# Failure of local Fontaine Mazur

This question unfortunately has a very similar name to this one, but I what want to ask here is different.

Let $$K$$ be a finite extension of $$\mathbb{Q}_p$$. It seems to be well known that the local version of the Fontaine-Mazur conjecture is false, in the sense that if $$\rho:Gal(\overline{K}/K)\rightarrow GL_d(\mathbb{Q}_p)$$ is some de Rham representation, it may very well be that $$\rho$$ does not come from a variety defined over a $$p$$-adic field. For instance, this is claimed by Mark Kisin in his survery paper "What is... a Galois representation?".

Maybe this question is naive, but can anyone give me a concrete example of such a representation?

Also, a slightly unrelated question: does one expect the answer to this version of the local Fontaine-Mazur conjecture to change if we allow representations coming from the etale cohomology of analytic spaces?

• Take a crystalline representation. If it comes from a global object, the eigenvalues of Frobenius should be algebraic numbers (over the rationals). So just choose a crystalline representations not having this last property. – Laurent Berger Sep 12 at 9:12
• Dear @LaurentBerger, thank you for your comment. In the case of a variety defined over $\mathbb{Q}_p$, but not necessarily over $\mathbb{Q}$, would it still be true that the eigenvalues are algebraic over $\mathbb{Q}$ (and not just over $\mathbb{Q}_p$)? (Note in my question above I asked also about representations coming from varieties defined over $\mathbb{Q}_p$, as in Kisin's article alluded to above) – xlord Sep 13 at 11:45