Flagfolds: an approach to multi-dimensional varifolds

TL;DR

The paper introduces Flagfolds, a novel geometric framework embedding Grassmannians into a stratified space of covariance matrices, enabling advanced shape modeling and geodesic computation between subspaces of varying dimensions.

Contribution

It develops a new embedding of Grassmannians into a stratified covariance matrix space and generalizes varifolds to flagfolds for multi-dimensional shape analysis.

Findings

Embedded Grassmannians into a stratified space of covariance matrices.

Defined a Riemannian metric compatible with stratification.

Enabled geodesic computation between subspaces of different dimensions.

Abstract

By interpreting the product of the Principal Component Analysis, that is the covariance matrix, as a sequence of nested subspaces naturally coming with weights according to the level of approximation they provide, we are able to embed all $d$--dimensional Grassmannians into a stratified space of covariance matrices. We observe that Grassmannians constitute the lowest dimensional skeleton of the stratification while it is possible to define a Riemaniann metric on the highest dimensional and dense stratum, such a metric being compatible with the global stratification. With such a Riemaniann metric at hand, it is possible to look for geodesics between two linear subspaces of different dimensions that do not go through higher dimensional linear subspaces as would euclidean geodesics. Building upon the proposed embedding of Grassmannians into the stratified space of covariance matrices, we generalize the concept of varifolds to what we call flagfolds in order to model multi-dimensional shapes.