Measuring large scale structure using angular cross-correlations

Student thesis: Doctoral thesis


In this thesis we propose to use galaxy clustering, more concretely angular cross-correlations, as a tool to understand the late-time expansion of the Universe and the growth of large-scale structure. Galaxy surveys measure the position of galaxies (what traces the dark-matter field) in spherical coordinates (z,θ,φ). Most galaxy clustering analyses convert these positions to distances assuming a background cosmology. This approach thus requires doing the full data analysis for each background cosmological model one wants to| test. Instead we propose to select galaxies in radial shells, according to their redshifts, and then measure and analyze the angular (2D) correlations in each bin circumventing the model assumption. On the one hand our approach projects and looses 3D information along the line-of-sight for distances smaller than the shell width. On the other hand, it allows a single analysis, as no cosmological model needs to be assumed. Remarkably we find that if we include in the analysis also the angular cross-correlations between different shells, we can recover the radial modes corresponding to the separations between radial bins. We found that the optimal binning to recover 3D information is given by the largest between the minimum scale considered for spatial clustering, 2π/kmax, and the photometric redshift error. Photometric galaxy surveys, such as Physics of the Accelerating Universe (PAU) and Dark Energy Survey (DES), access higher number densities and higher redshifts than current spectroscopic surveys, at the price of loosing radial accuracy. Angular analysis in redshift bins is then the natural framework for such surveys. We found that, for such photometric surveys, the constraints on the growth index of structure improve by a factor two when we include the cross-correlations. In addition, we show that one can use two different galaxy populations to trace dark matter and hence reduce sample variance errors. The cross-correlations of both populations in the same field leads to an overall gain of a factor five. This allows measurements of the growth rate of structure to a 10% error at high redshifts, z > 1, complementing low-z results from spectroscopic surveys. This gain is maximized for high bias difference and high densities. We also worked with N-body simulations to include non-linear gravitational effects and turn them on and off (e.g redshift space distortions or the radial distortions produced by photometric redshifts). We built galaxy survey mocks from the MICE simulations and measure galaxy clustering to compare with our previously mentioned models of angular correlations. We found a good agreement between theory and simulation measurements. In the future, we expect to apply this framework for cosmological parameter estimation, especially focusing on DES and PAU surveys.
Date of Award27 Sept 2013
Original languageEnglish
Supervisor Héctor Martín Crocce (Director) & Enrique Fernandez Sanchez (Tutor)


  • Cosmology
  • Large scale structure
  • Angular clustering

Cite this