3
$\begingroup$

Could anyone epxlain to me how to identify the residue field of $\overline{\mathbb{Q}}_p$ with $\overline{\mathbb{F}}_p$ ? I am basically trying to go from a $p$-adic galois representation of $G_\mathbb{Q}$ to a representation of the type Serre uses...so my ultimate goal is to obtain a representation $$ \rho: \operatorname{Gal}(\overline{\mathbb{Q}}/\mathbb{Q}) \rightarrow \operatorname{GL}_2(\overline{\mathbb{F}}_p) $$ from a more natural one like $\rho: \operatorname{Gal}(\overline{\mathbb{Q}}/\mathbb{Q}) \rightarrow \operatorname{GL}_2(\mathbb{Q}_p)$.

I was initially looking for an argument involving choosing a lattice in $V = \overline{\mathbb{Q}}_p^2$, but since then I read (and understood the proof that) the residue field of $\overline{\mathbb{Q}}_p$ is an algebraic closure of $\mathbb{F}_p$. Would that be an easier way to solve my problem?

Thank you very much !

$\endgroup$
  • $\begingroup$ I didn't understand, residue field of $\overline{\mathbb{Q}}_p$with respect to what? Recall this field is not complete with respect to the $p$-adic valuation. $\endgroup$ – Vinicius M. Nov 6 '13 at 23:20
  • $\begingroup$ @ViniciusM. For the residue field we take the valuation subring modulo its unique maximal ideal. Why would completeness be necessary? $\endgroup$ – anon Nov 6 '13 at 23:22
  • $\begingroup$ @anon you're right, I misunderstood the concepts. tks $\endgroup$ – Vinicius M. Nov 6 '13 at 23:26
  • $\begingroup$ Ultimately I would like to understand how to get a representation $\rho:G_\mathbb{Q} \rightarrow \operatorname{GL}_2(\overline{\mathbb{F}}_p)$. If I am correct that identifying those two fields will help me then yes :) However if you have a clever way to construct $\rho:G_\mathbb{Q} \rightarrow \operatorname{GL}_2(\overline{\mathbb{F}}_p)$ I would be extremely happy! $\endgroup$ – user62423 Nov 6 '13 at 23:29
  • 3
    $\begingroup$ @user62423: Dear user, You may want to learn about the theory of unramified extensions in more detail. For every finite extension $k$ of $\mathbb F_p$, there is a corresponding uniquely determined unramified extension of $\mathbb Q_p$ with residue field $k$ (whose ring of integers is often denoted $W(k)$, and called the ring of Witt vectors of $k$). This idea/construction is fundamental in the theory of $p$-adic Galois representations for lots of different reasons, so it's good to get it straight. Best wishes, $\endgroup$ – Matt E Nov 7 '13 at 12:21
5
$\begingroup$

The residue field of $\overline{\mathbf{Q}}_p$ is algebraically closed because every monic polynomial over the ring of integers $\overline{\mathbf{Z}}_p$ of $\overline{\mathbf{Q}}_p$ has a root in $\overline{\mathbf{Z}}_p$, and it is algebraic over $\mathbf{F}_p$ because every element in the ring of integers over $\overline{\mathbf{Z}}_p$ is integral over $\mathbf{Z}_p$. So the residue field of $\overline{\mathbf{Q}}_p$ is an algebraic closure of $\mathbf{F}_p$. It doesn't completely make sense to say that you want to "identify" the residue field with $\overline{\mathbf{F}}_p$, because there isn't really one algebraic closure of $\mathbf{F}_p$; any two are isomorphic, but non-canonically.

Regarding how to take a Galois representation and reduce it modulo $p$, you have to prove that $G_\mathbf{Q}$ preserves an $\overline{\mathbf{Z}}_p$-lattice in a finite-dimensional $\overline{\mathbf{Q}}_p$-vector space on which it acts continuously. This is easier to see I guess for a representation on a finite-dimensional vector space $V$ over a finite extension $L$ of $\mathbf{Q}_p$. Start with any $\mathscr{O}_L$-lattice $T$ in $V$ (i.e. a finite $\mathscr{O}_L$-module that spans $V$ over $L$) and let $\Gamma$ be its stablizer in $\mathrm{GL}_n(L)$. This is an open subgroup of $\mathrm{GL}_n(L)$, so the inverse image $\Gamma^\prime$ is an open subgroup of $G_\mathbf{Q}$. Now take the sum of the translates of $T$ by the finitely many coset representatives for $G_\mathbf{Q}/\Gamma^\prime$. This is still a lattice, and it is $G_\mathbf{Q}$-stable by construction.

What this means concretely is that you can conjugate $\mathrm{GL}_n(L)$ so that the image of $G_\mathbf{Q}$ is contained in $\mathrm{GL}_n(\mathscr{O}_L)$. Now you apply the natural reduction map $\mathrm{GL}_n(\mathscr{O}_L)\rightarrow\mathrm{GL}_n(\mathbf{F})$, where $\mathbf{F}$ is the residue field of $L$.

The reason you don't have to really think about representations over $\overline{\mathbf{Q}}_p$ when you do this is that the image of any continuous homomorphism $G_\mathbf{Q}\rightarrow\mathrm{GL}_n(\overline{\mathbf{Q}}_p)$ has image in $\mathrm{GL}_n(L)$ for some $L$ finite over $\mathbf{Q}_p$. This is a consequence of the Baire category theorem and works with $G_\mathbf{Q}$ replaced by any profinite group. I guess maybe the argument I gave above for the existence of a stable lattice still works with $\overline{\mathbf{Q}}_p$, but I worry about whether a sum of translates of a finite free $\overline{\mathbf{Z}}_p$-module spanning a finite-dimensional $\overline{\mathbf{Q}}_p$-vector space will still be free...anyway, you don't need it because you can always descend to a finite extension of $\mathbf{Q}_p$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Amazing answer ! Thank you very much for taking the time to write that, it clears up more than my original question ! $\endgroup$ – user62423 Nov 7 '13 at 8:22

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.