Constrained Optimization in the Presence of Noise

TL;DR

This paper develops a modified SQP method for nonlinear constrained optimization that remains stable and converges near the solution despite noisy function evaluations, with practical improvements demonstrated through numerical experiments.

Contribution

It introduces a noise-aware relaxation of the line search in SQP, enabling convergence without regularization or trust regions under noisy conditions.

Findings

Relaxed line search improves practical performance with noisy evaluations.

Convergence to a neighborhood of the solution is achieved under certain conditions.

Method applicable to derivative-free optimization with finite differences.

Abstract

The problem of interest is the minimization of a nonlinear function subject to nonlinear equality constraints using a sequential quadratic programming (SQP) method. The minimization must be performed while observing only noisy evaluations of the objective and constraint functions. In order to obtain stability, the classical SQP method is modified by relaxing the standard Armijo line search based on the noise level in the functions, which is assumed to be known. Convergence theory is presented giving conditions under which the iterates converge to a neighborhood of the solution characterized by the noise level and the problem conditioning. The analysis assumes that the SQP algorithm does not require regularization or trust regions. Numerical experiments indicate that the relaxed line search improves the practical performance of the method on problems involving uniformly distributed noise. One important application of this work is in the field of derivative-free optimization, when finite differences are employed to estimate gradients.