Intrinsic dimensionality estimation for the galaxy’s distribution structure analysis

Abstract

The proposed local intrinsic dimensionality method (TIDIM algorithm) demonstrates significant potential in detecting specific filament-like and globular cluster-like structures. It provides a non-parametric, reproducible, and resolution-flexible framework for identifying complex structures within three-dimensional distributions. This approach offers a complementary perspective to traditional statistical tools by focusing on local features such as voids, filaments, and globular clusters. Additionally, it allows for the reconstruction of discrete tracers using the same flexible framework. Its capacity to localize anisotropies and apply TIDIM on a spatial grid makes it particularly useful for comparisons with simulations. The Sobol grid-based TIDIM analysis complements galaxy-based assessments by enabling the detection of structures in sparsely populated regions.