I have two questions related to the stable motivic homotopy categories of Morel-Voevodsky. The first is probably simple; I wonder what is known on the second one.

  1. For the algebraic cobordism theory $MGL$ and a smooth variety $X$ over a (perfect?) field is it true that $MGL^{2n+i,n}(X)=0$ for any $n\in \mathbb{Z},i>0$? More generally, are there any reasonable restrictions on a (oriented?) ring spectrum $E$ in $SH$ that ensure the vanishing of
    $MGL^{2n+i,n}(E)$. In particular, is this question related with some sort of effectivity for spectra?

  2. It is well known that 'shifts and twists' of the spectra $\Sigma(X_+)$ generate $SH$, where $X$ runs through all smooth $k$-varieties. If the characteristic of $k$ is $0$, resolution of singularities yields that it suffices to consider only smooth projective varieties here. Now, what statements of this sort are known for $k$ of characteristic $p>0$? I suspect that that one can deduce a similar result for $SH\otimes \mathbb{Z}_{(l)}$ for any prime $l\neq p$, ffrom the Gabber's l'-alterations theorem. Is this true? If this is too difficult, can one prove a similar statement for the triangulated category of $MGL$-modules?

What are the best references for these questions?


2 Answers 2


(1) is true if $char(k)=0$. This follows from a combination of results. First of all, it is true over any field that the spectrum $MGL$ is connective, which means that


if $p>q+dim(X)$, $X\in Sm/k$ [1, Cor. 2.9]. (Slightly more is true: for any $p\geq q+dim(X)$, the orientation map $MGL\to H\mathbb{Z}$ induces an isomorphism $MGL^{p,q}(X)\cong H\mathbb Z^{p,q}(X)$ [1, Lem 6.4].)

Second, we have the Hopkins-Morel equivalence [1, Thm. 6.11]

$$MGL/(x_1,x_2,\dots) \simeq H\mathbb Z.$$

Assuming this, Spitzweck has shown in [2] that that the slices of MGL are given by

$$s_rMGL\simeq \Sigma^{2r,r}H(MU_{2r}),$$

Moreover, he gives an explicit description of the $r$-effective cover $f_rMGL$ as a homotopy colimit of spectra of the form $\Sigma^{2i,i}MGL$ for $i\geq r$, which shows that $f_rMGL$ is also $r$-connective (being a homotopy colimit of $r$-connective spectra). Because the homotopy $t$-structure is right complete [1, Cor. 1.4], this implies that


Now, let $E=\Sigma^{-p,-q}\Sigma^\infty X_+$ with $p>2q$. Take any map $E\to MGL$. Since $H^{p+2r,q+r}(X,A)=0$ for any abelian group $A$ and $r\in\mathbb Z$, this map lifts through all stages of the slice filtration, hence comes from a map $E\to \mathrm{holim}_{r\to\infty}f_rMGL=0$. QED.

(Incidentally, this shows that there is a strongly convergent spectral sequence $H^{\ast\ast}(X,MU_{2\ast})\Rightarrow MGL^{\ast\ast}(X)$.)

If $char(k)>0$ ($k$ need not be perfect), the Hopkins-Morel equivalence is also known if $char(k)$ is inverted [1], so we can at least deduce that $MGL^{p,q}(X)$ is $char(k)$-torsion for all $p>2q$, $X\in Sm/k$ (and it is zero if $p>q+dim(X)$ by connectivity, so for fixed $X$ and $q$ at most finitely many of these groups can be nonzero).

Some comments about (2): if you go through the proof of the characteristic zero case in [3] and try to use Gabber's theorem instead of resolution of singularities, at some point in the proof an isomorphism is replaced by a finite flap map $f: Y\to X$ of degree prime to a given prime $l\neq p$, and the proof will work if that map has a section. Even if you work $\mathbb Z_{(l)}$-locally, you still need a map $g: X\to Y$ such that $fg=deg(f)\cdot\mathrm{id}$, and I don't see why you'd have such a map in $SH\otimes\mathbb Z_{(l)}$. But for $MGL_{(l)}$-modules I guess that the Gysin map [4] should work.

[1] M. Hoyois, From algebraic cobordism to motivic cohomology (pdf)

[2] M. Spitzweck, Relations between slices and quotients of the algebraic cobordism spectrum (pdf)

[3] O. Röndigs, P. Østvær, Modules over motivic cohomology (pdf)

[4] F. Déglise, Around the Gysin triangle II (pdf)


It was proved by Riou in Appendix B of http://arxiv.org/abs/1311.2159 that the spectra of smooth projective varieties do (compactly) generate $SH(k)_{\mathbb{Z}_{(l)}}$ for any $l$ distinct from $\operatorname{char} k$.

Note also that $SH(k)_{\mathbb{Z}_{(l)}}$ "differs a little" from $SH(k)\otimes {\mathbb{Z}_{(l)}}$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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