On page 5 of this paper by Dwyer and Mitchell, it is said that Thomason's étale descent spectral sequence from his paper Algebraic K-theory and étale cohomology, which reads

$$H^p_{\acute{e}t}(X, \mathbb{Z}_l(-q/2)) \Rightarrow \pi_{-q-p}\hat{L}KX$$

where $KX$ is the algebraic K-theory spectrum of $X$ and $\hat{L}$ denotes the $\ell$-completed Bousefield localization at topological K-theory, implies the natural isomorphism

$$\pi_i\widehat{L}(KR) \cong \pi_i\text{Map}_{\Gamma_F'}(X_+^\theta, \hat{\mathcal{K}})$$

where here, $R$ is the ring of integers localized away from $\ell$ of some totally real field $F$, $X$ is a space realizing the étale homotopy type of $R$, $\theta$ is the character of $\pi_1(X)$ corresponding to the $\ell$-adic cyclotomic character, $X^\theta$ is the cover of $X$ corresponding to the kernel of $\theta$, with the corresponding action of $\theta$ of the Galois/fundamental group. The subscript plus, as usual, denotes the unreduced suspension spectrum. The Galois group of the $\ell$-adic cyclotomic extension is in particular $\Gamma_F'$, and lives inside $\Gamma'\cong \mathbb{Z}_l^\times$ (i.e. the corresponding Galois group over $\mathbb{Q}$.)

The $\text{Map}$ is then an equivariant mapping spectrum, with the action of $\Gamma'$ via Adams operations on the target.

(It is assumed we have a rigid, i.e. not just up to homotopy, action of $\Gamma'$ on $\hat{\mathcal{K}}$.)

So, how does this follow? There is no real explanation given, which makes me think it has to be simple. But I'm confused what the relationship of the localized algebraic K-theory groups to étale cohomology has to do with this twisted equivariant mapping spectrum.

Apologies if this is some very formal thing; I'm pretty un-versed in these matters. I'll delete this question if it turns out to just be very obvious somehow.

  • $\begingroup$ I recommend in this context the recent paper of Clausen and Mathew arxiv.org/abs/1905.06611 - this has some cleaner and more general results than the original result by Thomason. (Note that $p$-completed $TC$ of a ring vanishes if $p$ is invertible in the ring.) $\endgroup$ Aug 18, 2019 at 1:36
  • $\begingroup$ thanks lennart! I actually did take a look at that paper and I think resolved my issue (see the answer below), though I cited the étale-descent of localized algebraic K-theory to thomason rather than clausen-mathew for reasons of priority. $\endgroup$
    – xir
    Aug 18, 2019 at 1:38

1 Answer 1


I think I understand the situation a little better now. Thomason's descent spectral sequence requires the base scheme to be ``$\ell$-good'', or have all the $\ell$-power roots of unity, since the point is that what is proven is that we have a weak equivalence $\hat{L}KS \simeq \hat{L}\mathbb{H}_{\acute{e}t}(S,K)$, where the latter is the localized version of Thomason's hypercohomology spectrum, with coefficients in the sheaf of algebraic K-theory spectra. This his way of constructing the upgraded version of the usual descent spectral sequences for sheaves taking values in the usual modules.

The point is that if $X$ is $\ell$-good, and a few other nice technical conditions, the étale-local stalks of the sheaf $K$ are the K-theory spectra of strict Henselian rings, so their homotopy groups agree with the algebraic K-theory of their (separably closed) residue fields, hence (by a theorem of Suslin) are locally $\mathbb{Z}_l(n)$ in degree $n$ after being hit by $\hat{L}$. (I'm not super clear why we can pass the $\hat{L}$ through the hypercohomology spectrum functor; after all, localization at $K$-theory is smashing, but $\ell$-completion is not. Maybe just passing the K-theory localization through is enough if we take periodic $K$-theory, and then we can just $l$-complete at the end?)

Since $X$ globally contains all the $\ell$-power roots of unity by the $\ell$-good hypothesis, the localized sheaf of homotopy groups in degree $n$ is actually the constant sheaf $\mathbb{Z}_l(n)$; said differently, the sheaf of spectra is weakly equivalent to the constant sheaf $\widehat{\mathcal{K}}$ of $\mathcal{l}$-completed (topological, complex) K-theory spectra. So then by applying the same descent argument using the complete topological K-theory functor, we also get a weak equivalence with $\hat{L}\text{Map}(X,\hat{\mathcal{K}})=\text{Map}(X,\hat{\mathcal{K}})$.

So, returning to the situation at hand, if $\text{Spec }R$ is just the $\ell$-inverted ring of integers of a totally real field with étale homotopy type $X$, the thing we want to do is apply Thomason's construction to the $\ell$-good cyclotomic extension $R_\infty$, whose étale homotopy type is exactly $X^\theta$. Then we find that $\hat{L}KR_\infty \simeq \hat{L}\mathbb{H}_{\acute{e}t}(R_\infty,\widehat{\mathcal{K}})\simeq \mathcal{\widehat{K}}(X^\theta)$. Étale descending, the homotopy fixed points of the LHS is our $\hat{L}KR$. (after Thomason, $K(1)$-localized algebraic K-theory does satisfy étale descent, unlike the original.)

On the RHS, to compute the homotopy fixed points, we can compute that the $\Gamma$-action on the target $\mathcal{\widehat{K}}$ is by the Adams operations, remembering that it comes originally from algebraic K-theory via Suslin's theorem. (For example, in our case, since all the residue fields are algebraic over a finite field, this is just Quillen's computation.) Hence the descended version is just precisely our equivariant mapping space $\text{Map}_\Gamma(X^\theta, \mathcal{\widehat{K}})$.

I'm definitely not an expert on these matters, so please point out if there are glaring errors or missing pieces!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.