lost me also. $F_1$ is certainly not a union of cyclotomic fields, because it contains, among other things, the splitting field of $x^{11}-p$ for any prime $p$ (with a non-abelian Galois group). And that splitting field also contains subfields of degree five because $5\mid p-1$. I still think that the construction works and the argument can be saved, but I don't see it right away. May be show that no subfield of $F_1$ can have Galois group $C_5\rtimes C_4$ (= the Galois group of $x^5-2$)?2017-01-24