Knowledge Extraction from Huge Astronomical Data Sets using Massively Parallel Processing

Abstract

The aim of thesis is survey of the field of processing large data via parallel accesses and subsequent implementation of one algorithm and its parallelization on the chosen architecture. My task is to create an application that will be able to synthesize the model light spectrum of Be star, using this expression to calculate the necessary values for comparison with spectra of unknown astronomical objects. For the purposes of synthesis I use analytic programming, a method of symbolic regression, using evolutionary algorithms. Comparison of spectra is implemented on a multicore processor architecture.