3
$\begingroup$

In this paper, Quillen mentioned a spectral sequence as follows.

Let $f:X\to Y$ is a poset map and $\mathcal{F}:X\to Ab$, where $Ab$ is the category of abelian groups, a functor which is contravariant i.e., $x\leq y$ meaning there is an arrow from $y\to x$ and the functor induces morphism $\mathcal{F}(x)\to \mathcal{F}(y)$. For $y\in Y$, define the fibers $$f/y=\{x\in X~|~f(x)\leq y\}.$$ Then there is a Grothendieck spectral sequence $H_i(Y, y\mapsto H_j(f/y,\mathcal{F})) \Longrightarrow H_{i+j}(X,\mathcal{F})$.

I was expecting a similar spectral sequence for cohomology. In order to obtain such, I take $f:X\to Y$ as above and $\mathcal{G}:X\to Ab$ a covariant functor then I can define an induced functor which is also covariant on $Y$ and it maps $y\mapsto H^k(f/y,\mathcal{G})$. My expectation is to get $$H^i(Y,y\mapsto H^j(f/y,\mathcal{G}))\Longrightarrow H^{i+j}(X,\mathcal{G}).$$

Now in order to get such a convergent spectral sequnece I must expect the following to be true. Let $$Ab^X\xrightarrow{L_1} Ab^Y\xrightarrow{L_2} Ab,$$ where $Ab^X$ denotes the category of covariant functors from $X\to Ab$, $L_1$ is induced from $f$ and $L_2$ is $\lim_{\leftarrow Y}$ and the composition must be $\lim_{\leftarrow X}$ such that $L_1$ takes injective objects in $Ab^X$ to $L_2$-acyclic objects. However, being new in this field, I am not very confident whether this is going to work.

I could not find any reference so far where such a spectral cohomology sequence for posets is mentioned. My questions are:

Am I going in the right direction? Should I obtain a cohomology spectral sequence of posets with functor coefficient of such kind out of this and if not why? If such a thing is mentioned (i.e., Gorthendieck spectral sequence of cohomology for posets with functor coefficient) some where in the literature can anyone suggest me a reference?

I apologize for asking too many questions at the same time!!

$\endgroup$
1
  • 1
    $\begingroup$ In cohomology, the formula you want is OK if you consider contravariant functors (the idea is that cohomology simply is homology with values in the opposite category of the category of abelian groups). If you want covariant functors, the $E_2$ term of your spectral sequence should be a cohomology group of the poset $\{y\leq f(x)\}$. This is a very standard computation that dates back to the 60's. This is what is explained in the introduction of this paper (p. 197): numdam.org/item/AMBP_2003__10_2_195_0 $\endgroup$ Commented Nov 21 at 22:45

0

You must log in to answer this question.