OPM, a collection of Optimization Problems in Matlab

TL;DR

OPM provides a compact, Matlab-native collection of CUTEst nonlinear optimization problems, enabling installation-free testing of optimization algorithms within Matlab. It introduces a CPS-aware interface that exposes both full objective evaluations and element-wise functions, along with a CPS decomposition structure to exploit sparsity in large-scale problems. The key contributions include a formal and flexible problem API, asset-light access to 134 continuous problems, and mechanisms to reuse problem structure for efficient derivative evaluations. This approach lowers barriers to rapid prototyping and benchmarking of optimization methods in Matlab and supports advanced CPS-based strategies and derivative-free techniques. The library's design facilitates future extensions and community-driven growth of Matlab-compatible optimization test problems.

Abstract

OPM is a small collection of CUTEst unconstrained and bound-constrained nonlinear optimization problems, which can be used in Matlab for testing optimization algorithms directly (i.e. without installing additional software).