Complexity guarantees and polling strategies for Riemannian direct-search methods

TL;DR

This paper extends direct-search optimization methods to Riemannian manifolds, establishing complexity guarantees and comparing strategies for generating positive spanning sets, especially on the unit hypersphere.

Contribution

It introduces a framework for Riemannian direct-search with complexity bounds and analyzes two methods for creating positive spanning sets on manifolds.

Findings

Generating directions directly in tangent space improves complexity on the hypersphere.

Projection-based PSSs are less efficient than tangent space methods.

Numerical experiments show the impact of dimension and codimension on performance.

Abstract

Direct-search algorithms are derivative-free optimization techniques that operate by polling the variable space along specific directions forming positive spanning sets (PSSs). When the problem variables are constrained to lie on a Riemannian manifold, polling must be performed along tangent directions. Although Riemannian variants of direct search have already been proposed and endowed with asymptotic guarantees, a proper generalization of PSSs on manifolds remains to be investigated. In particular, a measure of quality for those PSSs is required to obtain complexity bounds for direct search. In this paper, we derive complexity guarantees for a class of Riemannian direct-search techniques, and study two ways of generating positive spanning sets in tangent spaces. We pay particular attention the unit hypersphere case, for which we establish that generating directions directly within the tangent space leads to better complexity properties than projecting PSSs from the ambient space onto the tangent space. Our numerical experiments highlight the impact of dimension and codimension in more general settings.