English

Linear programming on the Stiefel manifold

Optimization and Control 2023-11-01 v2

Abstract

Linear programming on the Stiefel manifold (LPS) is studied for the first time. It aims at minimizing a linear objective function over the set of all pp-tuples of orthonormal vectors in Rn{\mathbb R}^n satisfying kk additional linear constraints. Despite the classical polynomial-time solvable case k=0k=0, general (LPS) is NP-hard. According to the Shapiro-Barvinok-Pataki theorem, (LPS) admits an exact semidefinite programming (SDP) relaxation when p(p+1)/2nkp(p+1)/2\le n-k, which is tight when p=1p=1. Surprisingly, we can greatly strengthen this sufficient exactness condition to pnkp\le n-k, which covers the classical case pnp\le n and k=0k=0. Regarding (LPS) as a smooth nonlinear programming problem, we reveal a nice property that under the linear independence constraint qualification, the standard first- and second-order {\it local} necessary optimality conditions are sufficient for {\it global} optimality when p+1nkp+1\le n-k.

Keywords

Cite

@article{arxiv.2301.06918,
  title  = {Linear programming on the Stiefel manifold},
  author = {Mengmeng Song and Yong Xia},
  journal= {arXiv preprint arXiv:2301.06918},
  year   = {2023}
}
R2 v1 2026-06-28T08:13:29.885Z