Simulation of single-particle diffraction experiments - Requirements for hardware and software

by Duane Loh (SLAC)

Friday, December 9, 2011 from to (Europe/Berlin)
at AER 19, Rm. 3.11
High-resolution, 3D single-particle diffraction imaging relies on the assembly of many noisy, unoriented 2D diffraction data from copies near-identical particles into a compatible 3D intensities distribution. This talk is meant to highlight the various computational challenges that arise from various statistics that bear on this imaging workflow. I will attempt to capture the effects of these statistics with simplified models. I will then describe an algorithm (the Expansion-Maximization-Compression algorithm) that demonstrably succeeds in assembling the 3D intensities under considerable levels of noise. The conditions on this success are then related to realistic constraints on future single-particle imaging experiments.