The paper tackles a central question in modern number theory: when is a certain type of mathematical object called a Galois representation "modular," meaning it secretly encodes information from a classical object called a modular form? This connection, part of the Langlands program, underlies famous results like the proof of Fermat's Last Theorem. The specific representations studied here are two-dimensional, defined over totally real number fields, and have a residual representation (a simplified version modulo a prime) that splits into two simpler pieces. The tricky feature is that these representations behave differently at different primes above p: reducible (splitting into pieces) at some and irreducible (not splitting) at others. This "mixed case" had not been handled by previous landmark work of Pan and Zhang, which required uniform behavior everywhere.
The authors prove modularity in this mixed setting under a key condition: the representation must be non-split at every prime above p where it is reducible. The strategy combines two sophisticated tools. The first is Zhang's pro-modularity theorem, which provides a foundation for lifting modularity from residual to full representations. The second is a classicality criterion using Emerton's theory of ordinary parts, a technique from p-adic representation theory that helps verify when a p-adic object genuinely comes from a classical modular form. Together, these tools allow the authors to bridge the gap left by previous methods.
The results have several concrete applications. They establish modularity for lambda-adic representations attached to certain abelian varieties, which are higher-dimensional generalizations of elliptic curves. They also cover potentially modular representations of parallel weight two under mild conditions, as well as representations whose local behavior at primes above p looks like a discrete series, a natural class in the theory. Additionally, they handle certain cases over real quadratic fields involving a technical condition called Steinberg type. Overall, the paper makes meaningful progress on the Fontaine-Mazur conjecture, one of the guiding frameworks for understanding which Galois representations come from geometry and automorphic forms.