Abstract: State-of-the-art galaxy redshift surveys will provide the most detailed mapping of the universe and the tightest constraints on the cosmological parameters. However, the huge amount of data also presents a computational challenge. For
instance, it may take several months to produce thousands of simulations to estimate the covariance matrix which measures the uncertainty of our measurements. Furthermore, our constraints on the cosmological parameters are weakened because of the noises and biases in the simulation. In this talk, I will present my new work on using data compression and analytical models to dramatically reduce the computational cost and noise associated with this procedure. I validated my approach with the largest and most constraining state- of-the-art galaxy surveys and found computing the analytical covariance matrix
for these surveys can be done within a day while simulations will generally take several months. The data compression reduces the length of the data vector and speeds up the likelihood evaluation by a factor of two. Furthermore, I find even without knowing the best-fit parameters a priori, the analytical covariance matrix returns similar constraints on the cosmological parameters to the simulated covariance matrix. My method can significantly speed up the analysis for future surveys such as DESI and Euclid.