2
$\begingroup$

Let $S$ be a semigroup. An ideal (of $S$) is a subset $I$ of $S$ such that $SI$ and $IS$ are both contained in $I$. The non-empty ideals constitute a subsemigroup, $\mathfrak I(S)$, of the power semigroup of $S$ under the operation of setwise multiplication induced by $S$ on its parts; and for some reasons, one may want to ensure that the finitely generated ideals constitute a subsemigroup of $\mathfrak I(S)$. However, this is not going to be the case if a finitely generated ideal is taken to be a set of the form $X \cup XS \cup SX \cup SXS$ for some finite $X \subseteq S$ (which looks like the most obvious/natural thing to do). It goes the same with principal ideals, which are commonly defined as monogenic ideals, that is, sets of the form $\{a\} \cup aS \cup Sa \cup SaS$ with $a \in S$.

On the other hand, letting a finitely generated ideal be an ideal $I$ such that $I = X \cup XS = Y \cup SY$ for some finite sets $X, Y \subseteq S$ would turn the set of all finitely generated ideals into a subsemigroup $\mathfrak I_{\rm fin}(S)$ of $\mathfrak I(S)$ (conditional is mandatory here). Under this "alternative definition", a monogenic ideal would then be an ideal $I$ such that $I = \{a\} \cup aS = \{b\} \cup Sb$ for some $a, b \in S$, and the monogenic ideals would form a subsemigroup of $\mathfrak I_{\rm fin}(S)$. In a way, this makes a lot of sense, as we are asking that $I$ is finitely generated (resp., principal) both as a left and as a right ideal; and these left and right conditions are, in principle, independent from each other.

Question(s). Have these "alternative notions" (of finitely generated ideals and principal ideals) been considered before? If yes, under which names?

Note that the classical notions and the alternative ones coincide when $S$ is a commutative semigroup (or, more generally, a duo semigroup). In addition, any finitely generated (resp., principal) ideal in the alternative sense is a finitely generated (resp., principal) ideal in the classical sense.

$\endgroup$
4
  • $\begingroup$ I’ve not seen it. I think it very rare for an ideal to be principal as both a left and right ideal in general. $\endgroup$ Commented Mar 20 at 11:43
  • $\begingroup$ I don't disagree. However, there are instances where the semigroup of (two-sided) ideals, which are finitely generated both as left and as right ideals, is sufficiently large to encode a significant fragment of the structure of the original semigroup. And that's where my motivation comes from. $\endgroup$ Commented Mar 20 at 14:40
  • $\begingroup$ The finitely generated case is more reasonable but I don't think very studied $\endgroup$ Commented Mar 20 at 14:53
  • $\begingroup$ In the commutative case (where the two definitions agree), $\mathfrak I_{\rm fin}(S)$ has been extensively studied by people in multiplicative ideal theory as an important special case of a (weak) ideal system. With that said, one could always define $\mathfrak I_{\rm fin}(S)$ as the subsemigroup of the ideal semigroup generated by the ideals that are finitely generated in the usual sense. In hindsight, I'm more or less convinced that this would be more natural than adopting the alternative definition of 'finitely generated ideal' suggested in the OP. $\endgroup$ Commented Mar 28 at 2:56

0

You must log in to answer this question.

Browse other questions tagged .