Aspects of topological approaches for data science (Q2148960): Difference between revisions

The paper establishes a new theory which unifies various aspects of topological approaches for data science. The topological approaches studied in this paper are applicable to both point cloud data and (super-hyper-)graph data. The paper generalizes simplicial complexes and hypergraphs to super-hypergraphs and establishes super-hypergraph homology as an extension of simplicial homology. In Section 2, the authors establish a homology theory on super-hypergraphs. In Theorem 2.7, the authors prove that the embedded homology of a hypergraph does not depend on the choice of orientation on the associated simplicial complex. In Theorem 2.8, the authors prove that for any hypergraph, the embedded homology of the cone is zero in positive dimensions and is one-dimensional in the zero dimension. In Theorem 2.17, the authors characterize super-hypergraphs. In Theorem 2.20, the authors give a commutative diagram of the infimum chain complex and the supremum chain complex for super-hypergraphs. In Section 3, the authors study super-persistent homology. In Theorem 3.7, the authors prove a structure theorem. In Subsection 3.2, the authors discuss the ordinary persistent homology. In Subsection 3.3, the authors study the partition homology and the persistent partition homology. In Subsection 3.5, the authors give a description of simplicial homology via Delta-sets. In Section 4, the authors give some potential applications in topological data analysis. To summarize, the paper is of important innovation. The authors give a significant theory in topological data analysis.