Suspension, shifts and unbounded modules over a simplicial ring

Written by Claude Opus 5.5 (Anthropic) on 23 October 2026. The question was posed by Claude while writing this course and solved by GPT-6 Astra (OpenAI) in ChatGPT's Pro mode; the solution was refereed by Claude Opus 5.5 in a separate session, and this lesson was written from it and refereed and revised by Claude Opus 5.5 in two further sessions. Public domain (CC0).

For a simplicial ring \(R\) of a topos with normalization \(A=NR\), this lesson proves five results.

Prerequisites. Normalization of modules over a simplicial ring (Theorem 4.1, Corollary 4.2); Triangles and Ext groups of simplicial modules (Conventions, 1.1); Differential graded modules in a topos (Lemma 1.1, Proposition 1.2); AI-integrated Stacks project: homotopic maps and normalized complexes [Tag 019S], maps out of bounded above complexes of projectives [Tag 064B].

Conventions

As in the lesson on normalization. For an \(R\)-module \(E\), put \(\gamma E=\gamma_{\mathbb Z}\otimes E\) and \(\sigma E=\sigma_{\mathbb Z}\otimes E\), degreewise, with \(R\) acting on \(E\). By 1.1 of the lesson on triangles, \(\mathsf c_E\colon0\to E\xrightarrow{i}\gamma E\xrightarrow{v}\sigma E\to0\) is exact. The contraction of \(\gamma_{\mathbb Z}\) is induced by a map of simplicial sets, so \(\gamma E\) is homotopy equivalent to \(0\) and \(N\gamma E\) is acyclic [Tag 019S].

For a left module \(M\) over \(A\), put \[ M[r]^i=M^{i+r},\quad d_{M[r]}=(-1)^rd_M, \] with action \(a\cdot m=(-1)^{r|a|}am\); this is again a module.

For a map \(f\colon X\to Y\) of modules, the cone \(C(f)\) has \(C(f)^i=Y^i\oplus X^{i+1}\) and \(d(y,x)=(dy+f(x),-dx)\), with action \(a(y,x)=(ay,(-1)^{|a|}ax)\).

1. Suspension and shift

Proposition 1.1. For an \(R\)-module \(E\), let \(s\colon C(Ni)\to N\sigma E\) be \((y,x)\mapsto v(y)\) and \(q\colon C(Ni)\to NE[1]\) be \((y,x)\mapsto x\). Both are \(A\)-linear quasi-isomorphisms, natural in \(E\). So \(\theta_E=[q][s]^{-1}\colon N\sigma E\to NE[1]\) is an isomorphism of \(D(A)\), natural in \(E\). Moreover:

Proof. The map \(s\). It is a chain map since \(vi=0\), and it is onto. Its kernel \(\{(i x',x)\}\) is isomorphic to the cone of the identity of \(NE\), which is contractible by the homotopy \((y,x)\mapsto(0,y)\).

The map \(q\). It is a chain map since \(d_{NE[1]}=-d\), and it is \(A\)-linear by the action rules. Its kernel \(N\gamma E\) is acyclic.

Consequences. Naturality is clear. Then \(H^{k}(N\sigma f)\cong H^{k+1}(Nf)\) naturally, so \(\sigma\) preserves quasi-isomorphisms. Naturality of \(\theta\) extends to the localization. By Theorem 4.1 of the lesson on normalization, \(a\mapsto N(a)\) is bijective, and \(N(a)\mapsto N(a)[1]\) is bijective. Hence \(a\mapsto\sigma a\) is bijective on morphism sets. \(\square\)

Put \(\Theta_0=1\) and \(\Theta_{k+1,X}=\) \(\Theta_{k,X}[1]\circ\theta_{\sigma^kX}\colon\) \(N\sigma^{k+1}X\to\) \(NX[k+1]\). Naturality of \(\theta\) gives, for \(a\colon\sigma^nE\to\sigma^{n+p}F\), \[ \begin{aligned} &\Theta_{n+p+1}N(\sigma a)\Theta_{n+1}^{-1}\\ &\quad=(\Theta_{n+p}N(a)\Theta_n^{-1})[1]. \end{aligned}\tag{1.2} \]

Theorem 1.3 (stable Ext groups). For \(p\in\mathbb Z\) put \[ \begin{aligned} &\operatorname{Ext}^p(E,F)=\\ &\varinjlim_{n\geq\max(0,-p)} \operatorname{Hom}(\sigma^nE,\sigma^{n+p}F), \end{aligned} \] the morphisms being those of \(D_\bullet(R)\) and the transition maps \(a\mapsto\sigma a\). The maps \(a\mapsto(\Theta_{n+p}N(a)\Theta_n^{-1})[-n]\) induce an isomorphism, natural in \(E\) and \(F\), \[ \operatorname{Ext}^p(E,F)\cong\operatorname{Hom}_{D(A)}(NE,NF[p]). \] For commutative \(R\), these are the Ext groups of Definition 2.1 of the lesson on triangles.

Proof. Each stage map is bijective, by Theorem 4.1 of the lesson on normalization and because shifts are automorphisms. By (1.2), these maps commute with the transition maps. \(\square\)

Proposition 1.4 (inverting the suspension). Let \(\operatorname{SW}\) be the category whose objects are pairs \((E,n)\), with \(E\in D_\bullet(R)\) and \(n\geq0\), and whose morphisms are \[ \begin{aligned}&\operatorname{Hom}((E,n),(F,m))=\\&\varinjlim_{k\geq n,m}\operatorname{Hom}(\sigma^{k-n}E,\sigma^{k-m}F),\end{aligned} \] with composition at a common \(k\). Then \((E,n)\mapsto NE[-n]\) is an equivalence from \(\operatorname{SW}\) onto the full subcategory \(D^-(A)\) of \(D(A)\) of objects with bounded above cohomology.

Proof. The transition maps are bijective (Proposition 1.1), so composition is well defined. The maps \[ a\mapsto(\Theta_{k-m}N(a)\Theta_{k-n}^{-1})[-k] \] are bijections onto \(\operatorname{Hom}_{D(A)}(NE[-n],NF[-m])\). By (1.2) they are compatible with the transitions, and they are compatible with composition. Every \(NE[-n]\) has \(H^i=0\) for \(i>n\). Conversely, if \(H^iM=0\) for \(i>n\geq0\), then \(M[n]\cong NE\) for some \(E\) by Theorem 4.1 of the lesson on normalization, so \(M\cong NE[-n]\). \(\square\)

2. Telescopes

Proposition 2.1. Let \(M\) be a left \(A\)-module and \(M_n=\tau^{\leq n}M\) for \(n\geq0\), a submodule by Lemma 1.1 of the lesson on differential graded modules. Put \(B=\bigoplus_nM_n\), and let \(s\colon B\to B\) send the summand \(n\) to the summand \(n+1\) by inclusion. Then the map \(C(1-s)\to M\), \((y,x)\mapsto\sum_ny_n\), is an \(A\)-linear quasi-isomorphism. Each \(M_n\) is isomorphic in \(D(A)\) to \(NE_n[-n]\) for an \(R\)-module \(E_n\).

Proof. The exact sequence. In degree \(i\), \(M_n^i=M^i\) for \(n>i\). The sequence \[ 0\to B\xrightarrow{1-s}B\xrightarrow{\Sigma}M\to0 \] is exact on presheaf sums, by inspecting the first nonzero component and the standard description of a sequential colimit. It stays exact after sheafification.

The quasi-isomorphism. The map \(C(1-s)\to M\) is a chain map because \(\Sigma(1-s)=0\). It is onto, and its kernel is \(\{(y,x)\colon y\in\operatorname{Im}(1-s)\}\), which is isomorphic to the cone of the identity of \(B\). That cone is contractible by the homotopy \((z,x)\mapsto(0,z)\), so the map is a quasi-isomorphism.

The pieces. The module \(M_n[n]\) has no positive cohomology. Theorem 4.1 of the lesson on normalization gives \(M_n[n]\cong NE_n\). \(\square\)

3. Complexes of simplicial modules

Proposition 3.1. Let \(T\) be the topos of sets and \(D^b_{\mathrm{out}}\) the bounded derived category of the abelian category \(s\mathrm{Ab}\) of simplicial abelian groups. For a bounded complex \(E^\bullet\) put \(B^{p,q}=N_{-q}E^p\) for \(q\leq0\). Let \(\operatorname{Tot}NE\) be its total complex, with differential \(d_{\mathrm{out}}+(-1)^pd_{\mathrm{in}}\) on \(B^{p,q}\). Then:

  1. \(\operatorname{Tot}N\) induces a functor \(D^b_{\mathrm{out}}\to D(\mathbb Z)\);
  2. this functor is not full.

Proof. (1) Let \(E^\bullet\) be acyclic and \(z\) a total cocycle of degree \(r\), with outer support in \([a,b]\).

The total complex of an outer cone is the cone of the total complexes. So outer quasi-isomorphisms go to quasi-isomorphisms.

(2) The constant group \(c\mathbb Z\) is projective, because \(\operatorname{Hom}(c\mathbb Z,V)=V_0\) is exact in \(V\). Let \(K\) be the nerve of the group \(\mathbb Z\), so that \(NK=\mathbb Z\) in degree \(-1\) and \(K_0=0\). Put \(X=c\mathbb Z\) in outer degree \(-1\) and \(Y=K\) in outer degree \(0\). The Hom complex from \(X\) to \(Y\) is built from \(\operatorname{Hom}(c\mathbb Z,K)=K_0=0\), so \(\operatorname{Hom}(X,Y)=0\) [Tag 064B]. Both totalizations are \(\mathbb Z\) in degree \(-1\), with endomorphism group \(\mathbb Z\) [Tag 064B]. \(\square\)

4. Exercises

Exercise 4.1. Show that for \(R=\mathbb Z\) in the topos of sets, \(\operatorname{Ext}^p(\mathbb Z,\mathbb Z)\) is \(\mathbb Z\) for \(p=0\) and \(0\) otherwise.

5. Complete solutions

4.1. By Theorem 1.3, \(\operatorname{Ext}^p(\mathbb Z,\mathbb Z)=\operatorname{Hom}_{D(\mathbb Z)}(\mathbb Z,\mathbb Z[p])\). Since \(\mathbb Z\) is a bounded complex of projectives, this is computed by homotopy classes of chain maps [Tag 064B]. The result is \(\mathbb Z\) for \(p=0\) and zero otherwise.

Sources

Stable Ext groups of simplicial modules and their comparison with differential graded modules are due to Illusie, in his work on the cotangent complex. The main argument follows a solution produced with ChatGPT (OpenAI) in October 2026 and completed here.