Multiscale projective coordinates via persistent cohomology of sparse filtrations (Q1702351)

A data set \(X\) can be thought of as a subset of some metric space \(M\), usually a manifold or Euclidean space. If \(X\) is sampled yielding a set \({\mathbb X}\), the topological inference problem is to infer the topological properties of \(X\) from \({\mathbb X}\). This paper shows how one can use persistent cohomology of a sparse filtration by sample sets to construct transition functions for line bundles over neighbourhoods of the data. The characteristic classes of such line bundles give maps into projective spaces. By compressing these maps into finite dimensional projective subspaces the author infers dimensional reductions for the space \(X\).