A Study of First-Order Methods with a Deterministic Relative-Error Gradient Oracle

Nadav Hallak, Kfir Yehuda Levy

Abstract: This paper studies the theoretical guarantees of the classical projected gradient and conditional gradient methods applied to constrained optimization problems with biased relative-error gradient oracles. These oracles are used in various settings, such as distributed optimization systems or derivative-free optimization, and are particularly common when gradients are compressed, quantized, or estimated via finite differences computations. Several settings are investigated: Optimization over the box with a coordinate-wise erroneous gradient oracle, optimization over a general compact convex set, and three more specific scenarios. Convergence guarantees are established with respect to the relative-error magnitude, and in particular, we show that the conditional gradient is invariant to relative-error when applied over the box with a coordinate-wise erroneous gradient oracle, and the projected gradient maintains its convergence guarantees when optimizing a nonconvex objective function.

Submission Number: 5230