Let $\alpha \in \mathbb{C}$ be an algebraic integer of degree $n$, not a unit, and let $R = \mathbb{Z}[\alpha]$. Then every element $\beta \in R$ can be written uniquely in the form

$$c_0+ c_1 \alpha+\ldots +c_{n-1} \alpha^{n-1},$$ with $c_i=c_i(\beta) \in R$. Let $S \subset R$ denote the set of $\beta \in R$ such that $c_0(\beta) \equiv 1 \mod N(\alpha)$. Equivalently, $S$ is the set of $\beta \in R$ such that $\beta \equiv 1 \mod (\alpha)$.

Let $F(S) \subset \mathrm{Frac}(R)$ denote the set of ratios of elements of $S$, i.e. the (multiplicative) abelian group generated by $S$. Here we have used the fact that $\alpha$ is not an algebraic unit to make sure that $0 \not\equiv 1 \mod (\alpha)$, so $0 \notin S$ and this construction makes sense.

Claim: As an abelian group, $F(S)$ has infinite rank.

It seems that the logical way to do this would be to show that $S$ itself contains infinitely many irreducible elements (considered irreducible as elements in $R$), but I am not sure how to proceed from here. One avenue for exploration is based on the fact that irreducible elements in $\mathcal{O}_k$ (for $k=\mathbb{Q}(\alpha)$) that happen to lie in $S$ remain irreducible over our smaller ring. Thus, it would suffice to find infinitely many irreducible elements in $\mathcal{O}_k \cap S$, which may be a better understood problem. (Perhaps with the Chebotarev density theorem?)

A second method that I have considered attempts to find infinitely many irreducible elements of a particular form in $S$, e.g. irreducibles of the form $1+m\alpha$, with $m \in \mathbb{Z}$. If the norm form polynomial $$f(x):=N(1+x\alpha) = \prod_{i=1}^n \left(1-x \alpha^{(i)}\right) \in \mathbb{Z}[t]$$ assumes infinitely many prime values, then the sequence $\{1+m\alpha\}$ contains infinitely many irreducibles (as a factorization of $1+m\alpha$ would yield a non-trivial factorization of the norm). Here, $f(x)$ satisfies the hypotheses of Bunyakovsky's conjecture, so this claim is likely true (even if this avenue is not the best for proving it). Perhaps a variant of this method could pan out, if we ask for $f(x)$ to admit infinitely many distinct prime divisors (as $x$ varies), instead of infinitely many prime values?


Let $p\ge 2$ be an integer dividing $\alpha$ in $R$. It exists because $\alpha$ is not a unit.

Then $F(1+p\mathbb Z)=(1+p\mathbb Z)(1+p\mathbb Z)^{-1}$ is a subgroup of $F(S)$, and it is enough to show that it is not finitely generated as group. If it was, then only finitely many prime numbers appear in the factorization of elements of $F(1+p\mathbb Z)$. But by Dirichlet there are infinitely many prime numbers in $1+p\mathbb Z$. Contradiction.

| cite | improve this answer | |

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.