Reconstructing the Sky with CMB Lensing methods

Abstract

Current and future high significance CMB lensing-galaxy cross-correlations will soon precisely probe the growth of structure and provide powerful tests for parameter tensions, non-Gaussianity, neutrino mass, and modifications to Λ−CDM and gravity. However, such cross-correlations are sensitive to CMB foreground contamination, and the resulting biases represent an important challenge for cosmological analyses. In this thesis we describe new CMB lensing cross-correlation measurements and a novel multi-frequency cleaning-based technique to mitigate foreground effects in lensing maps. In particular, we present a tSZ cleaned cross-spectrum between Atacama Cosmology Telescope lensing and BOSS galaxies using a modified version of the standard quadratic estimator that retains most of the signal-to-noise. To minimize lensing foreground biases, geometry-based methods that modify the form of the standard quadratic estimator have been developed as well. We discuss how to optimally combine these with multi-frequency cleaned data to mitigate the impact of foreground contamination while maintaining maximal significance. For a Simons Observatory-like experiment, we find that for the CMB lensing auto-spectrum it is possible to reduce the foreground induced bias by a large factor, compared to the standard quadratic estimator, at a modest noise cost. We find a similar result for a cross-correlation with an LSST-like sample, with a large reduction in bias at small noise cost. Finally, we turn to the three-dimensional large scale structure of the Universe probed by galaxy surveys. We show how we can reconstruct the large-scale density field from galaxy modes using CMB lensing-inspired methods and use this reconstruction to improve constraints on local primordial non-Gaussianity.