Statistical detection of clumps and gaps in a random sample by continuous wavelet transforms: the 2D case

Abstract

We present a self-consistent framework to perform the wavelet analysis of two-dimensional statistical distributions. The analysis targets the 2D probability density function (p.d.f.) of an input sample, in which each object is characterized by two peer parameters. The method performs a probabilistic detection of various `patterns', or `structures' related to the behaviour of the p.d.f. Laplacian. These patterns may include regions of local convexity or local concavity of the p.d.f., in particular peaks (groups of objects) or gaps. In the end, the p.d.f. itself is reconstructed based on the least noisy (most economic) superposition of such patterns. Among other things, our method also involves optimal minimum-noise wavelets and minimum-noise reconstruction of the distribution density function. The new 2D algorithm is now implemented and released along with an improved and optimized 1D version. The code relies on the C++11 language standard, and is fully parallelized. The algorithm has a rich range of applications in astronomy: Milky Way stellar population analysis, investigations of the exoplanets diversity, Solar System minor bodies statistics, etc.