A realization result for systems of sets of lengths

Divisibility and factorizations in commutative rings (13A05) Dedekind, Prüfer, Krull and Mori rings and their generalizations (13F05) Arithmetic theory of semigroups (20M13)

Abstract: Let

m a t h c a l L^{*}

be a family of finite subsets of

m a t h b b N_{0}

having the following properties. (a).

0, 1 i n m a t h c a l L^{*}

and all other sets of

m a t h c a l L^{*}

lie in

m a t h b b N_{g e 2}

. (b). If

L_{1}, L_{2} i n m a t h c a l L^{*}

, then the sumset

L_{1} + L_{2} i n m a t h c a l L^{*}

. We show that there is a Dedekind domain

D

whose system of sets of lengths equals

m a t h c a l L^{*}

.

Review text

Let \(D\) be an integral domain or a monoid and \(a\in D\) be nonzero. A positive integer \(k\) is a \emph{factorization length} of \(a\) if there are irreducible elements \(u_1,\ldots,u_k\) of \(D\) such that \(a=u_1\cdots u_k\); the set of all factorization lengths is denoted by \(\mathsf{L}(a)\). The set \(\mathcal{L}(D)\) of all \(\mathsf{L}(a)\), as \(a\) ranges in \(D\) (the \emph{system of sets of lengths} of \(D\)) is a useful way to study the non-uniqueness of factorization in \(D\). It always satisfies two conditions: \begin{itemize} \item[1.] \(\{0\},\{1\}\in\mathcal{L}(D)\), and all other sets of \(\mathcal{L}(D)\) are contained in \(\mathbb{N}_{\geq 2}\); \item[2.] if \(L_1,L_2\in\mathcal{L}(D)\), then \(L_1+L_2\subseteq L_3\) for some \(L_3\in\mathcal{L}(D)\). \end{itemize} (Here \(L_1+L_2=\{l_1+l_2\mid l_1\in L_1,l_2\in L_2\}\) is the sumset of \(L_1\) and \(L_2\).) There are several examples where every finite subset of \(\mathbb{N}_{\geq 2}\) is in \(\mathcal{L}(D)\) (for example some ring of integer-valued polynomials, cf. [\textit{S. Frisch}, Monatsh. Math. 171, No. 3--4, 341--350 (2013; Zbl 1282.13004)]. It is not known which families \(\mathcal{L}^\ast\) with the previous properties arise as systems of sets of lengths. In this paper, the authors prove that, if \(\mathcal{L}^\ast\) is a family of finite subsets that satisfies the previous conditions and that is closed by sumsets (i.e., such that \(L_1+L_2\in\mathcal{L}^\ast\) if \(L_1,L_2\in\mathcal{L}^\ast\)) then there are a Krull monoid \(H\) and a Dedekind domain \(D\) such that \(\mathcal{L}(H)=\mathcal{L}^\ast=\mathcal{L}(D)\). The proof of the monoid case is obtained in two steps. The first step involves constructing, from an arbitrary finite subset \(L\subset\mathbb{N}_{\geq 2}\), a Krull monoid \(H\) whose atoms have some specific properties, and such that \(\mathcal{L}(H)\) contains exactly \(\{0\}\), \(\{1\}\) and the sets \(\{y+nL\mid y,n\in\mathbb{N}_0\}\); the second step consists of applying this construction to the indecomposable elements of \(\mathcal{L}^\ast\) (which, under the working hypothesis, is a semigroup), and then take the disjoint union of these monoids. The result of the domain case is obtained from the monoid case by the use of a transfer homomorphism.

Recommendations

Cites work

Cited in

(19)

This page was built for publication: A realization result for systems of sets of lengths

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2143228)