Scientific Discovery Through Advanced Computing (SciDAC-4)
High-energy particle physics has entered a new era with the discovery of the Higgs boson at the Large Hadron Collider. If hints for a solution to the hierarchy problem and the particle origin of dark matter are to be discovered, the analysis of scattering data, in particular those from the Large Hadron Collider (LHC), will require ever more theoretical precision and experimental accuracy. Monte-Carlo event generators are used to make the theoretical predictions, based on a perturbative expansion of the underlying quantum field theory, and based on the treatment of renormalization group predictions in a particle interpretation using Markovian Monte-Carlo methods. Both types of calculations profit from large-scale computing resources, but the use of leadership class facilities is hampered by the reliance on adaptive Monte-Carlo methods that require a large amount of communication during their adaptation stage. We are therefore invested in the design of a new software package to redesign optimization algorithms for improved scaling, and to devise strategies for preconditioning integrators such as to better incorporate the known behavior of the integrand in special regions of the phase space. We also work on the construction of improved event processing frameworks.