Construction 20.8.1. Let \(f\colon A\to B\) be a morphism of Waldhausen categories. Recall that \(i_n\colon [n]\to [n+1]\) is the upper embedding defined through \(i\mapsto i+1\). We form the pullback square below. Commutative diagram; see the PDF for its LaTeX source This construction clearly is functorial in \([n] : \Delta\catop\).

Proposition 20.8.2 (Waldhausen). Let \(C\) be an admissible class of Waldhausen categories. We consider a morphism \(f\colon A\to B\) in \(\WaldCat_C\). For any additive functor \(E\colon\WaldCat_C\to \An_{\bullet}\), there is a canonical pullback square of the form below. Commutative diagram; see the PDF for its LaTeX source In particular, there is a canonical fiber sequence of the form below. Commutative diagram; see the PDF for its LaTeX source

Proof

The last assertion follows from the first by Proposition 20.4.13. The proof of this proposition is in fact a variation on the proof of the latter. Since \[ {\underset{[n] : \Delta\catop}{\colim}\, }E(\Scons_{n+1}(B))=0\, , \] by Lemma 13.4.4, it suffices to prove that, for each \(n : \N\), the induced commutative square Commutative diagram; see the PDF for its LaTeX source is a pullback square. Through the identifications \(\Scons_n(A)=A^{[n-1]}\) for \(n>0\), one can identify \(\Scons_n(f\colon A\to B)\) as a category with terms the pairs \((x,y)\) where \(x\) is a term in \(A^{[n-1]}\) and \(y\) a commutative diagram in \(B\) of the form Commutative diagram; see the PDF for its LaTeX source in which all squares are pushout squares.

Let \(v\colon \Scons_n(B)\hookrightarrow\Scons_{n+1}(B)\) be the full embedding defined by sending a sequence \[ x_1\to\cdots\to x_n \] to the sequence \[ 0\to x_1\to\cdots\to x_n\, . \] Since \(u\) is a section of \(i^*_n\), there is a canonical pullback square of the form below Commutative diagram; see the PDF for its LaTeX source (in particular \(u\) is fully faithful and a section of \(p_n\)). There is also a full embedding \[ j\colon B\hookrightarrow\Scons_n(f\colon A\to B) \] defined by sending \(x\) to the pair consisting of the sequence of length \(n\) \[ x\xrightarrow{\id_x}\cdots\xrightarrow{\id_x}x \] and of the commutative diagram Commutative diagram; see the PDF for its LaTeX source in which all the vertical maps are identities. The functor \(j\) is a section of the functor \(\pi\) sending a pair \((x,y)\) to the object \(y_1\). There is an obvious functorial exact sequence of the form Commutative diagram; see the PDF for its LaTeX source and therefore a direct application of Proposition 20.4.10 shows that the pair of functors \[ (p_n,\pi)\colon\Scons_n(f\colon A\to B)\to \Scons_n(A)\times B \] induces an equivalence \[ E(\Scons_n(f\colon A\to B))= E(\Scons_n(A))\times E(B)\, . \] It follows that the squares Commutative diagram; see the PDF for its LaTeX source are equivalent, hence are both cartesian, which achieves this proof.

Theorem 20.8.3 (Thomason-Trobaugh). Let \(B\) be a small stable category and \(A\) a thick subcategory of \(B\). Then the pullback square Commutative diagram; see the PDF for its LaTeX source induces a fiber sequence Commutative diagram; see the PDF for its LaTeX source in \(\An_\bullet\).

Proof

We let \(C\) be the Waldhausen category defined by \(C=B\) as categories, with cofibrations being all maps but with weak equivalences being the maps whose cofiber lies in \(A\). We observe then that the equivalence \[ \Scons_{n+1}(B)=B^{[n]} \] induces an equivalence of the form \[ \Scons_n(f\colon A\to B)=w_n(C)\, . \] Applying Proposition 20.8.2 to the admissible class of Waldhausen categories given by small stable categories, with \(E=K\), and taking into account Remark 20.7.9, we see that there is a canonical fiber sequence of the form below. Commutative diagram; see the PDF for its LaTeX source It suffices now to see that the morphism of Waldhausen categories \(C\to B/A\) induced by the functor \(B\to B/A\) yields an equivalence of the form \[ K(C)=K(B/A)\, . \] We want to understand the map \[ {\underset{[m] : \Delta\catop}{\colim}\, } {\underset{[n] : \Delta\catop}{\colim}\, } w_m(\Scons_{n}(C))^\simeq\to {\underset{[m] : \Delta\catop}{\colim}\, } {\underset{[n] : \Delta\catop}{\colim}\, } \Iso_m(\Scons_{n}(B/A)) \] where \(\Iso_n=\Fun([n],(-)^\simeq)\). The assignment \[ [m]\mapsto w_m(\Scons_n(C))^\simeq \] is the nerve of the category of weak equivalences \(w(\Scons_n(C))\), whereas \[ [m]\mapsto \Fun([m],\Scons_{n}(B/A)^\simeq) \] is the nerve of the anima \(\Scons_{n}(B/A)^\simeq\). Taking the colimit in the variable \([m]\) amounts to consider the map \[ \Pi_\infty(w(\Scons_n(C)))\to \Scons_{n}(B/A)^\simeq\, . \] It suffices to check that the latter is an equivalence for all \(n\). The case \(n=0\) is obvious (we are comparing two terminal categories). For \(n>0\), identifying \(\Scons_n\) with \((-)^{[n-1]}\), taking into account Theorem 18.4.7 with \(P=[n-1]\), this is a particular case of Proposition 17.1.14.