The infinite fern in higher dimensions

Abstract

If $\bar\rho$ is an automorphic modulo $p$ Galois representation, it is natural to wonder if automorphic points are Zariski dense in the deformation space of $\bar\rho$. We prove new results in this direction in the case of a unitary group split (and unramified) at $p$. Namely, if $\bar\rho$ is associated to an automorphic form for a unitary group (which contributes to coherent cohomology), we prove that the "infinite fern" (i.e. the image of an appropriate Eigenvariety) in the polarised deformation space of $\bar\rho$ is Zariski dense in a non-empty union of irreducible components. This generalises in particular results of Gouv\^ea-Mazur for $GL_2/\mathbb Q$, Chenevier for $U(3)$ and recently Hellmann-Margerin-Schraen. The novelty is that we use the local model of Breuil-Hellmann-Schraen to control tangent spaces in the local deformation rings, and a geometric argument on the Eigenvariety originally due to Bella\"iche-Chenevier and Ta\"ibi to reduce to points with enormous image. At those points, we can use a recent result of Newton-Thorne to control the vanishing of a Selmer group. In particular, we do not need to assume any "Taylor-Wiles" hypothesis on $\bar\rho$, which can in particular be irreducible. If we moreover add Taylor-Wiles hypothesis on $\bar\rho$ and an extra hypothesis at $p$, we have by a result of Allen the Zariski density everywhere.