Boundedly controlled algebraic \(K\)-theory of spaces and its linear counterexamples (Q1181427): Difference between revisions

\textit{F. Waldhausen} has defined algebraic K-theory as a functor that maps \textit{categories with cofibrations and weak equivalences} to \textit{spaces}. An excellent account is available in his paper ``Algebraic K-theory of spaces'' [Lect. Notes Math. 1126, 318-419 (1985; Zbl 0579.18006)]. When a suitable category, \({\mathbf R}_{fd}(X)\), of retractive spaces over a fixed space \(X\) is fed into this functor, one arrives at the algebraic K- theory of the space \(X\), usually denoted \(A(X)\). Waldhausen's theory is also known as the non linear K-theory in contradistinction from Quillen's linear K-theory. For a space \(X\), the two are connected by a linearization map \(\lambda: A(X) \to K({\mathbb{Z}}\pi_ 1(X))\). Let \(B\) be a fixed, metric space, and let \(Z_ i\) be a space parametrized over \(B\) by means of a map \(p_ i: Z_ i \to B\), \((i = 1, 2)\). A map \(f: Z_ 1 \to Z_ 2\) is boundedly controlled, if dist\((p_ 2f(z),p_ 1(z))\) is bounded (\(z \in Z_ 1\)). In an earlier paper [Acta Math. 165, 161-187 (1990; Zbl 0722.19002)], the author has generalized \({\mathbf R}_{fd}(X)\) to a functor \({\mathbf R}_{fd}(Z,p)\) on the category \(\mathbf{ {Top}}^{bc}/B\) of boundedly controlled maps between spaces parametrized over \(B\). Thus, there is a similar extension of \(A(X)\) to a functor \(A(Z,p)\). In this paper, the author constructs a linearization map \(\lambda: A(Z,p) \to K({\mathbf P}_{\pi({\mathbf Z})})\) where the right hand side is Quillen's K-theory of the category of finitely generated, projective modules over the boundedly controlled fundamental groupoid \(\pi({\mathbf Z})\) of \((Z,p)\), in the sense of the \textit{D. R. Anderson} and the reviewer [Boundedly controlled topology. Foundations of algebraic topology and simple homotopy theory (Lect. Notes Math. 1323) (1988; Zbl 0646.57001)]. He also shows that \(\lambda\) induces an isomorphism on \(\pi_ 0\), and he uses this fact and results by \textit{E. K. Pedersen} and \textit{C. A. Weibel} [Lect. Notes Math. 1126, 166-181 (1985; Zbl 0591.55002)], to show that the spectrum formed by the spaces \(A(X \times {\mathbb{R}}^ i,\text{proj})\) (\(i = 1, 2, 3 \dots\)) is non connective (in general). Actually, in the earlier paper, the author made a somewhat unnatural restriction on the objects \((Z,p)\) considered. The review given in section 1 of the present paper removes this restriction. To do this, one needs some facts about mapping telescopes and the reviewer has some reservations about the formalities of the setup presented to get at these -- in particular, he cannot see how the commutative diagram \((**)\) on page 198 can give rise to a map of telescopes as claimed. However, he does believe that these inaccuracies can be repaired. [Also, in case of infinitely many components of the space \((Z,p)\), the reviewer is sceptical about the decomposition result on p. 202, but this is of no consequence for the rest of the paper]. Section 2 gives a succinct and very readable resume of the necessary algebraic topological results from the above mentioned book by Anderson and the reviewer. [Misprint: In line 4, p. 206, the target category should be {\textbf{Gpoid}} rather than {\textbf{Top}}]. Section 3 is at the heart of the present paper. It shows that (except for a limit formation as \(n \to \infty\)), the category \({\mathbf R}_{fd}(Z,p)\) can be replaced by a subcategory \({\mathbf R}_{fd}^ n(Z,p)\) of ``\(n\)- spherical objects''. These are defined in terms of the boundedly controlled fundamental groupoid and homology functors of Anderson and the reviewer [loc. cit.], and the proof involves a good deal of (classically inpired) manipulations with these notions. [The reviewer does not understand the last two sentences in the proof of Lemma 3.3]. Since \(n\)-spherical objects have exactly one non trivial homology ``group'', there is an obvious functor \({\mathbf R}_{fd}^ n(Z,p) \to {\mathbf P}_{\pi({\mathbf Z})}\); this passes to (the limit and) K-theory to give the desired linearization. Section 4 also contains a sketch of a more direct approach to linearization, where, instead of restricting to \(n\)-spherical objects on the non linear side, one extends to chain complexes on the linear side. One can then use the full chain complex to construct the desired functor.