Back to Volume
Paper: Likelihood Simulation for Bounding Non-Gaussianity in the CMB
Volume: 371, Statistical Challenges in Modern Astronomy IV
Page: 421
Authors: Jennings, K.; Treaster, A.; Cayon, L.
Abstract: The distribution of the CMB temperature fluctuations predicted by several alternative models to the Standard Inflationary can be parametrized by the so-called fNL parameter. We present a simulation-based scheme which uses a classification algorithm to designate the most likely parameter value. Methods for parameter space search, calibration, and importance sampling can dramatically increase accuracy and efficiency. Confidence intervals are constructed using a version of simulation hypothesis testing. Output of the search algorithm can be interpolated to provide reasonably accurate estimates of bounds; a further search can be used to provide more accurate placement.
Back to Volume