The universal unramified module for \(\mathrm{GL}(n)\) and the Ihara conjecture (Q2045157)

Let \(p\) be a prime number. Let \(F\) be a finite extension of \(\mathbb{Q}_p\). Let \(k\) be an algebraically closed field of characteristic \(\ell>0\) and \(\ell\neq p\). Let \(\mathcal{\mathcal{O}}_F\) be the ring of integers of \(F\). Let \(G=\mathrm{GL}_n(F)\) and \(K=\mathrm{GL}_n(\mathcal{O}_F)\). The space \(k[G/K]\) of finitely supported functions on \(G/K\) is naturally a module over the Hecke algebra \(k[K\backslash G/K]\). Let \(\lambda: k[K\backslash G/K]\rightarrow k\) be a homomorphism. The universal unramified module for \(\lambda\) is the \(k[G]\)-module defined by \(\mathcal{M}_{\lambda,k}:=k[G/K]\otimes_{k[K\backslash G/K],\lambda}k.\)\par The paper under review proves several interesting results for \(\mathcal{M}_{\lambda,k}\). Under the Satake isomorphism the morphism \(\lambda:k[K\backslash G/K]\rightarrow k\) gives rise to characters \(\chi:T\rightarrow k^{\times}\) where \(T\) is the diagonal torus of \(G\). The paper proves a conjecture of Lazarus that \(\mathcal{M}_{\lambda,k}\) shares the same semisimplification with the unramified principal series induced from \(\chi\). To prove this result, the author studies universal unramified modules \(W_{\lambda,R}\), replacing the coefficient ring \(k\) by \(W(k)\)-algebras \(R\). The author proves that \(\mathcal{M}_{\lambda,R}\) is flat over \(R\) for any \(\lambda:W(k)[K\backslash G/K]\rightarrow R\) and uses the known comparisons of the universal unramified modules and unramified principal series in characteristic \(0\).\par To apply to the Ihara conjecture concerning the genericity of certain representations generated by mod-\(\ell\) automorphic forms, the author compares \(\mathcal{M}_{\lambda,k}\) with the Whittaker space \(\mathrm{Ind}_U^G\psi\) where \(U\) is the unipotent radical of the standard Borel subgroup of \(G\) and \(\psi\) is a fixed non-degenerate character of \(U\) extended from an additive character \(\psi:F\rightarrow k^{\times}\) in the usual way. For each \(\lambda\), the author defines a canonical map \(\mathcal{M}_{\lambda,k}\rightarrow \mathrm{Ind}_U^G\psi\). The main theorem of the paper states that this map \(\mathcal{M}_{\lambda,k}\rightarrow \mathrm{Ind}_U^G\psi\) is an injection. The proof uses induction on \(n\), and involves the study of Jacquet modules of both \(\mathcal{M}_{\lambda,k}\) and \(\mathrm{Ind}_U^G\psi\). These results allow the author to give equivalent easier statements of the Ihara's lemma in [\textit{L. Clozel} et al., Publ. Math., Inst. Hautes Étud. Sci. 108, 1--181 (2008; Zbl 1169.11020)].