---
title: cdh-Motivic Cohomology
url: https://www.emergentmind.com/topics/cdh-motivic-cohomology
type: topic
---

# cdh-Motivic Cohomology

Searching arXiv for the cited papers and recent work on cdh/pro-cdh motivic cohomology.
cdh-motivic cohomology is the collection of motivic cohomology theories, descent formalisms, and comparison results organized by the cdh topology, the Grothendieck topology generated by Nisnevich coverings and abstract blow-up squares. In its classical \( \mathbb A^1 \)-invariant form, it is obtained by cdh-sheafifying motivic complexes coming from smooth schemes; in more recent non-\( \mathbb A^1 \)-invariant and pro-cdh forms, it is designed to retain sensitivity to singularities and nilpotents while preserving blow-up descent, projective bundle formulas, and Atiyah–Hirzebruch-type spectral sequences to \(K\)-theory [2508.09915] [2309.08463] [2412.06635] [2409.14295].

## 1. The cdh topology and the cdh-local motivic setting

The cdh topology refines the Nisnevich topology by adjoining abstract blow-up squares. Concretely, an abstract blow-up square is a cartesian square
\[
\begin{tikzcd}
Y' \ar[r] \ar[d] & X' \ar[d] \\
Y \ar[r] & X
\end{tikzcd}
\]
with \(Y \to X\) a finitely presented closed immersion, \(X' \to X\) finitely presented and proper, and \(X' \setminus Y' \simeq X \setminus Y\). In this sense, the cdh topology is tailored to capture excision for blow-ups and descent for singularities, and a Nisnevich sheaf is a cdh sheaf precisely when it sends abstract blow-up squares to cartesian squares [2508.09915].

At the stable motivic level, the cdh topology does not produce a different motivic homotopy category. For every qcqs base scheme \(S\), the cdh-local motivic homotopy category \(SH_{\mathrm{cdh}}(S)\) is canonically equivalent to \(SH(S)\), so any motivic spectrum can be viewed as cdh-local and therefore satisfies cdh descent [2310.13372]. This equivalence explains why cdh descent is pervasive in motivic constructions without implying that all concrete motivic complexes on singular schemes are \( \mathbb A^1 \)-invariant.

The topos-theoretic background is also significant. For a qcqs scheme of finite valuative dimension, its cdh \(\infty\)-topos is hypercomplete, extending Voevodsky’s theorem beyond the noetherian case [2002.11647]. This hypercompleteness is a technical input in valuation-theoretic descent, point detection, and excision arguments.

## 2. \( \mathbb A^1 \)-invariant cdh-motivic cohomology

A modern \( \mathbb A^1 \)-invariant cdh theory is constructed by left Kan extending motivic cohomology from smooth schemes and then cdh-sheafifying. In the notation of Bachmann, Elmanto, Morrow, and collaborators,
\[
\mathbb Z(j)^{\mathrm{lse}} := \mathrm{Lan}(\mathbb Z(j)^A|_{\mathrm{Sm}_B}),
\qquad
\mathbb Z(j)^{\mathrm{cdh}} := L_{\mathrm{cdh}} \mathbb Z(j)^{\mathrm{lse}},
\]
with \(\mathbb Z(j)^{\mathrm{cdh}}=0\) for \(j<0\). These complexes are finitary cdh sheaves on qcqs schemes and form the graded pieces of a cdh-motivic filtration on homotopy \(K\)-theory:
\[
E_2^{i,j} = H^i_{\mathrm{cdh}}(X,\mathbb Z(-j)) \Rightarrow KH_{i-j}(X).
\]
The same work proves the comparison
\[
L_{\mathbb A^1}\mathbb Z(j)^{\mathrm{cdh}}(X) \xrightarrow{\ \sim\ } \mathbb Z(j)^{A,\mathrm{cdh}}(X),
\]
and, after base-change stability of slices, identifies \(\mathbb Z(j)^{A,\mathrm{cdh}}\) with the \( \mathbb A^1 \)-invariant motivic cohomology represented by the zeroth slice of \(KGL\) [2508.09915].

This framework realizes a version of Voevodsky’s slice-filtration program over arbitrary qcqs schemes. It also clarifies the role of cdh descent: cdh-motivic cohomology is not merely a substitute for Nisnevich descent on singular schemes, but the descent theory that makes the filtration on \(KH\) computable through blow-ups, valuation rings, and syntomic/prismatic comparison maps [2508.09915].

## 3. Non-\( \mathbb A^1 \)-invariant theories and pro-cdh descent

Elmanto–Morrow constructed a non-\( \mathbb A^1 \)-invariant motivic cohomology theory for equicharacteristic qcqs schemes by gluing the cdh-local filtration on \(KH\) to trace-theoretic filtrations on \(TC\). The weight-\(j\) complex \( \mathbb Z(j)_{\mathrm{mot}} \) is the \(j\)th graded piece of the resulting motivic filtration on \(K\), and it is a Nisnevich sheaf, finitary, and equipped with an Atiyah–Hirzebruch spectral sequence
\[
E_2^{p,q} \cong H^p(X,\mathbb Z(-q)) \Rightarrow K_{-p-q}(X).
\]
In characteristic \(0\) it fits into a pullback square involving \(R\Gamma(X,L\Omega^{\le j}_{/\mathbb Q})\), while in characteristic \(p\) it fits into a pullback square involving syntomic complexes \( \mathbb Z_p(j)_{\mathrm{syn}} \). It is explicitly not \( \mathbb A^1 \)-invariant on singular schemes; rather,
\[
L_{\mathrm{cdh}}\mathbb Z(j)_{\mathrm{mot}} \simeq \mathbb Z(j)_{\mathrm{cdh}},
\qquad
L_{\mathbb A^1}\mathbb Z(j)_{\mathrm{mot}} \simeq \mathbb Z(j)_{\mathrm{cdh}},
\]
and the theory detects nonreduced structure, for example through explicit relative calculations on \(k[x]/x^e\) [2309.08463].

The mixed-characteristic extension constructs \( \mathbb Z(i)_{\mathrm{mot}} \) for qcqs schemes by means of a global filtration on \(TC\) whose graded pieces refine derived de Rham and Bhatt–Morrow–Scholze syntomic cohomology. For noetherian schemes, these complexes satisfy pro-cdh descent: for any abstract blow-up square and any \(i \ge 0\), the induced pro-square
\[
\begin{tikzcd}
\mathbb Z(i)_{\mathrm{mot}}(X) \ar[r] \ar[d] &
\mathbb Z(i)_{\mathrm{mot}}(X') \ar[d] \\
\{\mathbb Z(i)_{\mathrm{mot}}(rY)\}_r \ar[r] &
\{\mathbb Z(i)_{\mathrm{mot}}(rY')\}_r
\end{tikzcd}
\]
is weakly cartesian [2412.06635].

A further refinement in mixed characteristic proves that the motivic complexes \( \mathbb Z(i)_{\mathrm{mot}} \) are characterized on noetherian schemes by finitariness, Nisnevich descent, pro-cdh descent, and normalization on local rings; equivalently,
\[
\mathbb Z(i)_{\mathrm{procdh}}(X) \simeq \mathbb Z(i)_{\mathrm{mot}}(X)
\]
for every noetherian \(X\) and \(i \ge 0\) [2507.16501]. This universal characterization places pro-cdh descent at the center of the theory rather than as an auxiliary excision property.

## 4. Formal schemes, pro-cdh topology, and optimal dimension bounds

Kelly–Saito introduced a pro-cdh topology on formal schemes in order to remove a defect in the earlier scheme-level pro-cdh topology. On locally noetherian formal schemes, pro-cdh coverings are generated by Nisnevich coverings and formal abstract blow-up coverings
\[
\mathcal X_Z \amalg \mathcal Y \to \mathcal X,
\]
where \(\mathcal X_Z\) is the formal completion along a closed subscheme \(Z\) and \(\mathcal Y \to \mathcal X\) is a locally algebrizable admissible proper finite type morphism, an isomorphism over \(\mathcal X \setminus Z\). The scheme-level pro-cdh \(\infty\)-topos is recovered from the formal-scheme level by imposing continuity covers \(\{\mathcal X_n \to \mathcal X\}_{n \in \mathbb N}\), corresponding to the condition
\[
F(\mathcal X)=\lim_n F(\mathcal X_n).
\]
This makes the formal theory strictly finer and better adapted to nilpotents and completions [2409.14295].

The principal theorem is an optimal homotopy-dimension bound. If \(\mathcal X\) is a formal scheme with \(\dim |\mathcal X| = d\), then
\[
\mathrm{hd}\bigl(Shv_{\mathrm{pro\mbox{-}cdh}}(fSch^{ft}_S/\mathcal X,S)\bigr) \le d,
\]
hence for any abelian sheaf \(F\),
\[
H^n_{\mathrm{pro\mbox{-}cdh}}(\mathcal X,F)=0 \quad \text{for } n>d.
\]
This remedies the earlier \(2d\) bound on the scheme-level pro-cdh \(\infty\)-topos and is obtained by comparison with a Nisnevich–Riemann–Zariski site whose inverse system of Nisnevich topoi has homotopy dimension \(\le d\) [2409.14295].

The improved bound has immediate motivic consequences. Kelly–Saito derive a topos-theoretic interpretation of Weibel’s vanishing and of Elmanto–Morrow’s motivic cohomology bounds via the descent spectral sequence
\[
E_2^{p,q}=H^p_{\mathrm{pro\mbox{-}cdh}}(X,a_{\mathrm{pro\mbox{-}cdh}}\pi_q F_{\mathrm{cont}})
\Rightarrow \pi_{-p-q}F(X).
\]
For nonconnective \(K\)-theory this yields \(K_i(X)=0\) for \(i<-\dim X\), and for Elmanto–Morrow motivic complexes over \( \mathbb Q \) or \( \mathbb F_p \) it yields
\[
H^i_{\mathrm{mot}}(X,\mathbb Z(n))=0 \quad \text{for } i>\dim(X)+n
\]
under the stated hypotheses [2409.14295]. In mixed characteristic, the same vanishing range is established in motivic form:
\[
H^j_{\mathrm{mot}}(X,\mathbb Z(i))=0 \quad \text{for } j>i+d
\]
when \(X\) is noetherian of dimension \(\le d\) [2507.16501].

## 5. Structural theorems and comparison isomorphisms

A central structural property is the projective bundle formula. For the mixed-characteristic theory, if \(E\) is a vector bundle of rank \(r+1\) on a qcqs scheme \(X\), then
\[
\bigoplus_{j=0}^{r}\mathbb Z(i-j)_{\mathrm{mot}}(X)[-2j]
\xrightarrow{\ \sim\ }
\mathbb Z(i)_{\mathrm{mot}}(\mathbb P_X(E)),
\]
with the \(j\)th summand induced by pullback and multiplication by powers of the motivic first Chern class of \(\mathcal O(1)\) [2507.16501]. Equicharacteristic and mixed-characteristic constructions also prove regular blow-up formulas, giving cartesian squares for regular closed immersions and thereby Mayer–Vietoris and Gysin-type consequences [2309.08463] [2412.06635].

The theories admit comparison maps to classical arithmetic invariants in the expected ranges. For henselian local rings \(A\), the mixed-characteristic theory proves a finite-coefficient comparison
\[
K_i^M(A)/n \to H^{i,i}(A)/n
\]
which is an isomorphism for all \(i\ge 0\) and \(n\ge 1\) [2507.16501]. For \(\ell \neq p\) invertible on \(X\), the mixed-characteristic motivic complexes satisfy a Beilinson–Lichtenbaum-type comparison
\[
\mathbb Z/\ell^k(i)_{\mathrm{mot}}(X) \xrightarrow{\ \sim\ }
\tau_{\le i}R\Gamma(X_{\acute et},\mu_{\ell^k}^{\otimes i}),
\]
and there is a corresponding \(p\)-adic comparison with syntomic complexes in degrees \(\le i\) [2412.06635].

Low weights recover familiar invariants. Weight \(0\) is identified with cdh cohomology of the constant sheaf,
\[
\mathbb Z(0)_{\mathrm{mot}}(X)\simeq R\Gamma_{\mathrm{cdh}}(X,\mathbb Z),
\]
and in weight \(1\) one has the expected identifications
\[
H^1_{\mathrm{mot}}(X,\mathbb Z(1))\cong \mathcal O_X^\times,
\qquad
H^2_{\mathrm{mot}}(X,\mathbb Z(1))\cong \mathrm{Pic}(X),
\]
together with a natural first Chern class map \(R\Gamma_{\mathrm{Nis}}(X,\mathbb G_m)[-1]\to \mathbb Z(1)_{\mathrm{mot}}(X)\) [2412.06635]. These formulas place cdh-motivic cohomology within the expected formalism of orientations, Chern classes, and projective bundle decompositions while retaining singular sensitivity outside the \( \mathbb A^1 \)-invariant regime.

## 6. Related descent theories, weight zero, and interpretive issues

One recurring misconception is that cdh-motivic cohomology is intrinsically \( \mathbb A^1 \)-invariant. The recent literature makes the opposite point: the non-\( \mathbb A^1 \)-invariant theories were introduced precisely because algebraic \(K\)-theory fails \( \mathbb A^1 \)-invariance on singular schemes, and the cdh or pro-cdh formalism supplies the blow-up descent needed to compensate for that failure rather than erase it [2309.08463]. A related misconception is that cdh descent by itself always yields optimal vanishing bounds; Kelly–Saito show that the passage to formal schemes and a refined pro-cdh topology is what repairs the earlier \(2d\) homotopy-dimension defect [2409.14295].

The cdh topology is also closely related to ldh and valuation-theoretic descent. Under explicit hypotheses \((G1)\) and \((G2)\), Kelly and Morrow prove that for a \(\mathbb Z\)-linear presheaf with traces,
\[
H^n_{\mathrm{cdh}}(S,F_{\mathrm{cdh}})
\stackrel{\sim}{\longrightarrow}
H^n_{\mathrm{ldh}}(S,F_{\mathrm{ldh}}),
\]
and they emphasize that motivic cohomology with \(\mathbb Z[1/p]\)- or \(\mathbb Z_{(l)}\)-coefficients should be regarded as invariant under universal homeomorphisms, not merely nilpotent thickenings [1807.00158]. In a complementary direction, Bhatt–Mathew–inspired results show that, for qcqs schemes of finite valuative dimension, cdh hypercompleteness and henselian valuation-ring criteria yield Milnor excision for torsion motivic spectra over fields [2002.11647].

Relative and weight-zero variants further enlarge the scope of the subject. For a closed immersion \(D \subset X\), the spectral sequence for relative homotopy \(K\)-theory has \(E_2\)-terms given by cdh-hypercohomology of relative motivic complexes, and for smooth affine pairs in top codimension it identifies the relevant relative motivic cohomology with Chow groups with modulus [1801.00922]. In weight zero, over a trivially valued field of characteristic \(0\), one has canonical isomorphisms
\[
H^n(|X^{an}|,\mathbb Z)
\cong
H^n_{\mathrm{cdh}}(X,\mathbb Z)
\cong
H^n(X,\mathbb Z(0)),
\]
so singular cohomology of the Berkovich analytification computes weight-zero motivic cohomology [2411.18274]. This result isolates the cdh topology as the precise bridge between blow-up descent and a topological realization of motivic weight zero.

Taken together, these developments show that cdh-motivic cohomology is best understood not as a single theory, but as a stratified framework. Its \( \mathbb A^1 \)-invariant branch is represented in the cdh-local motivic homotopy category and computes \(KH\)-filtrations; its non-\( \mathbb A^1 \)-invariant branch encodes singularities, nilpotents, and arithmetic comparison data; and its pro-cdh refinement supplies the formal-geometry input needed for sharp vanishing theorems, Weibel-type bounds, and universal characterizations on noetherian schemes [2508.09915] [2412.06635] [2507.16501].

Source: https://www.emergentmind.com/topics/cdh-motivic-cohomology