Cosmological Constraints from Non-Gaussian Features in Weak Gravitational Lensing Maps
Weak gravitational lensing measures the distortion of the images of distant galaxies, as the light from these galaxies passes by massive galaxy clusters on its way to the observer on Earth. The path of the light gets bent due to the gravity of the foreground cluster, which makes this technique directly sensitive to all matter present, including dark matter. With upcoming wide-field surveys like the Large Synoptic Survey Telescope (LSST), weak lensing will be able to substantially tighten our constraints on cosmological parameters - if we learn how to extract this information properly and get systematic errors under control. In this talk I will describe our efforts to do this from ray tracing of cosmological N-body simulations, from which we extract information beyond the power spectrum (i.e. from the non-Gaussian part due to nonlinear clumping of gravity) and deal with systematics like magnification bias, baryonic physics, and instrument errors. On the technical side, this will take us from cosmology to big-data supercomputing and analysis with numerical/statistical issues like interpolation on an irregular grid using Gaussian processes, which are of general interest beyond the focus of this research.