The method for ab initio investigations of QCD is Lattice QCD, as originally proposed by Kenneth Wilson in 1974. It is based on the path integral formulation of QCD in Euclidean space-time, which allows the usage of Monte-Carlo methods in analogy to statistical physics.
Lattice QCD allows for the computation of quantities that are not accesible to perturbation theory. Among others these include hadron masses and matrix elements. Lattice QCD can for instance contribute important inside to our understanding of the CKM matrix and flavour physics.
Since Lattice QCD has a strong demand for computer resources a naturally connected research direction is high performance computing and the development of algorithms. Many of the projects carried out in this group for instance use the Blue Gene installation at FZ Jülich.