27
$\begingroup$

Assume that $p(x)\in \Bbb{Q}[x]$ is irreducible of degree $n\ge3$.

Is it possible that $p(x)$ has three distinct zeros $\alpha_1,\alpha_2,\alpha_3$ such that $\alpha_1-\alpha_2=\alpha_2-\alpha_3$?

As also observed by Dietrich Burde a cubic won't work here, so we need $\deg p(x)\ge4$. The argument goes as follows. If $p(x)=x^3+c_2x^2+c_1x+c_0$, then $-c_2=\alpha_1+\alpha_2+\alpha_3=3\alpha_2$ implying that $\alpha_2$ would be rational and contradicting the irreducibility of $p(x)$.

This came up when I was pondering this question. There the focus was in minimizing the extension degree $[\Bbb{Q}(\alpha_1-\alpha_2):\Bbb{Q}]$. I had the idea that I want to find a case, where $\alpha_1-\alpha_2$ is fixed by a large number of elements of the Galois group $G=\operatorname{Gal}(L/\Bbb{Q})$, $L\subseteq\Bbb{C}$ the splitting field of $p(x)$. One way of enabling that would be to have a lot of repetitions among the differences $\alpha_i-\alpha_j$ of the roots $\alpha_1,\ldots,\alpha_n\in\Bbb{C}$ of $p(x)$. For the purposes of that question it turned out to be sufficient to be able to pair up the zeros of $p(x)$ in such a way that the same difference is repeated for each pair (see my answer).

But can we build "chains of zeros" with constant interval, i.e. arithmetic progressions of zeros.

Variants:

  • If it is possible for three zeros, what about longer arithmetic progressions?
  • Does the scene change, if we replace $\Bbb{Q}$ with another field $K$ of characteristic zero? (Artin-Schreier polynomials show that the assumption about the characteristic is relevant.)
$\endgroup$
  • 1
    $\begingroup$ I do not have time to think this through until tomorrow, but for your second variant, it pops to my mind that one should look at the quotient field of $\mathbb Q[x,y,z,w,u]/(x-y-u,y-z-u)$ and consider $(T-x)(T-y)(T-z)(T-w)$ over the smallest field, you can define this (i.e. the field generated by the elementary symmetric polynomials in x,y,z,w). $\endgroup$ – MooS Jul 1 '16 at 10:59
  • $\begingroup$ If this does not work, i.e. if this polynomial turns out to be reducible, I dont see that it could work with concrete numbers instead of indeterminants. Of course this does only cover the $n=4$-case. $\endgroup$ – MooS Jul 1 '16 at 11:01
  • 1
    $\begingroup$ On the other hand if this would work with indeterminates, couldn't we invoke Hilbert's irreducibility theorem to deduce that it is also possible over $\mathbb Q$? $\endgroup$ – MooS Jul 1 '16 at 11:12
  • $\begingroup$ A promising plan @MooS! Take your time :-) I will be gone for 30 hrs or so. $\endgroup$ – Jyrki Lahtonen Jul 1 '16 at 13:36
23
+100
$\begingroup$

The answer to your question is NO. Suppose by contradiction that such a polynomial $P$ exists, and denote by $S$ the set of roots. By hypothesis, some $a\in S$ can be written $a=\frac{b+c}{2}$ where $b,c$ are distinct elements of $S$. But since the Galois group acts transitively on $S$, this property holds for all $a\in S$. This motivates the following definition :

Definition. A (non-empty) set $S\subseteq {\mathbb C}$ is AP-extensive if any $a\in S$ can be written $a=\frac{b+c}{2}$ where $b,c$ are distinct elements of $S$.

Note that an AP-extensive $S\subseteq {\mathbb R}$ cannot have a largest element. In particular, any (non-empty) AP-extensive $S\subseteq {\mathbb R}$ is necessarily infinite. This still holds in $\mathbb C$ :

Lemma. If $S\subseteq {\mathbb C}$ is AP-extensive, then $S$ is infinite (or empty).

Proof of lemma. Suppose by contradiction that $S$ is finite and nonempty. Then the set $\lbrace b\in{\mathbb R}\ | \ \exists a, a+ib\in S\rbrace$ is finite also and therefore has a largest element $b_0$. Let $S_1=\lbrace z\in S \ | \ Im(z)=b_0 \rbrace$. It is easy to see that if $a=\frac{b+c}{2}$ with $a,b,c\in S$ and further $a\in S_1$, then $b$ and $c$ must be in $S_1$ also. So $S_1$ is AP-extensive as well. Next, let $S_2=S_1-ib_0$. Then $S_2$ is AP-extensive also, but by construction $S_2\subseteq {\mathbb R}$. So $S_2$ (and hence $S_1,S$ also) must be infinite which is impossible. This concludes the proof.

$\endgroup$
  • $\begingroup$ A beautiful argument :-) $\endgroup$ – Jyrki Lahtonen Jul 3 '16 at 9:20
  • 1
    $\begingroup$ I got a lot of inspiration from the Krein-Milman theorem. $\endgroup$ – Ewan Delanoy Jul 3 '16 at 9:29
  • $\begingroup$ Unless I misunderstood, the same argument works for polynomials over any characteristic zero field. The set of zeros spans (affinely or linearly) a finite dimensional space over $\Bbb{Q}$, and we can then work with one coordinate at the time much the same way. In other words, we don't need the set $S$ to be a subset of $\Bbb{C}$ in particular. $\endgroup$ – Jyrki Lahtonen Jul 3 '16 at 9:41
  • $\begingroup$ @JyrkiLahtonen Indeed. Conversely, there is a counterexample in characteristic $p$ for any $p>0$. $\endgroup$ – Ewan Delanoy Jul 3 '16 at 14:34
  • 1
    $\begingroup$ Wow! The key idea to use the transitive action to deduce that any root could play the role of $\alpha_2$ is so simple but wonderful! $\endgroup$ – MooS Jul 4 '16 at 14:13

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.