Finsler Metric Clustering in Weighted Projective Spaces (2507.00001v1)

Abstract: This paper develops a hierarchical clustering algorithm for weighted projective spaces $\mathbb{P}{\mathbf{q}}$, utilizing a Finsler metric $ d_F([z], [w]) $ and its rational analogue $ d{F,\mathbb{Q}}([z], [w])$ to define distances that preserve the non-Euclidean geometry of these quotient manifolds. Defined via geodesic integrals of a scaling invariant Finsler norm weighted by the grades $\mathbf{q} = (q_0, q_1, \dots, q_n)$, these metrics satisfy true metric properties including the triangle inequality, overcoming the limitations of the non-metric dissimilarity measure from prior work.