I was thinking about how to show whether or not a polynomial irreducible over $\mathbb Q$ could be irreducible over $\mathbb Q(i)$

I am aware of $x^4 + 1$ being irreducible over $\mathbb Q$ but not $\mathbb Q(i)$. However, it also has strictly complex roots. I was wondering if whenever there are only two non real roots and at least one real root, that the polynomial is irreducible over both $\mathbb Q$ and $\mathbb Q(i)$.

Are there any more examples of polynomials irreducible over $\mathbb Q$ but not $\mathbb Q(i)$ that have only two complex roots and at least one real root? I've realised that my original reasoning is actually flawed but I still suspect that maybe there isn't such a polynomial.

Also, a related question:

I am aware of many tests for irreducibility over $\mathbb Q$, but what I can I do other than check the roots, to see that a polynomial is or isn't irreducible over $\mathbb Q(i).$ I am aware that Eisenstein's criterion may be generalised to integral domains, but what if the constant of the polynomial is $2$? In that case Eisenstein wouldn't apply, so is there anything else that I could do?

  • $\begingroup$ $x^2+1$ is a slightly easier example than yours, and the first I thought of before reading your entire question. Still has the same problems, though. $\endgroup$
    – Arthur
    Jan 2, 2018 at 13:13
  • $\begingroup$ What do you make of $x^4-3$ and $x^4-5$? $\endgroup$ Jan 2, 2018 at 13:19
  • $\begingroup$ @MarkBennet: Why, yes! The day after New Year's Day can be hard sometimes… $\endgroup$
    – Bernard
    Jan 2, 2018 at 13:29
  • $\begingroup$ Newton's polygon gives an occasionally useful extension of Eisenstein's criterion. The condition is that if the polygon of a degree $n$ polynomial consists of a single line segment with slope $-a/n,\gcd(a,n)=1$, then the polynomial is irreducible. Consider, for example, $p(x)=x^5-4x+2$ over $\Bbb{Q}(i)$ and the valuation related to the prime $p=1+i$. The constant term is divisible by $p^2$, the linear with $p^4$, and the leading with $p^0$. The polygon thus reduces to a line with slope $-2/5$ and we are done. $\endgroup$ Jan 2, 2018 at 15:53
  • $\begingroup$ (cont'd) Compare with this segment from WP on Eisenstein. $\endgroup$ Jan 2, 2018 at 15:55

1 Answer 1


Let $f$ be irreducible over $\Bbb Q$ but not over $\Bbb Q(i)$. Then over $\Bbb Q(i)$ we have the factorisation $f=g\overline g$ where the coefficients of $\overline g$ are the complex conjugates of those of $g$. The zeroes of $\overline g$ the complex conjugates of those of $g$. A real zero of $f$ is then both a zero of $g$ and of $\overline g$ and so a double zero of $f$, which we'd assumed irreducible...

  • $\begingroup$ Does this imply that $f$ is of even degree, since $g, \bar g$ would have the same degree? I'm also not really sure what your response implies. What is the importance of this double root? $\endgroup$
    – user366818
    Jan 2, 2018 at 17:51
  • 1
    $\begingroup$ @user366818 If $f$ is irreducible over $\Bbb Q$ but not over $\Bbb Q(i)$ that it has even degree. A polynomial over $\Bbb Q$ with a multiple root cannot be irreducible. $\endgroup$ Jan 3, 2018 at 8:41

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.