Revisit Gradient Descent for Geodesically Convex Optimization

Abstract: In a seminal work of Zhang and Sra, gradient descent methods for geodesically convex optimization were comprehensively studied. In particular, based on a refined use of the triangle comparison theorem of Toponogov, Zhang and Sra derived a comparison inequality that relates the current iterate, the next iterate and the optimum point. Since their seminal work, numerous follow-ups have studied different downstream usages of their comparison lemma. However, all results along this line relies on strong assumptions, such as bounded domain assumption or curvature bounded below assumption. In this work, we introduce the concept of quasilinearization to optimization, presenting a novel framework for analyzing geodesically convex optimization. By leveraging this technique, we establish state-of-the-art convergence rates -- for both deterministic and stochastic settings -- under substantially weaker assumptions than previously required.