Riemannian Optimization on Unknown Manifolds via Stochastic Proximal Refinement
Abstract
We study Riemannian optimization on an unknown manifold accessed only through feasible samples. The central challenge is to turn feasible samples into optimization steps, i.e., the samples are drawn from the unknown feasible manifold, but do not directly provide the tangent spaces, projections, or retractions required by classic Riemannian methods. Moreover, regions with low objective values may be sparsely represented in the data, so directly sampling from a distribution fitted to these observations can require many draws to find better feasible solutions. We introduce Proximal Refinement on Implicit Smooth Manifolds (PRISM), which turns feasible-sample access into an iterative optimization procedure by repeatedly sampling an objective-aware proximal distribution conditioned on the current iterate. Theoretically, we show that PRISM recovers the effect of Riemannian-gradient refinement, leading to provable objective descent, feasibility (manifold adherence), and convergence toward first-order stationarity. We demonstrate the effectiveness of our approach through synthetic experiments illustrating its geometric properties and applications to data-driven building control and diverse real world optimization tasks, highlighting its potential as a general framework for data-driven optimization across scientific and engineering domains.
Related
- A Stochastic Riemannian Alternating Descent Ascent Method for Nonsmooth Composite Expectation Optimization on Riemannian Manifolds
- Online Covariance Estimation for Averaged Stochastic Gradient Descent on Riemannian Manifolds
- Riemannian optimization on low-rank tensor train manifolds for the Gross-Pitaevskii equation
- Technical Report: Partially Asynchronous Distributed Optimization on Riemannian Manifolds
- A Stochastic Proximal-Gradient Method for Regularized Generally Constrained Continuous Optimization
BibTeX
@misc{wu2026riemannian,
title = {{Riemannian Optimization on Unknown Manifolds via Stochastic Proximal Refinement}},
author = {Ruixiang Wu and Jiahao Ai and Tinko Sebastian Bartels and Xuanzhuo Liu and Yifei Wu and Yu Mao and Tongxin Li and Costas Courcoubetis},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.41379},
eprint = {2610.41379},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.41379},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Riemannian Optimization on Unknown Manifolds via Stochastic Proximal Refinement
AU - Ruixiang Wu
AU - Jiahao Ai
AU - Tinko Sebastian Bartels
AU - Xuanzhuo Liu
AU - Yifei Wu
AU - Yu Mao
AU - Tongxin Li
AU - Costas Courcoubetis
PY - 2026
DA - 2026/10/30
PB - arXiv
DO - 10.48550/arxiv.2610.41379
UR - https://arxiv.org/abs/2610.41379
AB - We study Riemannian optimization on an unknown manifold accessed only through feasible samples. The central challenge is to turn feasible samples into optimization steps, i.e., the samples are drawn from the unknown feasible manifold, but do not directly provide the tangent spaces, projections, or retractions required by classic Riemannian methods. Moreover, regions with low objective values may be sparsely represented in the data, so directly sampling from a distribution fitted to these observations can require many draws to find better feasible solutions. We introduce Proximal Refinement on Implicit Smooth Manifolds (PRISM), which turns feasible-sample access into an iterative optimization procedure by repeatedly sampling an objective-aware proximal distribution conditioned on the current iterate. Theoretically, we show that PRISM recovers the effect of Riemannian-gradient refinement, leading to provable objective descent, feasibility (manifold adherence), and convergence toward first-order stationarity. We demonstrate the effectiveness of our approach through synthetic experiments illustrating its geometric properties and applications to data-driven building control and diverse real world optimization tasks, highlighting its potential as a general framework for data-driven optimization across scientific and engineering domains.
ER -