In that paper https://web.math.rochester.edu/people/faculty/doug/mypapers/hkr.pdf Hopkins-Kuhn-Ravenel introduced the idea of generalized character corresponding to a complex-oriented cohomology theory $E^*$. They construct an $E^*$-algebra $L(E^*)$, which is defined to be the colimit: ${\rm colim} E^*(B\mathbf{Z}^n_p)$ (More details regarding the whole construction can be found in the link.)

Later on, they use this construction to define the invariant ring $L(E^*)^{{\rm Aut}(\mathbf{Z}^n_p)}$, where ${\rm Aut}(\mathbf{Z}^n_p)$ acts as $E^*$-algebra homomorphisms ($\mathbf{Z}_p$ denotes the additive group of $p$-adic integers). To prove this they define $L_r(E^*)=E^*(B\mathbf{Z}_{p^r})$, and the natural ${\rm Aut}(\mathbf{Z}_{p^r})$ action on gives $L_r(E^*)^{{\rm Aut}(\mathbf{Z}_{p^r})}=p^{-1}E^*$.

My question has to do with the proof of the above: What I understand is that the invariant rings $L_r(E^*)^{{\rm Aut}(\mathbf{Z}_{p^r})}=p^{-1}E^*$, induce a direct system of $E^*$-algebras and the colimit must be $L(E^*)^{{\rm Aut}(\mathbf{Z}^n_p)}$. However, they don't give any proof hence should be somehow straightforward why the colimit is the invariant ring $L(E^*)^{{\rm Aut}(\mathbf{Z}^n_p)}$ (not at all to me). Can you explain me please if my understanding makes sense? If yes, probably an explanation why the above colimit converges on the invariant ring $L(E^*)^{{\rm Aut}(\mathbf{Z}^n_p)}$ would be really helpful. If not, a sort of insight would be very appreciable!


I think what you understand is correct. Write $A_r= L_r(E^*)$, which fit into a direct system $A_r\to A_{r+1}\to \cdots$. Let $G=\mathrm{Aut}(\mathbb{Z}_p^n)$, which acts compatibly on every $A_r$ (it actually acts on $A_r$ through the quotient group $\mathrm{Aut}((\mathbb{Z}/p^r)^n)$).

If $\cdots \to A_r\to A_{r+1}\to \cdots$ is a sequence of injective maps between $G$-sets, then it is straightforward to show that $\mathrm{colim} (A_r^G) \to (\mathrm{colim} A_r)^G$ is an isomoprhism. In fact, you only need them to be injective for all $r\geq R$ for some $R$.

It remains to show that the $A_r\to A_{r+1}$ are injective. I don't see a proof of this in the paper, but it is surely true.

Here's a proof that $A_r\to A_{r+1}$ is injective if $r\geq1$. There's probably a better proof, but it's what comes to mind now. I use the description in the proof of [HKR, 6.5]: $$ A_r = S_r^{-1} E^*B\Lambda_r, $$ where $\Lambda_n=(\mathbb Z/p^r)^n$, and $S_r\subseteq E^*B\Lambda_r$ is a certain multiplicatively closed subset, which can be defined as follows: $$ S_r = \{ c(\alpha):=(B\alpha)^*(x)\; | \; \alpha\in \Lambda_r^*\smallsetminus\{0\} \},\qquad \Lambda_r^*:=\mathrm{Hom}(\Lambda_r, U(1)), $$ where $x\in E^*BU(1)$ is a chosen coordinate of the formal group. Although the set $S_r$ depends on the choice of $x$, the fraction ring $A_r$ doesn't, because any two coordinates differ by a unit in $E^0BU(1)$. Note that the direct system of $A_r$ comes from the inverse system $\cdots\to\Lambda_{r+1}\to \Lambda_r\to\cdots$

Note that $c(\alpha^p)=[p](c(\alpha))= c(\alpha)f(c(\alpha))$ where $[p](x)$ is the $p$-series of the formal group and $f(x)=[p](x)/x$ is a power series. This implies that if we invert $c(\alpha^p)$ then we automatically invert $c(\alpha)$ as well. For every $\alpha\in \Lambda_r^*\smallsetminus\{0\}$ with $r\geq1$, some $\alpha^{p^k}$ is in the image of $\Lambda_1^*\to \Lambda_r^*$, so in fact $$ A_r = S_1^{-1}E^*B\Lambda_r, $$ where $S_1$ really means the image of $S_1\subset E^*B\Lambda_1$ under the map induced by $\Lambda_r\to \Lambda_1$. Since $E^*B\Lambda_r\to E^*B\Lambda_{r+1}$ is injective and $S_1^{-1}E^*B\Lambda_1$ is flat over $E^*B\Lambda_1$, the claim follows.

  • $\begingroup$ Thanks for your effort Charles. I haven't read your answer through yet. Though I have a question from the very beginning. How would you explicitly describe that: $G$ acts compatibly on every $A_r$ through the quotient group ${\rm Aut}((\mathbf{Z}/p^r)^n)$? $\endgroup$ – user430191 Oct 21 '18 at 17:24
  • $\begingroup$ My problem isn't the context in that paper. But rather, how you can write things up more explicitly. I understand that many things have been swept under the rug for simplicity. But it's not entirely obvious to me how this action can be written? $\endgroup$ – user430191 Oct 21 '18 at 17:27
  • $\begingroup$ We have an action of $G=GL_n(\mathbb{Z}_p)$ on $\Lambda=\mathbb{Z}_p^n$, which passes to an action on every quotient $\Lambda_r=\Lambda/p^r\Lambda$. Everything comes from that. $\endgroup$ – Charles Rezk Oct 21 '18 at 19:40
  • $\begingroup$ Charles, thanks for the answer. It was very illuminating indeed. A last question only. How you come up with "injective" morphisms $E^*B\Lambda_r \to E^*B\Lambda_{r+1}$? $\endgroup$ – user430191 Oct 22 '18 at 10:05

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.