The classical definition of Hilbert modular cuspforms as given in say, Hida's "On $p$-adic Hecke algebras for $\mathrm{GL}_2$ over totally real fields", defines them as holomorphic functions $f: \mathbb{H}^d \to \mathbb{C}$ satisfying the usual moebius transformation properties with respect to a chosen level subgroup and weight. (Here $\mathbb{H}$ is the usual upperhalf plane, and $d=[F:\mathbb{Q}]$ with $F$ your totally real field).

There are also, some slight variants of this construction (as can also be found in Hida's paper), where one picks some subset $J$ of the set of embeddings of $F$ into $\mathbb{C}$ and then require that these functions be holomorphic in the $J$ components and anti-holomorphic in the $J^c$ components (here ${}^c$ is complement). Let me call them $J$-holomorphic. These show up for example in the Eichler--Shimura isomorphism for Hilbert modular forms.

Now, I know the holomorphic forms can be seen as the global sections of some sheaf on some Hilbert modular variety (a la Katz). But what about these $J$-holomorphic ones? Are they $H^0$s of some sheaf? or maybe $H^k$s for some $k>1$ of the modular sheaf? If its the second option then is $k=|J|$?



1 Answer 1


Your second guess is spot on: these $J$-holomorphic forms are exactly $H^{k}$ of a sheaf on (a smooth compactification of) the Hilbert modular variety, where $k = |J^c|$. This is an instance of a much more general theory which is largely due to Harris. The canonical reference is

Harris, Michael. Automorphic forms of $\bar{\partial}$-cohomology type as coherent cohomology classes. J. Differential Geom. 32 (1990), no. 1, 1--63.

  • $\begingroup$ Thanks, that is what I hoped for. $\endgroup$ Commented Jun 17, 2020 at 8:01

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.