Berkeley Lab Scientific Computing Seminar

The purpose of our multi-scale, multi-physics, three-dimensional parallel codes and methods development is to make the step from qualitative to quantitative predictions. The simulation codes require the accurate three-dimensional modeling of large and complicated accelerator structures including space charge, beam lines, collimation, and, in the future, secondary effects.

Besides all this we are faced with the data challenge and the resulting difficulties in analyzing large data sets. I will present our attempts with respect to the data challenge by first reviewing the H5Part project and speculate on the future of general parallel data models in the context of particle accelerator simulation.