Date of Award

Fall 10-1-2021

Document Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Padmanabhan, Nikhil


The study of the cosmological large-scale structure seeks to understand the makings and the evolution of the universe. In this subject, I worked on improving current techniques and their application to the existing large, high-precision cosmological data sets. Specifically, my dissertation explores boosting power spectrum measurements at large scales for 21-cm intensity maps through reconstruction, and at small scales for Ly α forest by developing and applying the optimal estimator to hundreds of high-resolution spectra.The cosmic tidal reconstruction is a novel technique for low redshift (z < 2) 21-cm intensity mapping surveys (e.g. CHIME and HIRAX) that recovers the lost large-scale line-of-sight signal from local small-scale anisotropies formed by tidal interactions. My thesis shows this algorithm is robust against redshift space distortions and can recover the signal with approximately 70% efficiency for k < 0.1 h Mpc−1 using N-body simulations. If angular modes are also lost (known as the foreground wedge), the efficiency drops down to 30–50% range. I further introduce an analytical framework based on perturbation theory, which correctly predicts the shape of the reconstructed field’s 2D power spectrum and reveals that the reconstruction mostly utilizes angular modes with k > 0.3 h Mpc−1. Through absorption lines in quasar spectra, the Ly α forest technique can probe matter in vast volumes far into the past (2 < z < 5) and at smaller scales than galaxy surveys (r <1 Mpc). The power spectrum at these scales is shaped by the thermal state of the intergalactic medium (IGM), reionization history of the universe, neutrino masses and the nature of the dark matter. 1D power spectrum P1D has emerged as a competitive framework to study new physics, but has come with various challenges and systematic errors in analysis. I implement the optimal quadratic estimator for P1D and generate synthetic spectra based on The Dark Energy Spectroscopic Instrument (DESI) specifications. Using these mock spectra, I prove robustness against relevant problems including quasar continuum errors and gaps in spectra due to bad pixels or masked high column density absorbers, show that an input fiducial power spectrum improves the accuracy, and provide simple 5-yr forecasts for DESI P1D measurements. I also apply the optimal estimator to the largest number of high-resolution, high-S/N spectra, obtained by combining Keck Observatory Database of Ionized Absorption toward Quasars (KODIAQ), The Spectral Quasar Absorption Database (SQUAD) and XQ-100 data sets. This project yields the most precise P1D measurement at small scales and should improve the mass of warm dark matter constraints by more than a factor of 2.