Serre weights for \(\mathrm{U}(n)\) (Q683512)

The classical Serre's conjecture for a continuous Galois modular representation \[{\bar r}: {\mathrm{Gal}}( \bar{\mathbb Q}/{\mathbb Q})\rightarrow \mathrm{GL}_2({\mathbb F}_p)\] predicts the set of weights \(W_{\mathrm{mod}}(\bar r)\) of cuspidal modular eigenforms that give rise to \(\bar r\). This conjecture has various generalizations both to various number fields and higher rank groups. The authors of the paper have already obtained remarkable results in this subject. In the current paper they consider representations: \[{\bar r}: {\mathrm{Gal}}( \bar{\mathbb Q}/{\mathbb Q})\rightarrow \mathrm{GL}_n({\mathbb F}_p)\] and their modular weights with respect to automorphic forms on the compact unitary group \(\mathrm{U}(n)\). Let \({\bar{\rho}}= {\bar r}|_{\mathrm{Gal}}({\mathbb{Q}_p}/{\mathbb Q})\) be the restriction of \(\bar r\) to the decomposition group at \(p\). The authors introduce the following sets of weights: \(W^{\mathrm{cris}}(\bar{\rho})\) -- the set of Hodge-Tate weights of crystalline lifts of \({\rho}\), \(W^{\mathrm{diag}}(\bar{\rho})\) -- the set of Hodge-Tate weights of potentially diagonalizable lifts of \({\rho}\), \(W^{\mathrm{obv}}(\bar{\rho})\) -- the set of obvious weights defined combinatorially from \({\rho}\). One of main results of the paper (Theorem 4.1.9) describes the set of weights of \(\bar r\) from the mod \(p\) reduction of algebraic Weyl modules with highest weight in \(W^{\mathrm{diag}}(\bar{\rho})\) and gives a lower bound in the complexity of modular weights for \(\bar{\rho}\). This result relies on the following inclusions \[W^{\mathrm{obv}}(\bar{\rho})\subset W^{\mathrm{diag}}(\bar{\rho})W^{\mathrm{cris}}(\bar{\rho}).\] Another main theorem of the paper characterizes the modular weights for \(\bar r : G_F \rightarrow \mathrm{GL}_3({\mathbb F}_l)\), an irreducible Galois representation and \(F\) an imaginary CM field. One assumes that if \(F^+\) is a maximal real subfield of \(F\) then the extension \(F/F^+\) is unramified at all finite places and \(l\) splits completely in \(F\). One also assumes existence of the automorphic representation \({\Pi}\) of \(\mathrm{GL}_3({\mathbb A}_F)\) satisfying an appropriate set of assumptions.