Been sitting with this question from for a whole bunch of hours now. I'm studying for an exam so I don't want to be stuck for too long, however much I dig endlessly pondering.

Let $f(t)$ be a separable polynomial in $\Bbb Q[t]$ with zeros $\alpha_i,i=1,\dots,4$. Let \begin{align*} \beta_1=\alpha_1\alpha_2+\alpha_3\alpha_4\\ \beta_2=\alpha_1\alpha_3+\alpha_2\alpha_4\\ \beta_3=\alpha_1\alpha_4+\alpha_2\alpha_3 \end{align*} and $g(t)=(t-\beta_1)(t-\beta_2)(t-\beta_3)$. Show that $g(t)\in \Bbb Q[t]$.

So, since the coefficients of $g$ are \begin{align*} \beta_1+\beta_2+\beta_3\\ \beta_1\beta_2+\beta_1\beta_3+\beta_2\beta_3\\ \beta_1\beta_2\beta_3 \end{align*} I just want to show that they're rational. A common approach is to try to express them in terms of \begin{align*} s_1&=\alpha_1+\alpha_2+\alpha_3+\alpha_4\\ s_2&=\alpha_1\alpha_2+\alpha_1\alpha_3+\alpha_1\alpha_4+\alpha_2\alpha_3+\alpha_2\alpha_4+\alpha_3\alpha_4\\ s_3&=\alpha_1\alpha_2\alpha_3+\alpha_1\alpha_2\alpha_4+\alpha_1\alpha_3\alpha_4+\alpha_2\alpha_3\alpha_4\\ s_4&=\alpha_1\alpha_2\alpha_3\alpha_4 \end{align*} which I know are rational 'cause they're the coefficients of $f$. But this is where I don't get any further. I know that if the coefficients of $g(t)$ are symmetric polynomials in the roots, then they can be expressed in terms of $s_1,s_2,s_3,s_4$ but f.ex. $\beta_1+\beta_2+\beta_3$ for sure ain't symmetric, since with $\begin{pmatrix}1 & 2 & 4 & 3\end{pmatrix}$ we get the term $\alpha_2\alpha_4$ which isn't a term of $\beta_1+\beta_2+\beta_3$.

What do you think? Am I not thinking enough?

  • 4
    $\begingroup$ Are you sure that your definition of $\beta_2$ is correct? I think it should be $\alpha_1\alpha_3+\alpha_2\alpha_4$. $\endgroup$
    – Alex Fok
    Dec 21, 2015 at 9:44
  • $\begingroup$ Haha, I was sure beginning to think there was a typo. I did the corresponding change of $\beta_1$ but that didn't yield much. I'll try yours too and report back. $\endgroup$ Dec 21, 2015 at 10:41
  • $\begingroup$ @AlexFok It worked, is there any other way of fixing the typo up? $\endgroup$ Dec 21, 2015 at 12:01
  • $\begingroup$ I made the correction for you. $\endgroup$
    – Lubin
    Dec 22, 2015 at 0:25
  • $\begingroup$ You don’t need to show that the $\beta_j$ are individually invariant under any permutation of the $\alpha_i$, but just that $\beta_1+\beta_2+\beta_3$ is, for instance. And the others you quote. $\endgroup$
    – Lubin
    Dec 22, 2015 at 0:27

1 Answer 1


You only want to show that the triple $\{\beta_1,\beta_2,\beta_3\}$ is invariant under all permutations of the roots $\{\alpha_i\}$, since the fixed field of $\Bbb Q(\alpha_1,\cdots,\alpha_4)$ under the permutation group $S_4$ is the field of the elementary symmetric functions in the $\{\alpha_i\}$, which are the coefficients of the polynomial, assumed to be in $\Bbb Q$.

So it only remains to show that every simple exchange of two of the $\alpha_i$ leaves the set $\{\beta_1,\beta_2,\beta_3\}$ unchanged. For instance, $\alpha_1\leftrightarrow\alpha_2$ leaves $\beta_1$ fixed, but interchanges $\beta_2$ and $\beta_3$. You’ll find the same result for any other simple exchange. So all is well.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .