Existence and uniqueness of linking systems: Chermak's proof via obstruction theory (Q382426)

The \(p\)-local structure of a finite group \(G\) is best described algebraically by its \textit{saturated fusion system}. Given a Sylow \(p\)-subgroup \(S\) this is a category whose objects are subgroups of \(S\) and whose morphisms are group homomorphisms induced by conjugation in~\(G\). There is an analogous abstract notion, defined over a finite group \(S\) without the presence of an ambient group \(G\), where morphisms are certain monomorphisms between subgroups. Topologically however this is not good enough to recover the \(p\)-completed classifying space \(BG^\wedge_p\). One needs a centric linking system as introduced by \textit{C. Broto} et al. [J. Am. Math. Soc. 16, No. 4, 779--856 (2003; Zbl 1033.55010)]. Together with the fusion system this forms an object known as a \textit{\(p\)-local finite group}. The author proved in [Math. Proc. Camb. Philos. Soc. 137, No. 2, 321--347 (2004; Zbl 1077.55006)] and [Mem. Am. Math. Soc. 848, 102 p. (2006; Zbl 1095.55008)] that two groups having isomorphic fusion systems also have isomorphic linking systems, which was the first piece of evidence -- and maybe the only one! -- why every abstract saturated fusion system over any \(p\)-group \(S\) should have a unique centric linking system, hence a classifying space. This was one of the major open problems in the field for the last ten years and has been proven now by \textit{A. Chermak} [Acta Math. 211, No. 1, 47--139 (2013; Zbl 1295.20021)]. In the article under review the author proposes a different proof based on the obstruction theory developed in [Zbl 1033.55010]. The idea is that a centric linking system corresponds to a rigidification of the homotopy functor associating to each centric subgroup \(P < S\) in \(\mathcal O(\mathcal F^c)\) its classifying space \(BP\). The obstructions to existence and uniqueness live respectively in \(\text{lim}^3\) and \(\text{lim}^2\) over \(\mathcal O(\mathcal F^c)\) of the functor \(\mathcal Z_{\mathcal F}\) sending \(P\) to \(C_S(P) = Z(P)\). In fact the main theorem establishes the vanishing of \textit{all} higher limits \(\text{lim}^k_{\mathcal O(\mathcal F^c)} \mathcal Z_{\mathcal F}\) for \(k \geq 2\) at the prime \(2\) and \(k\geq 1\) at odd primes. The proof is done in several steps. Construct first certain ``invariants intervals'' in \(\mathcal F^c\) and consider the limits of the associated subquotient functor of \(\mathcal Z_{\mathcal F}\). This is Chermak's reduction to a so-called general setup. Second, to prove the vanishing of these limits, reduce the problem to a ``reduced setup''. Assemble then the results by using long exact sequences of derived functors and results on higher limits which are recalled in the first section of the article. This is a very short and imprecise description of the strategy of the proof, which does not even mention the central role of the Thompson subgroup and best offenders. We refer the reader to the article itself, and in particular to the beautifully written introduction.