On the Semisimplicity of Pure Sheaves

L Fu
DOI: https://doi.org/10.1090/s0002-9939-99-05414-3
IF: 0.9709
1999-01-01
Proceedings of the American Mathematical Society
Abstract:We obtain a criteria for a pure sheaf to be semisimple. As a corollary, we prove the following: Let X 0 X_0 and S 0 S_0 be two schemes over a finite field F q \mathbf {F}_q , and let f 0 : X 0 → S 0 f_0: X_0\rightarrow S_0 be a proper smooth morphism. Assume S 0 S_0 is normal and geometrically connected, and assume there exists a closed point s s in S 0 S_0 such that the Frobenius automorphism F s F_s acts semisimply on H i ( X s ¯ , Q l ¯ ) H^i(X_{\bar s}, {\overline {\mathbf {Q}_l}}) , where X s ¯ X_{\bar s} is the geometric fiber of f 0 f_0 at s s (this last assumption is unnecessary if the semisimplicity conjecture is true). Then R i f 0 ∗ Q l ¯ R^if_{0\ast } {\overline {\mathbf {Q}_l}} is a semisimple sheaf on S 0 S_0 . This verifies a conjecture of Grothendieck and Serre provided the semisimplicity conjecture holds. As an application, we prove that the galois representations of function fields associated to the l l -adic cohomologies of K 3 K3 surfaces are semisimple. We also get a theorem of Zarhin about the semisimplicity of the Galois representations of function fields arising from abelian varieties. The proof relies heavily on Deligne’s work on Weil conjectures.
What problem does this paper attempt to address?