Given an abelian variety $A$ over a base scheme $\text{Spec } \mathcal{O}_{K_p}$, we define the functor $P$ as taking $A \mapsto \text{colim}_n A[p^n]$, its associated $p$-divisible group. What is the essential image of $P$?

Further, take the functor $\circ$, which takes a $p$-divisible group $B[p^\infty]$ to its associated formal group $B[p]^\circ$ (the connected component of the $p$-divisible group, under Tate's categorical equivalence of connected $p$-divisible groups and formal Lie groups). What is the essential image of $\circ$?

In other words, how much of $p-Div$ and $FG$ can we touch via the study of abelian varieties?

  • 2
    $\begingroup$ What do you mean when you write $K_p$? $\endgroup$ Commented Nov 21, 2017 at 22:23
  • $\begingroup$ I mean some extension of Q_p $\endgroup$ Commented Nov 22, 2017 at 4:26

1 Answer 1


In short, I still don't know over a general ring, but for $\mathbb{F}_q$, we have an answer: The idempotent completion of the essential image of $AbVar_{/\mathbb{F}_q}$ in $FG_{/\mathbb{F}_q}$ is categorically equivalent to $FG_{/\mathbb{F}_q}$.

In Manin's paper, The Theory of Formal Commutative Groups over Fields of Finite Characteristic, he calls formal groups "algebroid" if they are the completion of algebraic varieties over any field $k$. He proves that every formal group over $\mathbb{F}_q$ is isogenous to a subgroup of an algebroid formal group over $\mathbb{F}_q$. The proof outline is:

  1. Every formal group is isomorphic to $\oplus_i G_{n_i,m_i}$ up to isogeny.
  2. Every such $G_{n,m}$ (pick any coprime m, n) can be realized in the formal group decomposition associated to an abelian variety.

Notation: $G_{n,m}$ is the formal group whose Dieudonné $W(\mathbb{F}_p)$-module is isomorphic to $W(\mathbb{F}_p)/(F^n - V^m)$.

For step 2, he uses the Jacobian variety $J_a$ associated to the family of curves $y^p - y = x^{p^a-1}$ over $\mathbb{F}_{p^a}$, where $a = 1, 2, 3, ...$. The decomposition of the formal group $\hat{J}_a$ contains all $G_{n, m}$ such that $m+n \mid a(p-1)$. So, for any pair of coprime $(n, m)$ and any $p$, $\hat{J}_{n + m}$ contains a piece isogenous to $G_{n,m}$.

(There is the subtlety that we are working in $FG$ up to isogeny rather than isomorphism here, so this answer only tells us about the category $FG$ up to isogeny.)

A note on the difference in base field from my original question: I asked about $\mathcal{O}_{K_p}$ where $K_p$ is a finite extenstion of $\mathbb{Q}_p$, whereas Manin considers $\mathbb{F}_q$. It seems that since the varieties $A$ are defined over finite extensions of $\mathbb{Q}_q$, if they have good reduction we can extend them (via the Néron model?) to $\mathbb{Z}_q$, and take their special fiber to get to varieties defined over $\mathbb{F}_q$.


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.