Algebra and local presentability: how algebraic are they? (A survey) (Q680309): Difference between revisions

This survey paper reviews the results obtained by the authors and others over a question raised by Bill Lawvere, and which they quipped ``How algebraic is algebra'' in their common paper [\textit{J. Adámek} et al., Theory Appl. Categ. 8, 253--283 (2001; Zbl 0978.18006)]. More precisely, after recalling their definitions of the 2-categories \(\mathbf{VAR}\) of (many-sorted, finitary) varieties and \(\mathbf{LFP}\) of locally finitely presentable categories (i.e., of models of algebraic and essentially algebraic theories respectively), and noticing that they are not ``algebraic'' (i.e., the forgetful functors to \(\mathbf{CAT}\) are not monadic in an appropriate sense), they define what an algebraic hull for them should mean. As in General Algebra, this would be the category of all categories satisfying the equations satisfied by all varieties, resp. by all locally finitely presentable categories. For this they need to define what operations (and their arities) mean, as well as equations between them, which they recall in detail. A characterization of the algebraic hulls of \(\mathbf{VAR}\) and of \(\mathbf{LFP}\) with respect to ranked operations were obtained (as the 2-categories of so-called algebraically exact categories and of precontinuous categories respectively), but not the ones with respect to all operations, or to all the ones of small arities. However, they note that under a certain set-theoretical assumption, all operations of small arities on \(\mathbf{LFP}\) are actually ranked. Finally the characterization generalizes in a straightforward way to the infinitary case for \(\mathbf{LFP}\), but not for \(\mathbf{VAR}\), which remains an open problem (the reason being that the sifted categories, a concept which plays in varieties the role played by the filtered ones in the lfp categories, have their expected infinite generalizations provenly equivalent to the ones of filtered categories).