A long time ago, where were the galaxies far, far away?
Abstract
How did the universe get from then to now ? I examine this broad cosmological problem from two perspectives: forward and backward.
In the forward perspective, I implement a method of generating initial conditions for N -body simulations that accurately models real-space statistical properties, such as the mass variance in spheres and the correlation function. The method requires running ensembles of simulations because the power in the DC mode is no longer assumed to be zero. For moderately sized boxes, I demonstrate that the new method corrects the previously widely ignored underestimate in the mass variance in spheres and the shape of the correlation function. In the backward perspective, I use reconstruction techniques to transform a simulated or observed cosmological density field back in time to the early universe. A simple reconstruction technique is used to sharpen the baryon acoustic peak in the correlation function in simulations. At z = 0.3, one can reduce the sample variance error bar on the acoustic scale by at least a factor of 2 and in principle by nearly a factor of 4. This has significant implications for future observational surveys aiming to measure the cosmological distance scale. Another reconstruction technique, Monge-Ampere-Kantorovich reconstruction, is used on evolved N -body simulations to calibrate its effectiveness in recovering the linear power spectrum. A new "memory model" parametrizes the evolution of Fourier modes into two parameters that describe the amount of memory a given mode retains and how much the mode has been scrambled by nonlinear evolution. Reconstruction is spectacularly successful in restoring the memory of Fourier modes and reducing the scrambling; however, the success of reconstruction is not so obvious when considering the power spectrum alone. I apply reconstruction to a volume-limited sample of galaxies from the Sloan Digital Sky Survey and conclude that linear bias is not a good model in the range 0.01 h Mpc -1 [Special characters omitted.] k [Special characters omitted.] 0.5 h Mpc -1 . The most impressive success of reconstruction applied to real data is that the confidence interval on the normalization of the power spectrum is typically halved when using the reconstructed instead of the nonlinear power spectrum.- Publication:
-
Ph.D. Thesis
- Pub Date:
- 2007
- Bibcode:
- 2007PhDT.........5S
- Keywords:
-
- Galaxies;
- N-body simulations;
- Initial conditions;
- Baryon bump