Word problems and a homological finiteness condition for monoids (Q1105040)

A terminating Church-Rosser presentation (also known as a complete rewriting system) for a monoid S consists of a set R of ordered pairs of words in a free monoid F. This free monoid, together with the set of equalities arising from the pairs in R, should present S. If u,v\(\in F\), then v is a subword of u if \(u=avb\) for some a,b\(\in F\). A single replacement then consists of replacing a subword which is a left hand side of a pair in R by the right hand side of the pair. We require that there is no infinite sequence of such replacements. There is then a corresponding set of irreducible elements for which no replacement is possible and we further require that the natural map from irreducible elements of F to elements of S be bijective. In this interesting paper, such rewriting systems are exploited to give the beginnings - up to dimension 3 - of a free \({\mathbb{Z}}S\)-resolution of \({\mathbb{Z}}\). In the case of groups, this extends the well-known partial resolution which arises from the so-called `relation sequence'. Thus the term in dimension 0 is \({\mathbb{Z}}S\) and in dimensions 1 and 2 is a free module on a basis which is bijective with, respectively, the generators of F, and R. The author adds a term in dimension 3 for which the basis is bijective with the set of all overlaps of left-hand sides of rules if R. He then defines a reasonably natural (although not unique) boundary map and shows the necessary exactness. The result can also be used to give an example of group or monoid with a finite presentation and solvable word problem which can have no finite terminating Church-Rosser presentation. For the result shows that if S has a finite terminating Church-Rosser presentation then there must be a \({\mathbb{Z}}S\)-resolution of \({\mathbb{Z}}\) which is finitely generated in the first 3 dimensions; i.e. S has type \((FP)_ 3\). The interested reader should, however, be aware of results of \textit{D. J. Anick} [Trans. Am. Math. Soc. 296, 641-659 (1986; Zbl 0598.16028)] which appear to be related. With somewhat different, but apparently related, hypotheses, Anick gives a \({\mathbb{Z}}S\)-resolution of \({\mathbb{Z}}\) (in all dimensions). Where the hypotheses overlap, this resolution seems to agree with the one given here in the first three dimensions.