The lattice of full subsemigroups of an inverse semigroup. (Q871629)

Imposing restrictions on the lattice of all inverse subsemigroups of an inverse semigroup generally proves not to be fruitful, in that the resulting class of semigroups is very narrow. In a series of articles, the reviewer and others have shown that the sublattice comprising the full inverse subsemigroups -- those containing the semilattice of idempotents -- is much more rewarding. The authors of the paper under review show that many of the techniques applicable to the lattice of full inverse subsemigroups carry over to the lattice \(\text{Subf\,}S\) of all full subsemigroups of \(S\). In particular, this lattice is again a subdirect product of the corresponding lattices of the principal factors of \(S\) so that, for many lattice theoretic-properties of interest, one may focus on the 0-simple ones. Not surprisingly, a given lattice-theoretic property results in a narrower class of inverse semigroups when considering the lattice Subf. For instance, while the bicyclic semigroup has distributive lattice of full inverse subsemigroups, this is not so for the lattice of all inverse subsemigroups, with far-reaching consequences. The authors determine all inverse semigroups \(S\) for which \(\text{Subf\,}S\) is distributive, meet or join semidistributive, modular, or a chain. It is an open question whether the lattice of full subsemigroups of the bicyclic semigroup is upper semimodular.