Advances in progressive decoupling methodology for problem decomposition in optimization

Applied Nanoscience · Springer Nature · 2026-10-19 · DOI: 10.1007/s10107-026-02430-5

Аннотация

The progressive decoupling algorithm solves problems of optimization by iteratively suppressing complications coming from linear linkage relations. Here its formulation is significantly extended and a sharper theory of convergence is provided. Minimization steps no longer have to be executed exactly, and proximal terms in broadened form, potentially for numerical preconditioning, can vary from one iteration to the next. Linear convergence, for which no results were previously available, is tied precisely to a condition of metric subregularity at the solution, with its rate given by a formula based on the modulus of that property. It is established that, for problems with piecewise linear-quadratic objectives, linear convergence is guaranteed, and that for convex general objectives the lack of it is, in a perturbational sense, rare. The methodology is shown to be extendable also to problem decomposition with nonconvex objectives, but with shortcomings so far in practical implementability.

Читать у издателя →

Похожие статьи