Smooth, globally Polyak-{\L}ojasiewicz functions are nonlinear least-squares
Nicolas Boumal, Christopher Criscitiello, Quentin Rebjock

TL;DR
The paper proves that smooth, globally Polyak-Łojasiewicz functions on contractible manifolds are essentially sums of squares, revealing a rigid structure and dichotomy in their geometry and convexity properties.
Contribution
It establishes that such functions must be of a specific sum-of-squares form and characterizes the geometric nature of their minimizer sets, highlighting a fundamental rigidity.
Findings
Functions are sums of squares of submersions.
Minimizer sets are either Euclidean or exotic manifolds.
There exists a Riemannian metric making the function geodesically convex.
Abstract
The Polyak-{\L}ojasiewicz (P{\L}) condition is often invoked in nonconvex optimization because it allows fast convergence of algorithms beyond strong convexity. A function on a Riemannian manifold is globally P{\L} if for all , where and . How much does this pointwise, first-order inequality constrain and its set of minimizers ? We show that if is also smooth () and is contractible (e.g., if ), then the P{\L} condition imposes a firm global structure: such a function is necessarily of the form (a nonlinear sum of squares) where is a submersion, and is the codimension of in . The proof hinges on showing that…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
