15
$\begingroup$

The usual disclaimer applies: I'm new to all this stuff, so be gentle.

It seems like the spectrum, as defined by Balmer, of the stable homotopy category of finite complexes is something like $M_{FG}$, the stack of formal groups (that is, $Spec L/ G$ where $L$ is the Lazard ring and $G$ acts by coordinate changes). I'm not actually sure if that's true, I don't think I've seen it written quite like that, but the picture of the spectrum in Balmer's paper looks an awful lot like how I'd imagine $M_{FG}$ looking.

If the above is right, then there's another tensor triangulated category with the same spectrum, namely the derived category of perfect complexes on $M_{FG}$ (whatever that means for stacks...).

So my question is:

Just how far away is the stable homotopy category from actually being equivalent to this derived category? Is there a theorem to the effect that it can't be equivalent to such a thing? Do we even know that it's not equivalent?

I've heard that chromatic homotopy theory is about setting up a rough dictionary between algebro-geometric terminology regarding $M_{FG}$ and the stable homotopy category, so I guess the question is about whether or not we can make the dictionary into a proper functor.

$\endgroup$
5
  • $\begingroup$ Uh oh... I'm realizing that this $M_{FG}$ doesn't make much sense unless I pick a field... should I rephrase the question by localizing the stable homotopy category at a prime and picking a field of characteristic $p$? Or am I being paranoid? $\endgroup$ Aug 19, 2011 at 7:13
  • $\begingroup$ Section 2.4.2 of Morava's Complex cobordism and algebraic topology says something about this and provides references for sub-questions. Here's an arXiv link: arxiv.org/abs/0707.3216 . $\endgroup$ Aug 19, 2011 at 8:18
  • 1
    $\begingroup$ The Hopf map is an obstruction for the existence of an algebraic model for the stable homotopy category $\endgroup$ Aug 19, 2011 at 9:16
  • 2
    $\begingroup$ (You don't need to pick a field, the moduli of formal groups makes sense over any base.) $\endgroup$ Aug 19, 2011 at 13:05
  • $\begingroup$ @Fernando: I was thinking about this... I know there are no Hopf maps in things like $D^{perf}(X)$ when $X$ is a scheme, but I wasn't sure if this still held for stacks $\endgroup$ Aug 19, 2011 at 15:54

3 Answers 3

23
$\begingroup$

One useful thing to keep in mind is that the cohomological functor from the stable homotopy category to the category of quasi-coherent sheaves on the moduli stack $\mathcal{M}$ is not essentially surjective. For example, if you fix a prime $p$ and a height $n \geq 1$, then there is a closed substack $\mathcal{M}^{\geq n}$ consisting of formal groups over $\mathbb{F}_p$ having height $\geq n$. A standard problem in stable homotopy theory is to try to cook up finite spectra which map to the structure sheaf of $\mathcal{M}^{\geq n}$. You can generally only do this when $p$ is large compared with $n$. For small values of $p$ you generally have to make do with finite spectra whose image is the structure sheaf of some nilpotent thickening of $\mathcal{M}^{\geq n}$. These can always be found (a deep result of Devinatz-Hopkins-Smith) and this is what gives you such a strong connection between the topology of $\mathcal{M}$ and the "spectrum" of the stable homotopy category. But you have to work hard for it, and the connection is much weaker (closed subsets of $\mathcal{M}$ have an interpretation in the stable homotopy category, rather than closed substacks) than what you would expect if Adams-Novikov spectral sequences were to degenerate.

$\endgroup$
1
  • 1
    $\begingroup$ Ah, thank you! Just to make sure I know what you're talking about: The finite spectra mapping to the structure sheaf of $\mathcal{M}^{\ge n}$ would be these $V(n)$-type spectra, yeah? And a finite spectrum with image the structure sheaf of a nilpotent thickening of $\mathcal{M}^{\ge n}$ would be the ones constructed in the proof of the periodicity theorem? This language is wonderful, conceptually! $\endgroup$ Aug 19, 2011 at 16:01
24
$\begingroup$

The relationship between the derived category of $M_{FG}$ and the stable homotopy category $\mathcal{S}$ is somewhat like the relationship between $D(gr_I(R))$ and $D(R)$, where $R$ is a commutative ring, and $I$ is an ideal in $R$, and $gr_I(R)=\bigoplus_nI^n/I^{n+1}$ is the associated graded ring. Complex cobordism gives a homological functor from $\mathcal{S}$ to the abelian category of quasicoherent sheaves on $M_{FG}$, but there is no useful functor from $\mathcal{S}$ to the associated derived category $D(M_{FG})$. Computationally, the morphism groups in $\mathcal{S}$ are the target of an Adams-Novikov spectral sequence whose $E^2$ page can be described as morphism groups in $D(M_{FG})$.

$\endgroup$
1
  • $\begingroup$ This answer is great! I was torn between all of them... so I went with the popular vote. $\endgroup$ Aug 19, 2011 at 16:05
22
$\begingroup$

It's definitely known that the derived category of ${\cal M}_{FG}$ and the stable homotopy category are not equivalent. This is an instance of

The Mahowald Uncertainty Principle: Any spectral sequence converging to the homotopy groups of spheres with an $E_2$-term that can be named using homological algebra will be infinitely far from the actual answer.

(The naming is due to Ravenel; this quote is from Paul Goerss' "The Adams-Novikov Spectral Sequence and the Homotopy Groups of Spheres".) There is often a feeling that stable homotopy theory always deviates from algebra as soon as is possible.

As Neil said, the Adams-Novikov spectral sequence starts with morphisms in the derived category and computes stable homotopy groups of spheres. Every place where this spectral sequence does not degenerate indicates a point where the stable homotopy category deviates from the algebraic approximation. This includes the following phenomena.

  • Hidden additive extensions, such as the hidden additive extension making $\pi_3^s$ into $\mathbb{Z}/24$ rather than $\mathbb{Z}/12 \times \mathbb{Z}/2$.

  • Hidden multiplicative extensions. In the (2-local) stable homotopy groups there are elements $\eta \in \pi_1^s$, $\nu \in \pi_3^s$, and $\sigma \in \pi_7^s$. My recollection is that such that $\eta^2 \sigma = \nu^3$ on the $E_2$-term, but Toda showed that this relationship doesn't hold on-the-nose in stable homotopy groups of spheres.

  • Differentials. For any prime $p$, there is always a nontrivial differential in the Adams-Novikov spectral sequence, and the first differential is called the Toda differential.

$\endgroup$

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.