Algorithms for classification of astronomical object spectra

摘要

Obtaining interesting celestial objects from tens of thousands or even millions of recorded optical-ultraviolet spectra depends not only on the data quality but also on the accuracy of spectra decomposition. Additionally rapidly growing data volumes demandes higher computing power and/or more efficient algorithms implementations. In this paper we speed up the process of substracting iron transitions and fitting gaussian functions to emission peaks utilising C++ and OpenCL methods together with the NOSQL database. In this paper we implemented typical astronomical methods of detecting peaks in comparison to our previous hybrid methods implemented with CUDA.