Localization and duality in topology and modular representation theory. (Q2481809)

\textit{W.~G. Dwyer}, \textit{S. Iyengar} and the second author took several concepts from commutative algebra and worked out what they meant for \(\mathbb{S}\)-algebras [Adv. Math. 200, No. 2, 357-402 (2006; Zbl 1155.55302)]. Recall that in algebraic topology, an \(\mathbb{S}\)-algebra is a ring spectrum in a category of spectra whose smash product is \textit{strictly} commutative and associative. The authors pursue this line of development further, obtaining a duality theory for localizations. They use it to prove a conjecture in modular representation theory due to the first author [New York J. Math. 7, 201-215 (2001; Zbl 0994.20009)]. Let \(G\) be a finite group, \(k\) an algebraically closed field of characteristic~\(p\), and \(\mathfrak p\) a non-maximal homogeneous prime ideal in the cohomology ring \(H^*(G,k)\). It is known that there is an idempotent module \(\kappa_{\mathfrak p}\) which picks out the layer of the stable module category corresponding to \(\mathfrak p\). The theorem states that after a shift, the Tate cohomology of \(\kappa_{\mathfrak p}\) is the injective hull of \(H^*(G,k)/\mathfrak p\). The first author has since obtained another proof of this result, which is shorter but less conceptual [J. Pure Appl. Algebra 212, No. 7, 1744-1746 (2008; Zbl 1156.20042)]. This result is an application of the work on duality that forms the backbone of the paper. The context is a diagram consisting of two module categories and three derived categories. The map from the bounded derived category of \(kG\)-modules to the stable module category is extended to a map \(\Psi\) from the unbounded derived category. The main result (Theorem 12.1) is a duality statement which looks like it is induced by a certain Gorenstein map: only this map does not exist.