On representations of complex hyperbolic lattices (Q1574685)

Let \(G\) be a connected non-compact semisimple Lie group. Let \(\Gamma\) be a lattice in \(G\). It was proved by Borel and Harish Chandra that all arithmetic subgroups are lattices. Selberg and Piatetski-Shapiro in the sixties conjectured the converse for most semisimple Lie groups. The arithmeticity theorem of Margulis asserts that irreducible lattices in any simple Lie group of real rank greater than or equal to two are arithmetic subgroups. The arithmeticity of lattices in the real rank one groups \(\text{Sp}(n,1)\) and \(F^{-20}_4\) was proved in the nineties due to the work of Corlette and Gromov-Schoen. For the other two groups, namely \(\text{SO}(n,1)\) and \(\text{SU}(n,1)\), it was long known that non-arithmetic lattices do exist. In \(\text{SO}(n,1)\) such examples were constructed by Markov, Vinberg and Gromov-Piatetski Shapiro for all \(n\). In this paper the lattices in \(\text{SU}(n,1)\) are studied. The first non-arithmetic lattice in \(\text{SU}(2,1)\) was constructed by Mostow. Later, Thurston studied the moduli spaces of Euclidean cone metrics on the sphere and more non-arithmetic examples were obtained in this way. In this paper the following superrigidity type theorem for complex hyperbolic lattices is proved. Let \(X= \Gamma\setminus B^n\) be a compact complex ball quotient for \(n=2\) or 3. Suppose \(H^{1,1}(X,C)\cap H^2(X,Z)\) is generated by the Kähler class of \(X\). Then any representation of \(\Gamma\) in \(\text{GL}(n+ 1,C)\) can either be deformed to a unitary representation or be extended to a homomorphism from \(\text{SU}(n,1)\) into \(\text{GL}(n+ 1,C)\). The proof makes use of the idea of complex variations of Hodge structures.