24

In his paper Fonctions L p-adiques, Pierre Colmez says:

Tate a montré qu'il n'existait pas dans $\mathbf C_p$ d'analogue $p$-adique de $2i \pi$ et donc par conséquent que les périodes $p$-adiques des variétés algébriques ne pouvaient pas vivre dans $\mathbf C_p$.

(Tate showed that there does not exist in $\mathbf C_p$ a $p$-adic analogue of $2i\pi$, and therefore that the $p$-adic periods of algebraic varieties cannot live in $\mathbf C_p$.)

He further explains that this realization led Fontaine to construct his complicated 'rings of periods'. He gives, as reference for the quoted claim, Tate's paper p-divisible groups. However, nothing that I've read in Tate's paper seems to immediately justify this surprising claim. I've read a bit about Fontaine's rings and I have an idea of the role they play in studying $p$-adic representations, but I'm not quite sure how to formally express "that there does not exist in $\mathbf C_p$ a $p$-adic analogue of $2i\pi$".

My feeling is that he is thinking about periods on curves, where the formal residue theorem of Serre-Tate allows us to explicitly describe Riemann-Roch-Poincaré duality in terms of a formal integration pairing. Over $\mathbf C$, this pairing can be calculated analytically by Cauchy's Residue Theorem, but it requires "division by $2i \pi$"... and somehow, we can't mimick this inside $\mathbf C_p$? is this right? if so, why can't we mimick it?

I would welcome any clarification or explanation of this matter. Thanks!

Edit: This user comment on the blog post linked to by anon, seems to suggest that my hunch is correct. So how can we justify this claim, and get a good feel for the necessity of Fontaine's rings? And, how is this related to Tate's paper?

Bruno Joyal
  • 52,320
  • 6
  • 122
  • 220
  • 4
    Related blog post on SBS: http://sbseminar.wordpress.com/2009/02/18/there-is-no-p-adic-2-pi-i/. (My understanding of most of the content here is zero; the most I can do is quote Strassman's theorem or say that continuous periodic functions out of ${\Bbb C}_p$ are locally constant.) – anon Sep 30 '13 at 23:23
  • Thanks @anon, that is very helpful. – Bruno Joyal Sep 30 '13 at 23:30

1 Answers1

17

I think Colmez is referring to the fact that $H^0\bigl(G_{\mathbb Q_p},\mathbb C_p(1)\bigr) = 0.$

Any such invariant would be be a $p$-adic period for the cyclotomic character; but as Tate showed, this space of invariants vanishes, and so the $p$-adic cyclotomic character does not have a period in $\mathbb C_p$; one has to go to $B_{\mathrm{dR}}$ to get a period (the famous element $t$ of Fontaine).

Matt E
  • 118,032
  • 11
  • 286
  • 447
  • 1
    Dear Matt, thank you for your answer! Would you mind elaborating on why such an invariant would be a $p$-adic period for the cyclotomic character? Regards, – Bruno Joyal Oct 02 '13 at 05:06