If $X$ and $Y$ are two spectra, I denote by $F(X,Y)$ their mapping spectrum. This is uniquely determined by the existence of a natural isomorphism $[X\wedge Y, Z]\cong [X,F(Y,Z)]$.

I denote by $H_*$ the rational homology functor. If $X$ is a finite spectrum, I have an isomorphism $H_*(F(X,Y))\cong Hom(H_*(X),H_*(Y))$. Indeed, each side is a generalized cohomology theory in the $X$ variable that takes the value $H_*(Y)$ on the sphere. I would like to extend this result to a general $X$. What I can do is write a general spectrum $X$ as a filtered homotopy colimit of spectra $X_i$ that are finite. Then $F(X,Y)$ is the inverse limit of the spectra $F(X_i,Y)$. I get a Milnor exact sequence:

$$0\to \mathrm{lim}_i^1Hom_{n-1}(H_{*}(X_i),H_{*}(Y))\to H_nF(X,Y)\to \mathrm{lim}_i Hom_n(H_*(X_i),H_*(Y))\to 0$$

where $Hom_n$ denotes the set of homomorphisms of degree $n$.

I want to look at an example of the previous exact sequence with $X=\bigvee_{\mathbb{N}}S^0$ and $Y=S^0$. I can write $X$ as the homotopy colimit of $X_i=\bigvee_{1\leq k\leq i}S^0$. Then, $F(X,Y)$ is equivalent to $(S^0)^{\mathbb{N}}$, thus $H_0(F(X,Y))\cong \mathbb{Q}\otimes\mathbb{Z}^{\mathbb{N}}$. On the other hand, we have $$\mathrm{lim}_i\;Hom_0(H_*(X_i),H_*(Y))\cong \mathrm{lim}_i\;\mathbb{Q}^i\cong\mathbb{Q}^{\mathbb{N}}$$ The obvious map $\mathbb{Q}\otimes\mathbb{Z}^{\mathbb{N}}\to\mathbb{Q}^{\mathbb{N}}$ is not even surjective !

My question has two parts:

(1) Explain where the mistake is in the previous calculation. My guess is that I misunderstood the Milnor exact sequence.

(2) Is there a method for computing $H_*(F(X,Y))$ knowing $H_*(X)$ and $H_*(Y)$ ?

  • @NeilStrickland. Thanks, this fully answers my question. If you want to write it as an answer, I will accept it. Are there easy to verify conditions on $X$ that insure that $H_*F(X,Y)\cong Hom(H_*X,H_*Y)$ ? – Geoffroy Horel Mar 19 '15 at 11:11
up vote 6 down vote accepted

This was a comment, but at the OP's suggestion I have promoted it to an answer.

There is no Milnor exact sequence for homology groups of an inverse limit, only for the homotopy groups. If $I$ is the Brown-Comenetz dual of the sphere, then $H_∗(I)=0$ but $F(I,I)$ is the profinite completion of S and so $H_∗F(I,I)≠0$. Thus, you cannot hope for a method that works in complete generality.

Since you are doing everything rationally and stably, homotopy and homology are the same.


$H_* F(X,Y) = [X,Y]_* \otimes \mathbb{Q}$

If $Y$ is itself a rational spectrum, then this is the same as

$\text{Hom}_{\mathbb{Q}}(H_* X, H_* Y)$

but of course in general it will not be. Said another way,

$\text{Hom}_{\mathbb{Q}}(H_* X, H_* Y) = H_*F(X,LY)$

where $LY$ is the rational localization of $Y$.

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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