English

First Order Algorithm on an Optimization Problem with Improved Convergence when Problem is Convex

Optimization and Control 2025-08-20 v1

Abstract

We propose a first order algorithm, a modified version of FISTA, to solve an optimization problem with an objective function that is a sum of a possibly nonconvex function, with Lipschitz continuous gradient, and a convex function which can be nonsmooth. The algorithm is shown to have an iteration complexity of O(ϵ2)\mathcal{O}(\epsilon^{-2}) to find an ϵ\epsilon-approximate solution to the problem, and this complexity improves to O(ϵ2/3)\mathcal{O}(\epsilon^{-2/3}) when the objective function turns out to be convex. We further provide asymptotic convergence rate for the algorithm of worst case o(ϵ2)o(\epsilon^{-2}) iterations to find an ϵ\epsilon-approximate solution to the problem, with worst case o(ϵ2/3)o(\epsilon^{-2/3}) iterations when its objective function is convex.

Keywords

Cite

@article{arxiv.2508.13302,
  title  = {First Order Algorithm on an Optimization Problem with Improved Convergence when Problem is Convex},
  author = {Chee-Khian Sim},
  journal= {arXiv preprint arXiv:2508.13302},
  year   = {2025}
}

Comments

15 pages