Generalized André-Quillen cohomology (Q847566): Difference between revisions

The use of (co)homology theories has become increasingly widespread since the first examples appeared in algebraic topology, group theory and algebraic geometry. Since the work of Barr, Beck, André and Quillen it has been known how to set up such functors in non-abelian settings, and more recent work on deformation theory and obstruction theory has led to many more examples being introduced. Roughly speaking, a cohomology theory on a model category should be a contravariant functor on the associated homotopy category taking values in something like a triangulated additive category, while a homology theory should be a covariant functor. The present paper attempts to lay down the conditions required for a representable functor on some category with structure, \(\mathcal{C}\), to be a cohomology theory with useful properties. To do this, Bastaini and Ehresmann's notion of a \textit{sketch} is used. Then for a model category \(\mathcal{E}\) equipped with a suitable kind of sketch \(\Phi\), and enough extra structure made precise in the paper, it is shown how to define a cohomology theory on \(\mathcal{E}\) using the total left derived functor \(\mathbb{L}\text{map}_{\mathcal{E}}(-,G)\) associated to an object \(G\in\mathcal{E}\) which is an algebra over \(\Phi\). This takes values in a semi-triangulated category \(\mathcal{V}\) so that \(\mathcal{E}\) is enriched over \(\mathcal{V}\). A similar construction defines homology functors. The main benefits from this work seem to be a clearer understanding of the data required for defining (co)homology theories in very general situations, and the resulting general machinery for comparison using spectral sequences. It appears that most of the existing examples that have been studied previously fit into this framework.