Embedding theorems for HNN extensions of inverse semigroups. (Q886237): Difference between revisions

In general, given an inverse semigroup \(S\) and two inverse subsemigroups \(A\) and \(B\), together with an isomorphism \(\varphi\) between them, an HNN extension is an inverse semigroup \(T\), which one hopes contains \(S\), in which \(\varphi\) can be interpreted as some type of conjugation by elements of \(T\). The first part of this paper contains a remarkably lucid account of various concepts of HNN extensions that have been defined for inverse semigroups, and of the relationships among them. Examples demonstrate that the various concepts described therein generally yield different nonisomorphic extensions. The first primary purpose of the paper is to show that the definition given by \textit{N. D. Gilbert} [J. Algebra 272, No. 1, 27-45 (2004; Zbl 1041.20042)], in which \(A\) and \(B\) were required to be order ideals of \(S\), makes sense in general. The author notes that the original requirement on \(A\) and \(B\) naturally arises from a connection with groupoids and yields a normal form that may not be present in general. The second part of the paper uses Schützenberger graphs to show that Gilbert's extension embeds in one of the previously known extensions, one studied by the author in a sequence of papers; and also to show that \(S\) is full in Gilbert's extension (that is, no new idempotents are introduced) if and only if \(A\) and \(B\) are order ideals.