I'm currently solving a fairly long exercise related to Galois theory in which I've come across having to prove that $\sqrt{3+\sqrt{7}} \not\in \mathbb{Q}(\sqrt{3-\sqrt{7}})$ and $\sqrt{3-\sqrt{7}} \not\in \mathbb{Q}(\sqrt{3+\sqrt{7}})$. So far I haven't been able to find an "easy" or simple and understandable way to do so given that this isn't the main part of the problem.

Any help is appreciated!

  • $\begingroup$ $\sqrt{3+\sqrt{7}} = a+b\sqrt{3-\sqrt{7}} \implies 3+\sqrt{7} = a^2+2ab\sqrt{3-\sqrt{7}} + b(3-\sqrt{7})$ $ \implies c+d\sqrt{7} = e\sqrt{3-\sqrt{7}} \implies c^2+2cd\sqrt{7}+7d^2 = e^2(3-\sqrt{7}) \implies 2cd = -e^2$. But $-e^2 = -4a^2b^2$ and $2cd = 2(3-a^2-3b)(b+1)$, so we must have $2a^2b^2 = (a^2+3b-3)(b+1)$. Now solve for $a^2$ and hopefully this will give a contradiction. $\endgroup$ – mathworker21 Jan 8 at 15:39
  • 3
    $\begingroup$ @mathworker21 What is $a$ and $b$? $\{1,\sqrt{3-\sqrt{7}}\}$ is not a basis for $\mathbb{Q}(\sqrt{3-\sqrt{7}})$ over $\mathbb{Q}$ $\endgroup$ – mouthetics Jan 8 at 15:40
  • $\begingroup$ @mouthetics good point. imma bit rusty on my algebra. (they were supposed to be rationals) $\endgroup$ – mathworker21 Jan 8 at 15:41

Let $x_{\pm}=\sqrt{3 \pm \sqrt{7}}$.

It is easy to see that we have the following quadratic extensions:

$\mathbb{Q} \subset \mathbb{Q}(\sqrt{7}) \subset \mathbb{Q}(x_+)$,

$\mathbb{Q} \subset \mathbb{Q}(\sqrt{7}) \subset \mathbb{Q}(x_-)$.

Assume that $x_+ \in K=\mathbb{Q}(x_-)$. Then $\sqrt{2} = x_+x_- \in K$, thus $L=\mathbb{Q}(\sqrt{2},\sqrt{7}) \subset K$.

Since these fields have the same degree over $\mathbb{Q}$, $K \subset L$, ie $x_+=a+b\sqrt{2}+c\sqrt{7}+d\sqrt{14}$ for rationals $a,b,c,d$.

Taking squares, we get $3+\sqrt{7}=(a^2+2b^2+7c^2+14d^2) + (2ab+14cd)\sqrt{2} + (2da+2bc)\sqrt{14} + (2ca+4bd)\sqrt{7}$.

Thus $ab=-7cd$, $ad=-bc$, $2ca+4bd=1$, $a^2+2b^2+7c^2+14d^2=3$.

Assume $a=0$: then $d \neq 0$ thus $c=0$ and $bd=1/4$, $2b^2+14d^2=3$. Usual quadratic theory yields then a contradiction.

Thus $b=-7cd/a$, and $ad=7c^2d/a$ thus $a^2=7c^2$ hence $a=0$. A contradiction, hence the result.


In general, in this kind of problem, it is better not to mix the two operations + and $\times$. Let me give an illustration here, using only $\times$. Introduce the quadratic field $k=\mathbf Q(\sqrt 7)$. Adopting the notation $x_{\pm}=\sqrt {3 \pm \sqrt 7}$ suggested by @Mindlack, let us write $K_{\pm}=k(x_{\pm})$. These are two extensions of $k$ of degree at most $2$ :

  • if $K_{+}$ or $K_{-} =k$, i.e. $(3 \pm\sqrt 7)\in {k^*}^2$, norming down to $\mathbf Q$ shows that $N(3\pm\sqrt 7)=2$ is a square in $\mathbf Q^*$: impossible

  • if both degrees are 2, $K_{\pm}\subset K_{\mp}$ iff $K_{\pm}= K_{\mp}$, iff $k(x_{+})= k(x_{-})$, iff $2=(3+\sqrt 7)(3-\sqrt 7)\in {k^*}^2$ (no specific calculation, this is rudimentary Kummer theory over $k$), iff $\mathbf Q(\sqrt 2)=\mathbf Q(\sqrt 7)$, iff $2.7$ is a square in $\mathbf Q^*$(again by Kummer): impossible because $\mathbf Z$ is a UFD ./.


HINT: Show that both have minimal polynomial $f:=X^4-6X^2+2$ over $\Bbb{Q}$, and hence that $[\Bbb{Q}(\sqrt{3\pm\sqrt{7}}):\Bbb{Q}]=4$, but that the splitting field of $f$ over $\Bbb{Q}$ has degree greater than $4$.

  • $\begingroup$ My goal is to justify that the splitting field is $\mathbb{Q}(\sqrt{3+\sqrt{7}},\sqrt{3-\sqrt{7}})$, can I see that its degree is grater than 4 without having to prove what I was asking? $\endgroup$ – BBC3 Jan 8 at 16:27
  • $\begingroup$ How to prove that the splitting field of $f$ over $\mathbb{Q}$ has degree greater than $4$? Any other method? $\endgroup$ – mouthetics Jan 8 at 16:28
  • $\begingroup$ Do you mean $X^4 - 6X^2 + 2$? $\endgroup$ – Connor Harris Jan 8 at 17:08
  • $\begingroup$ @ConnorHarris Indeed I do, edited. $\endgroup$ – Servaes Jan 8 at 22:46
  • 1
    $\begingroup$ @mouthetics As the question is tagged Galois theory, determining the order of the Galois group was the approach I had in mind. $\endgroup$ – Servaes Jan 8 at 22:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.