Period search in large datasets
Alex Schwarzenberg-Czerny · Open Astronomy · 1998
The number of papers on the analysis of unevenly sampled time series is scarce.The present article is an attempt to provide a fairly simple introduction to this topic.We start with the demonstration that some procedures for analysis of unevenly sampled time series, such as the power spectrum, suffer from a number of faults and traps which make them unreliable in practice.Then we consider the application of orthogonal models in statistics and testing of statistical hypotheses.Next, we demonstrate, how these classical principles of statistics can be adapted to the analysis of unevenly sampled time series.In this way we derive new, reliable methods for analysis of unevenly sampled time series.We discuss the relevant statistics, i.e. periodogram functions and their performance, and provide tools for planning efficiency of time series observations.These methods should be particularly useful for astronomers, since astronomical time series are often sampled unevenly in time.