Integral cohomology of certain Picard modular surfaces (Q2436759)

The paper describes a method for computing the integral cohomology of the locally symmetric space associated to a congruence subgroup of the Picard modular group of an imaginary quadratic number field \(k\) of class number one. The explicit computations are carried out using a computer in the cases \(k=\mathbb{Q}(i)\) and \(k=\mathbb{Q}(\sqrt{-3})\) for congruence subgroups of various levels with not too large index in the Picard modular group. For large index the computation becomes too demanding for a computer. More precisely, let \(\mathbf{G}\) be the semi-simple algebraic group defined over \(\mathbb{Q}\), which is the identity connected component of the group of isometries of a non-degenerate Hermitian form on a three-dimensional vector space over an imaginary quadratic number field \(k\) whose signature over \(\mathbb{R}\) is \((2,1)\). Let \(G=\mathbf{G}(\mathbb{R})=\mathrm{SU}(2,1)\) be the group of real points of \(\mathbf{G}\), and \(K\) a maximal compact subgroup in \(G\), so that \(\mathcal{D}=G\slash K\) is the associated symmetric space. The Picard modular group \(\bar{\Gamma}=\mathrm{SU}(2,1,\mathcal{O}_k)\) of \(k\) is defined as the stabilizer in \(\mathbf{G}(\mathbb{Q})\) of an integral lattice in the Hermitian space defining \(\mathbf{G}\). Let \(\Gamma\subset\bar{\Gamma}\) be a congruence subgroup. The goal is computing the integral cohomology of the associated locally symmetric space \(\Gamma\backslash \mathcal{D}\). In general, according to [\textit{J. Franke}, Ann. Sci. Éc. Norm. Supér. (4) 31, No. 2, 181--279 (1998; Zbl 0938.11026)], cohomology with complex coefficients of a locally symmetric space associated with a congruence subgroup is the same as the relative Lie algebra cohomology of the space of automorphic forms for the congruence subgroup. However, for the integral cohomology this method would not yield the torsion classes. The method of this paper is to construct a cell complex that allows the computation of the integral cohomology of the locally symmetric space. This cell complex, known as spine, is analogous to the well-rounded retraction of [\textit{A. Ash}, Duke Math. J. 51, 459--468 (1984; Zbl 0542.22011)]. The existence of spine for locally symmetric spaces associated to congruence subgroups of any semi-simple algebraic group of \(\mathbb{Q}\)-rank one is proved in [\textit{D. Yasaki}, Sel. Math., New Ser. 12, No. 3--4, 541--564 (2006; Zbl 1188.11026)]. In this paper a method for the computation of the spine for the locally symmetric space \(\Gamma\backslash \mathcal{D}\) is outlined in the case \(k\) is of class number one.