Let $E = E_n$ be the $n$-th Morava $E$-theory and let $\{ M_{I} \}$ be a tower of generalised Moore spectra. Then (see this previous question) there is a Milnor exact sequence

$$0 \to \varprojlim_I {}^1 (E \wedge M_I)_{\ast+1}(X) \to E_*^\vee X \to \varprojlim_I (E \wedge M_I)_*X \to 0. $$

Here we write $E_*^\vee X$ for $\pi_*L_{K(n)}(E \wedge X)$.

The functor $\mathcal{K}_*X := \varprojlim_I (E \wedge M_I)_*X$ is sometimes called the Morava module of $X$, and seems to have first been introduced by Hopkins, Mahowald and Sadofsky whilst studying the $K(n)$-local Picard group.

In Hopkins-Mahowald-Sadofksy, under the condition that $\mathcal{K}_*X$ is finitely-generated, the authors show that the $E_2$-term of the $K(n)$-local $E_n$-based Adams spectral sequence can be given by $H_c^\ast(\mathbb{G}_n,\mathcal{K}_*X)$, where $\mathbb{G}_n$ is the $n$-th (extended) Morava stabilizer group.

It is known that the $\varprojlim{}^1$ term of the Milnor exact sequence vanishes under suitable conditions; for example if $E_*^\vee X$ is finitely-generated, or pro-free, as an $E_*$-module. In these cases then $E^\vee_*X \simeq \mathcal{K}_*X$. I suspect that the above spectral sequence should have $E_2$-term $H_c^\ast(\mathbb{G}_n,E^\vee_*X)$. This leads to the following question:

If $\mathcal{K}_*X$ is finitely-generated is there an isomorphism $\mathcal{K}_*X \simeq E^\vee_*X$?

By the comments above it would suffice to show that $E^\vee_* X$ is finitely-generated. Hovey-Strickland Proposition 8.6 gives a number of equivalent conditions, including that $X$ is $K(n)$-locally dualisable or that $K_*X$ is finite.


Here are some initial steps, which may be useful.

First, consider $A=L_{K(n)}(E\wedge X)$, which is a $K(n)$-local $E$-module with $\pi_*A=E^\vee_*X$. It is not hard to see that $A\wedge M_I=E\wedge X\wedge M_I$. Given this, we see that we have a short exact sequence $$ \lim_I{}^1\pi_{*+1}(A\wedge M_I) \to \pi_*(A) \to \lim_I\pi_*(A\wedge M_I), $$ and we want to know whether the middle term is finitely generated, assuming that the last term is. All this makes sense for an arbitrary $K(n)$-local $E$-module $A$, and I suspect that it is easiest to work in that generality. In particular, we can replace the operation $A\mapsto A\wedge M_I$ with $A\mapsto A\wedge_EE/I$. This is advantageous because we can construct $E/I$ for any ideal of the form $(u_0^{i_0},\dots,u_{n-1}^{i_{n-1}})$, and it is easy to analyse the maps between these objects.

Next, recall the category of $L$-complete $E_0$-modules described in Appendix A of Hovey-Strickland. Results given there suffice to show that all $E_0$-modules arising in any plausible approach to the above problem will be $L$-complete. Note in particular that if $P$ is $L$-complete then the map $P=L_0P\to\widehat{P}=\lim_IP/IP$ is always surjective.

I am not sure exactly where to go from here, but I would suggest doing the case $n=1$ first. There we need to show that if $\lim_k\pi_*(A/p^k)$ is finitely generated, so is $\pi_*(A)$. The obvious approach is to compare the tower $\{\pi_*(A/p^k)\}$ with the towers $\{\pi_*(A)/p^k\}$ and $\{\text{ann}(p^k,\pi_{*-1}(A))\}$.


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.