Putzka, Jens Frederik Bernhard: A Toolbox to Compute the Cohomology of Arithmetic Groups in Case of the Group Sp2(Z). - Bonn, 2013. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.
Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5n-31490
@phdthesis{handle:20.500.11811/5645,
urn: https://nbn-resolving.org/urn:nbn:de:hbz:5n-31490,
author = {{Jens Frederik Bernhard Putzka}},
title = {A Toolbox to Compute the Cohomology of Arithmetic Groups in Case of the Group Sp2(Z)},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2013,
month = mar,

note = {

It is the aim of this thesis to present a toolbox of methods which can be used to compute the group cohomology Hq(Sp2(Z), Mλ) of the Siegel modular group Sp2(Z) for a given integer q ≥ 0 and a highest weight module Mλ with respect to a weight λ. Many tools we introduce in this thesis can be applied with minor changes also to groups different from the symplectic group. By some spectral sequence argument we get under some week assumptions an isomorphism

Hq(Sp2(Z), Mλ) ≅ Hq(S2/Sp2(Z), Mλ)

between group cohomology of Sp2(Z) and sheaf cohomology of the sheaf Mλ over S2/Sp2(Z), which is associated to the module Mλ. Here S2 denotes the Siegel upper half-space. The cohomology groups behave very differently for different choices of λ. In particular, those cases are of special interest in number theory, e.g. for Harder's conjecture on eigenvalues of Hecke operators, which have non-vanishing so called cusp cohomology. This happens for the first time for the highest weight module of weight λ=(7,4), which has rank 1820 over the integers, which is not really small.
There are three main branches in this thesis:
Topological Model: It is known by a result of Mark McConnell and Robert MacPherson from the late eighties that there is a Γ'-equivariant deformation retract W of S2 for a neat arithmetic subgroupΓ'⊂Sp2(Z), which has the structure of a regular cell complex, and descends to a 4-dimensional retract W/Γ' of the 6-dimensional space S2/Γ', which has again a cell decomposition. Our first main result is the generalization of this model to the torsion case. To do this we have to replace cells by new objects called orbicells, which are the orbifold equivalent to cells in manifolds, and have very similar properties. We obtain a retract W/Γ' of S2/Γ' for any subgroup Γ'⊂Sp2(Z) of finite index. Each of these retracts has an orbicell decomposition, which derives from the cell decomposition of W. We implemented a computer program in Sage which computes various things related to these decompositions, e.g. closures, stabilizers, and neighbours of cells. We illustrate up to some level - drawing in 4-D is a little bit ambitious - how the building blocks look like.
Highest Weight Modules: It is essential for an efficient computation of the cohomology to be able to perform various actions with and on highest weight modules. Most things are quite simple if a basis is known. However it turned out that the computation of the action of a group element on the module is not that easy. Therefore, our second main result is an algorithm to compute this action. This involves the decomposition of a given group element into generic generators coming from some roots via the morphism between Lie algebras and Lie groups. To obtain the needed decomposition we introduce a structured Gaussian elimination which preserves the symplectic structure.
Compute Cohomology: There is the notion of constructible sheaves due to Alexander Grothendieck and his collaborators, which describes a category of sheaves which are locally constant restricted to objects which cover (in Zariski-sense) a variety. We were able to generalize this category to a category of sheaves on the obtained orbicell decomposition. This construction is quite beautiful, and works in a much more general context. Therefore, we present parts of it not only for W or W/Γ', but for an abstract space with a suitable decomposition and an action of a group. We use this language to get an abstract description of the cohomology groups we are looking for, which later could be used to compute the cohomology.
Finally we discuss the application of this construction to the orbicell decomposition of the deformation retract W/Sp2(Z) of the Siegel modular variety S2/Sp2(Z) and indicate what has to be done to complete the computation of the cohomology of the symplectic group.

},

url = {https://hdl.handle.net/20.500.11811/5645}
}

The following license files are associated with this item:

InCopyright