High Rank Matrix Completion via Grassmannian Proxy Fusion

Huanran Li, Jeremy Johnson, Daniel Pimentel-Alarcón · 2026

Read High Rank Matrix Completion via Grassmannian Proxy Fusion by Huanran Li, Jeremy Johnson, Daniel Pimentel-Alarcón (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02565.txt.

Description

Huanran Li and 2 co-authors fill in missing entries of data matrices whose columns gather near a handful of subspaces, in a 2026 arXiv paper. This high-rank completion problem shows up wherever readings are partial and structured. Prior solvers often came without guarantees, were hard to interpret, or needed more observed entries than information theory requires.

The new solver groups partial columns by assigning each a stand-in subspace. It then tunes two geometric costs on the Grassmannian, the manifold of subspaces. One cost is how far each column sits from its assigned subspace, measured chordally. The other is how the chosen subspaces relate to one another along geodesics. On synthetic and real data the method ties the best rivals when observations are plentiful. When observations grow scarce it pulls clearly ahead, closing much of the distance to the floor set by theory.

Catalog record

CREATORHuanran Li, Jeremy Johnson, Daniel Pimentel-Alarcón
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-02565

Browse the full library · About the Tower of Babel