Accelerating the analysis of large astrophysical datasets (Astronomy)
Data mining in large astrophysical data sets to better understand the nature of our universe and various structures within it: The astrophysics community regularly convenes national panels to evaluate the state of the field and to develop clear research priorities. On the recommendations of these review panels, a significant amount of theoretical and numerical progress has been made over the past few years in developing more accurate models for how galaxies form and evolve.
The most effective approach to observationally constrain galaxy formation models is to calculate correlation functions, which quantify the level of excess clustering over a random distribution. In spite of the existence of an entire family of correlation functions, known as n-point correlation functions, the vast majority of published analyses utilize only the lowest order function-the two-point correlation function (the Fourier Transform of the more widely-known Power Spectrum). Historically this is due to the computational complexity of calculating correlation functions. Naively, the complexity of an n-point function scales as O (N^n), where N = sources (e.g., galaxies) being analyzed. We propose to work with undergraduate students to leverage advanced parallel computational facilities and novel computational accelerators to calculate the full suite of n-point correlation functions that will enable a precise characterization of the dark matter and dark energy in the Universe.