Perfecting the nearly perfect (Q959021)

In their paper in [Compos. Math. 119, No. 2, 133--155 (1999; Zbl 1011.11075)], \textit{T. Chinburg, M. Kolster, G. Pappas} and \textit{V. Snaith} (CKPS for short) defined a notion of ``nearly perfect complexes'' (npc for short) of modules over the group ring of a finite group and associated to each such npc a canonical Euler characteristic. This characteristic was reinterpreted and refined by \textit{D. Burns, B. Köck} and \textit{V. Snaith} [J. Algebra 272, No. 1, 247--272 (2004; Zbl 1067.19001)] by using an explicit mapping cone construction. Here the author introduces a natural variant of the notion of npc which is much more amenable to working in derived categories: to each npc he associates a ``derived nearly perfect complex'' (or dnpc), which is canonical in many cases of arithmetic interest (in particular in all arithmetic applications given by CKPS); then he constructs a functor from the category of dnpc's to a suitable derived category of perfect complexes and he shows that, for any given npc, the Euler characterisitc of the associated perfect complex is equal to the Euler characteristic (in the serve of CKPS) of the underlying npc. The author's ``derived'' approach not only presents technical advantages which enable him for example to answer a question posed by Snaith (the CKPS Euler characteristic is additive on suitable exact triangles), but it also allows him to construct canonical complexes which, for the purpose of formulating both special value conjectures and arithmetic duality theorems, provide adequate substitutes for Lichtenbaum's (not yet fully working?) ``Weil-étale'' cohomology for arithmetic schemes.