PreprintCC BY 4.0arXiv
Certifiably Correct Fréchet Means in Orbit Spaces
Abstract
We study Fréchet means in orbit spaces of finite orthogonal group actions. Since computing these means is NP-hard in general, we take inspiration from group synchronization to develop tractable spectral and semidefinite relaxations. Pairwise alignment certifies global optimality and we develop a semidefinite certificate that can extend beyond this condition. For Gaussian data modulo sign, we prove that our certificate succeeds in a high-noise regime where alignment fails. Numerical experiments evaluate these methods in various settings, including summarizing ensembles of districting plans and clustering time series
Related
- Weighted Power Fréchet Means in Metric Spaces with Curvature Bounded Above
- Finite-Sample Bounds for Fréchet Means of Dependent Data in Metric Spaces with Curvature Bounded Above
- Invariant means, coset correct means and the Hahn-Banach Theorem
- Central Limit Theorems for Sample Fréchet Means of Manifold-Valued Markov Chains
- Dense uniqueness and dense nonuniqueness of Fréchet means
BibTeX
@misc{mixon2026certifiably,
title = {{Certifiably Correct Fréchet Means in Orbit Spaces}},
author = {Dustin G. Mixon and Tom Needham and Nathan Willey},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.24191},
eprint = {2610.24191},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.24191},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Certifiably Correct Fréchet Means in Orbit Spaces
AU - Dustin G. Mixon
AU - Tom Needham
AU - Nathan Willey
PY - 2026
DA - 2026/10/20
PB - arXiv
DO - 10.48550/arxiv.2610.24191
UR - https://arxiv.org/abs/2610.24191
AB - We study Fréchet means in orbit spaces of finite orthogonal group actions. Since computing these means is NP-hard in general, we take inspiration from group synchronization to develop tractable spectral and semidefinite relaxations. Pairwise alignment certifies global optimality and we develop a semidefinite certificate that can extend beyond this condition. For Gaussian data modulo sign, we prove that our certificate succeeds in a high-noise regime where alignment fails. Numerical experiments evaluate these methods in various settings, including summarizing ensembles of districting plans and clustering time series
ER -