On the geometric Serre weight conjecture for Hilbert modular forms

Let $p$ be a prime, $F$ be a totally real field in which $p$ is unramified and $\rho: \mathrm{Gal}(\overline{F}/F)\rightarrow \mathrm{GL}_2(\overline{\mathbb{F}}_p)$ be a totally odd, irreducible, continuous representation. The geometric Serre weight conjecture formulated by Diamond and Sasaki can be viewed as a geometric variant of the Buzzard-Diamond-Jarvis conjecture. They introduce the notion of geometric modularity in the sense that $\rho$ arises from a mod $p$ Hilbert modular form, and algebraic modularity in the sense that $\rho$ arises in the mod $p$ cohomology of a Shimura curve. Diamond and Sasaki conjecture that $\rho$ being geometrically modular of weight $(k,l)\in \mathbb{Z}^\Sigma_{\geq 2}\times\mathbb{Z}^\Sigma$ is equivalent to $\rho$ being algebraically modular of the same weight, if $k$ lies in the minimal cone, where $\Sigma$ is the set of embeddings from $F$ into $\overline{\mathbb{Q}}$. In this thesis, we prove that geometric modularity implies algebraic modularity for real quadratic fields $F$ in which $p \geq 5$ is inert, and for totally real fields $F$ in which $p \geq \max\{5, [F:\mathbb{Q}]\}$ totally splits.