3

Is it true that for every natural number $n \geq 3$, there exists an irreducible polynomial $f(x) = \sum_{i=0}^{n} a_{i}x^{i} \in\mathbb{Z}[x],$all of whose coefficients are non-zero and all of its roots are distinct irrational numbers?

Even for $n = 3$, I am finding it hard to prove. My attempt for $n = 3$ is as follows:

For $n = 3$, I figured that if $f(x) = ax^{3}+bx^{2}+cx+d$ then its discriminant is equal to $$D(f) := b^{2}c^{2}-4ac^{3}-4b^{3}d-27a^{2}d^{2}+18abcd.$$ The roots of $f(x)$ will all be real if $D(f) > 0$. So, the roots of $f(x)$ will be three distinct irrational when $D(f) > 0$ and $f(r) \neq 0$ for any $r$ in the following set $R$.

$$R:=\big\{ r = \frac{p}{q} : p \mid d \text{ and } q \mid a \big\}.$$

Since $(0, \infty)$ is open in $\mathbb{R}$, there must be an Euclidean open set worth of $(a,b,c,d)\in\mathbb{R}^{4}$ such that $D(f) > 0$. We call this Euclidean open set $O_{1}$. I do not see what else can be said about this $O_{1}$.

My strategy was to show that set $S$ of $(a,b,c,d)\in\mathbb{Q}^{4}$ such that $f(x)$ is irreducible is “quite large” in some sense such that it actually intersects with $O_{1}$. Then, we choose $(a,b,c,d)\in\mathbb{Q}^{4}\cap O_{1}$, which will correspond to a $f(x)$ which is irreducible and has three distinct irrational roots.

Comment- See even if my approach for $n=3$ works, this method would not work for higher degrees because $D(f)>0$ is not enough to have that all roots are irrational.

Maulik
  • 396
  • This should give some inspiration https://math.stackexchange.com/questions/2791944 – Sil Jul 21 '22 at 20:10
  • Eisenstein polynomials –  Jul 22 '22 at 12:19
  • @franzlemmermeyer Are Eisenstein’s polynomials just the polynomials that satisfy the Eisenstein’s Irreducibility Criterion? If yes, is there a reference for the fact that Eisenstein polynomials have distinct irrational roots? If not, what do you mean? – Maulik Jul 26 '22 at 00:31
  • Do you know an irreducible polynomial whose roots are not distinct irrationals? –  Jul 26 '22 at 05:17
  • Sure.. the polynomial $(x^{5}-6x^{4}-6x^{3}-6x^{2}-6x-6)$ is irreducible by Eisenstein’s criterion for $p=3$ or $p=2$. But it has one irrational root $\approx 6.99964$ and other four roots are complex but not real. – Maulik Jul 26 '22 at 05:24

1 Answers1

2

You can drop the condition that the coefficients are all nonzero: If $f(x) \in \mathbb{Z}[x]$ of degree $n \geq 2$ is irreducible and all of its roots are real (and hence irrational because $f$ is irreducible), then for any integer $N$, the polynomial $f(x + N)$ is also irreducible and has all roots real. For all but finitely many $N$, $f(x + N)$ will also have all coefficients nonzero.

Indeed, the $x^k$ coefficient of a polynomial is nonzero if and only if the $k$-th derivative of the polynomial has a nonzero value at zero, and the $k$-th derivative of $f(x + N)$ evaluated at zero is equal to the $k$-th derivative of $f(x)$ evaluated at $N$. So, we just need to choose $N$ so that $f(N), f'(N), f''(N), \ldots, f^{(n-1)}(N)$ are all nonzero, which means avoiding the roots of a finite list of polynomials, each of which has finitely many roots.

So, it suffices to find an irreducible integer polynomial of degree $n$ with all roots real. This is equivalent to finding a totally real number field of degree $n$. There are a number of ways of constructing totally real number fields of any given degree; this answer gives one such way.

Daniel Hast
  • 5,095
  • Could you give me some references on where I can read more (in detail) about Totally real number field, and some other ways we can construct them? I have never heard of this thing before and google didn’t turn up much. – Maulik Jul 21 '22 at 19:05
  • Do you know what a field is in abstract algebra? A number field is a field that's given by taking the rational numbers and adjoining some roots of polynomials. A totally real number field is one with the property that every embedding of it into the complex numbers lands inside the real numbers. Working with number fields rather than just polynomials allows us to apply the tools of Galois theory, which is a very useful toolkit for problems like this. – Daniel Hast Jul 21 '22 at 20:08